Method, apparatus, and system for anomaly detection and parts replacement cycle optimization of subscription-type solar power facilities based on digital twin

Abstract

One embodiment of the present invention relates to a method, apparatus, and system for detecting abnormal signs and optimizing the component replacement cycle of a digital twin-based subscription-type solar power facility, which utilizes a solar asset model replicated in a virtual space and heterogeneous multi-simulation to predict the degradation status of individual components and derive an optimal maintenance timing considering economic feasibility.

Description

Technology Field

[0001] The following embodiments relate to a method, apparatus, and system for detecting anomalies and optimizing component replacement cycles of a digital twin-based subscription-type solar power facility, which utilizes a solar asset model replicated in a virtual space and heterogeneous multi-simulation to predict the degradation status of individual components and derive an optimal maintenance timing considering economic feasibility. Background Technology

[0002] With the recent expansion of renewable energy adoption, subscription-based solar power generation business models that reduce the burden of initial installation costs are gaining popularity. In subscription business models, the profitability of the entire project can only be secured by guaranteeing consistent power generation performance during the contract period and minimizing operation and maintenance (O&M) costs. However, conventional solar asset maintenance methods have primarily relied on time-based maintenance, which involves replacing components according to a uniform cycle recommended by the manufacturer, or run-to-failure maintenance, which involves performing repairs immediately after a failure occurs.

[0003] These conventional methods failed to reflect the unique weather conditions and actual operational stresses faced by individual solar power plants, leading to problems such as unplanned downtime caused by unexpected component failures and massive losses in electricity sales. In particular, despite potential-induced degradation (PID) in solar panel modules and thermal fatigue in internal power semiconductors of inverters proceeding through mutually distinct mechanisms, conventional technology failed to analyze these factors comprehensively from both physical and data perspectives. Consequently, there were fundamental limitations in accurately predicting remaining lifespan (RUL) and optimizing replacement timing considering economic feasibility. Prior art literature

[0004] (Patent Document 0001) KR 10-2578639 B (Patent Document 0002) KR 10-2444254 B (Patent Document 0003) KR 10-2539017 B (Patent Document 0004) KR 10-2940218 B (Patent Document 0005) KR 10-2326571 B The problem to be solved

[0005] The problem that one embodiment of the present invention aims to solve is to provide a method, device, and system for detecting abnormal signs and optimizing the component replacement cycle of a digital twin-based subscription-type solar power facility, which, in order to overcome the limitations of the conventional uniform and reactive solar asset maintenance method described above, precisely predicts the remaining lifespan of individual components through ensemble simulation combining a physical model and a data-based model, calculates physical replacement costs and power sales losses, and dynamically determines the optimal component replacement time to maximize economic returns over the entire subscription cycle. means of solving the problem

[0006] According to one embodiment, a method for detecting abnormal signs and optimizing the component replacement cycle of a digital twin-based subscription-type solar asset, wherein the device includes a processor, memory, a communication module, and a non-transient storage medium, and the processor executes a program stored in the non-transient storage medium, comprises: a step of collecting integrated operational data including weather environment data, power market indicators, and real-time power generation status data of the solar asset; a step of analyzing the integrated operational data to generate operational analysis information; and a step of generating operational execution information including at least one of an asset operation plan, a resource control plan, a settlement plan, a revenue distribution plan, a maintenance plan, and a contract management plan based on the operational analysis information; wherein the step of generating the operational analysis information comprises: a step of synchronizing a digital twin object that replicates the physical specifications and real-time degradation status of the solar asset in a virtual space based on the integrated operational data; The present invention provides a method for detecting anomalies and optimizing the component replacement cycle of a digital twin-based subscription-type solar asset, comprising: a step of performing an accelerated degradation simulation of a target component by chronologically inputting weather environment data and asset operation parameters into the digital twin object, and deriving the remaining useful life (RUL) and failure probability for each component by calculating the state error rate with a preset normal operation profile; and a step of generating operational execution information comprising: a step of calculating the physical replacement cost of the target component, the expected loss of power sales due to failure, and the remaining subscription contract period of the corresponding subscriber based on the remaining useful life and failure probability; and a step of determining the optimal replacement time for each component to maximize the economic return of the entire subscription cycle.

[0007] In addition, the step of deriving the remaining lifespan and failure probability for each of the above components comprises: defining the solar panel module, power semiconductor, and DC link capacitor as target components; extracting meteorological factors from the meteorological environment data, including cumulative UV exposure, temperature variation range, humidity variation range, number of temperature and humidity cycles, wind speed variation, and wind load variation; extracting operational factors from the asset operation parameters and real-time power generation status data, including DC-side voltage / current (IV) trajectory, AC-side voltage / current waveform, output load rate, switching frequency, and conversion efficiency variation; and mapping the meteorological factors and operational factors as stress factors representing thermal stress, electrical stress, mechanical stress, and environmental stress for each target component with respect to the Electrical Equivalent Circuit Model of the solar panel module and the Multiphysics Thermal Model of the power conversion inverter constituting the digital twin object. Based on the above stress factors, a step of performing a physics-based first-order accelerated simulation for a solar panel module to calculate potential-induced degradation (PID), micro-crack progression rate, and series resistance increase; a step of performing a data-driven second-order simulation to estimate the cumulative residual of non-linear thermal fatigue of inverter internal power semiconductors (IGBTs) and DC-Link Capacitors by extracting high-frequency noise components, switching ripple components, harmonic distortion components, and instantaneous temperature fluctuation components from the real-time power generation status data and inputting them into a time-series deep learning model; and a step of calculating an ensemble future degradation trajectory for each target component by synchronizing the results of the first-order accelerated simulation and the second-order simulation on the same time axis and weightedly combining them.The method may include: a step of generating an expected power generation efficiency curve, an expected output deviation curve, and an expected internal resistance increase curve for each target component from the above-mentioned ensemble future degradation trajectory, and calculating a Dynamic SoH Error Rate by comparing them with the above-mentioned normal operation profile generated based on the manufacturer's shipment specifications and the initial installation environment of the solar asset; a step of correcting a performance curve corresponding to the degradation characteristics of each target component among the above-mentioned power generation efficiency curve, expected output deviation curve, and expected internal resistance increase curve for each target component based on the above-mentioned Dynamic SoH Error Rate; and a step of deriving a temporal confidence interval of the remaining lifespan for each target component and a failure probability for each time interval by integrating the probability density function (PDF) at the point in time when the performance curve corresponding to the degradation characteristics of each target component does not satisfy the performance guarantee threshold condition corresponding to each target component.

[0008] And, the step of calculating the ensemble future degradation trajectory comprises: generating a first degradation state vector including an expected power generation efficiency degradation rate, an expected series resistance increase rate, an expected cell damage area ratio, and an expected insulation performance degradation rate for each solar panel module from the results of the first acceleration simulation; generating a second degradation state vector including an expected junction temperature swing, an expected conduction loss increase rate, an expected switching loss increase rate for each power semiconductor, and an expected capacitance degradation rate, an expected equivalent series resistance increase rate, and an expected ripple tolerance degradation rate for each DC link capacitor from the results of the second simulation; rearranging the first degradation state vector and the second degradation state vector on the same time grid at preset sampling cycles and performing interpolation operations on missing time points to generate a time-series degradation matrix for each target component; and calculating reliability parameters including an error variance, a residual autocorrelation coefficient, a prediction uncertainty, and a recent time point response sensitivity for each target component based on the difference between the time-series degradation matrix and the normal operation profile. Based on the above reliability parameters, a step of calculating a first weight for the results of a first acceleration simulation for the solar panel module and a second weight for the results of a second simulation for the power semiconductor and DC link capacitor, and normalizing the first weight and the second weight to generate a weight vector for each target component; a step of generating an ensemble degradation state vector for each target component by linearly combining each degradation component of a time-series degradation matrix using the weight vector, and generating a degradation transition covariance matrix based on the time-series variance and covariance of the ensemble degradation state vector; a step of inputting the ensemble degradation state vector and the degradation transition covariance matrix into an ensemble Kalman filter to sequentially correct the time-series degradation state, and performing Monte Carlo sampling on the corrected degradation state to generate a plurality of future degradation scenarios;and may include the step of deriving an ensemble future degradation trajectory for each target component, including an expected power generation efficiency degradation curve, an expected output deviation curve, an expected internal resistance increase curve, and an expected thermal fatigue accumulation curve for each of the photovoltaic panel module, power semiconductor, and DC link capacitor, by superimposing the above future degradation scenarios on a time axis.;

[0009] In addition, the step of generating the operational analysis information further comprises: a step of generating a real asset tokenization profile by aligning the power generation capacity, construction cost, expected lifespan, and real-time power generation status data of the solar assets included in the integrated operational data, reflecting the issuance size of a Security Token Offering (STO) that divides the total ownership of the solar assets into predetermined units, the share of rights per token, the value depreciation rate, and characteristics of asset return rates by time period; and a step of calculating asset return analysis information including total power sales revenue based on real-time power generation and power unit price, platform operation fee deductions, asset maintenance reserves, and expected dividends per unit of the Security Token Offering by combining weather environment data and power market indicators with the real asset tokenization profile; and the step of generating the operational execution information further comprises: a step of generating token asset operation standard information including the dividend cycle, conditions for automatic transfer of proceeds, token transaction approval criteria, and tax withholding criteria of a smart contract running on a blockchain distributed ledger, based on the asset return analysis information. The method further includes the step of determining the settlement plan and profit distribution plan by inputting the actual generated power sales revenue and subscription fees into the smart contract logic based on the asset profit analysis information and token asset operation standard information, calculating dividends proportional to the token shareholding ratio held by each subscriber, and hashing and recording the distribution details on the blockchain network; and the step of generating the real asset tokenization profile comprises: the step of configuring the underlying asset calculation data by extracting the total construction cost, power generation capacity, expected lifespan, and real-time power generation status data of the solar asset from the integrated operation data; the step of calculating the total number of tokens to be issued to liquidate the total ownership of the solar asset by applying a pre-set fiat currency-based face value division unit to the total construction cost, and determining the total number of tokens to be issued as the issuance size of the Security Token Offering (STO);A step of determining the ratio of one unit token to the total ownership of the solar asset based on the total number of issued tokens as the right share per token, and calculating the collateral facility capacity per unit token corresponding to the right share per token; a step of calculating a straight-line depreciation rate according to financial accounting standards based on the expected lifespan and the elapsed operating period up to the present time; a step of estimating the physical cumulative degradation rate of the solar panel based on the DC voltage drop, the time-dependent output degradation rate, and the module temperature variability included in the real-time power generation status data; a step of dynamically combining the straight-line depreciation rate and the physical cumulative degradation rate according to a time-weighted function to calculate a value depreciation rate as a time-series function that tracks the decline in the actual value of the underlying asset of the token security; and a step of calculating the Levelized Cost of Energy (LCOE) of the solar asset based on the past time-series power generation pattern generated by the accumulation of the real-time power generation status data and the predicted power generation amount during the expected lifespan. A step of calculating an expected hourly revenue flow by reflecting a pre-established standard power unit price for asset valuation to the seasonal solar radiation deviation and hourly power production efficiency variability derived from the above-mentioned past time-series power generation pattern, and determining the characteristics of the asset return by hour by deriving a dynamic Internal Rate of Return (IRR) curve based on the expected hourly revenue flow and the above-mentioned life-cycle levelized cost of generation; a step of structuring the token security issuance scale, the right share ratio per token, the collateral facility capacity per unit token, the value depreciation rate, and the characteristics of the asset return by hour into off-chain data outside the blockchain platform;and may include the step of generating a hash value for data integrity verification regarding the off-chain data, and deriving a real asset tokenization profile by packaging the hash value into a metadata format for Oracle integration that can be referenced by a smart contract to be deployed on a distributed ledger within a blockchain platform.

[0010] In addition, the step of calculating the asset return analysis information comprises: verifying forgery or tampering of off-chain data corresponding to the real asset tokenization profile using hash values ​​included in the oracle integration metadata format, and obtaining characteristics of the value depreciation rate, the share of rights per token, and the asset return rate by time period from the verified off-chain data; calculating the effective power generation efficiency per unit hour by combining the real-time solar radiation forecast, ambient temperature, and time period information included in the weather environment data, the DC voltage drop and module temperature variability included in the real-time power generation status data, and the physical cumulative degradation rate included in the value depreciation rate; calculating the real-time power generation amount by time period based on the power generation capacity and effective power generation efficiency per unit hour of the solar asset; extracting the real-time System Marginal Price (SMP) and Renewable Energy Certificate (REC) trading unit price included in the power market indicator, and calculating the power unit price by time period by combining the real-time System Marginal Price and the REC trading unit price. A step of calculating hourly power sales revenue by applying hourly power unit prices to the real-time power generation amount for each hour, and calculating total power sales revenue by accumulating the calculated hourly power sales revenues; a step of calculating the deviation between the straight-line depreciation rate and the physical cumulative degradation rate, and estimating the probability of component failure within the remaining lifespan and the marginal maintenance cost based on the deviation; a step of calculating an asset maintenance reserve by applying a pre-set risk hedge ratio to the total power sales revenue to depreciate the marginal maintenance cost; and a step of calculating a platform operation fee deduction amount by summing the gas fee representing the network transaction processing cost of the blockchain platform, the oracle integration processing fee, and the operation cost based on the solar power plant entrusted operation rate.The method may include: a step of calculating a distributable net profit by subtracting an asset maintenance reserve and a platform operation fee deduction from the total power sales revenue; a step of calculating an estimated dividend per unit of token security by applying a dividend reflection coefficient based on the rights share ratio per token and the characteristics of asset return rates by time period to the distributable net profit; and a step of generating asset return analysis information by structuring the real-time power generation amount by time period, the power unit price by time period, the total power sales revenue, the platform operation fee deduction, the asset maintenance reserve, the distributable net profit, and the estimated dividend per unit of token security into an interconnected profit analysis record.

[0011] A device according to one embodiment may be combined with hardware and controlled by a computer program stored on a medium to execute the method of any one of the methods described above. Effects of the invention

[0012] According to one embodiment, by using a digital twin object synchronized in a virtual space to predict the remaining lifespan and failure probability of individual parts, and by comprehensively calculating the physical part replacement cost, the expected loss of power sales due to failure, and the remaining subscription contract period, it is possible to derive the optimal preemptive part replacement timing that can maximize economic returns from the perspective of the entire lifecycle of a subscription business model.

[0013] In addition, by combining the results of a physics-based first-order acceleration simulation for solar panel modules and the results of a time-series deep learning-based second-order simulation for internal components of power conversion inverters on the same time axis, the complex degradation characteristics of heterogeneous components involving a mixture of physical stress (dislocation-induced degradation, microcracks, etc.) and thermal stress (nonlinear thermal fatigue, etc.) can be integrated into a single ensemble future degradation trajectory, thereby dramatically improving the accuracy of life prediction.

[0014] In addition, by calculating error variance and reliability parameters based on the time-series degradation matrix derived from the results of heterogeneous simulations and applying them to the Ensemble Kalman Filter and Monte Carlo sampling to sequentially correct the degradation status at each time point, it is possible to stably derive multiple future degradation scenarios even in environments with local weather variability and prediction uncertainty, and maximize the statistical reliability of the predictive maintenance system.

[0015] According to one embodiment, by generating a tokenization profile and asset return analysis information that links the real asset value of a solar power plant with real-time power sales revenue, and inputting this into smart contract logic to automatically calculate dividends and record them in the ledger, a transparent and highly reliable fractional investment and profit distribution environment can be provided to multiple subscribers (investors).

[0016] Furthermore, by dynamically combining the financial straight-line depreciation rate of solar assets with the physical cumulative degradation rate based on actual weather and operation data to calculate the value depreciation rate as a time-series function, the actual decrease in value of the underlying asset can be accurately reflected in the token security. In addition, by structuring asset return characteristics, such as the dynamic internal rate of return (IRR), as off-chain data outside the blockchain and packaging them into metadata for oracle integration that includes hash values, it is possible to prevent the blockchain mainnet from being overloaded due to the processing of vast amounts of real-world data while fundamentally blocking data tampering.

[0017] Furthermore, real-time power generation and sales revenue are calculated based on external data streams verified through the Oracle network; by calculating the 'deviation' between the financial straight-line depreciation rate and the physical cumulative degradation rate, future failure probabilities and marginal maintenance costs can be preemptively estimated. By automatically hedging these estimated costs from power sales revenue into an asset maintenance reserve, the financial stability of dividend payments is strongly guaranteed even in the event of unexpected equipment failures. In addition, by explicitly reflecting gas fees for blockchain transaction processing and Oracle call fees in the platform operation fees to derive the final distributable net profit, this method offers an excellent effect in maximizing the economic sustainability of Web3-based energy business models. Brief explanation of the drawing

[0018] FIG. 1 is a schematic diagram showing a digital twin-based subscription-type solar power system for detecting abnormal signs and optimizing the component replacement cycle according to an embodiment of the present invention. FIG. 2 is a flowchart illustrating a method for detecting abnormal signs and optimizing the component replacement cycle of a digital twin-based subscription-type solar power facility according to an embodiment of the present invention. FIG. 3 is a flowchart illustrating the steps for generating operational analysis information of a method for detecting abnormal signs and optimizing the component replacement cycle of a digital twin-based subscription-type solar power facility according to an embodiment of the present invention. FIG. 4 is a flowchart illustrating the steps for generating operational execution information of a method for detecting abnormal signs and optimizing the component replacement cycle of a digital twin-based subscription-type solar power facility according to an embodiment of the present invention. Specific details for implementing the invention

[0019] Hereinafter, embodiments are described in detail with reference to the attached drawings. However, various modifications may be made to the embodiments, and thus the scope of the patent application is not limited or restricted by these embodiments. It should be understood that all modifications, equivalents, and substitutions to the embodiments are included within the scope of the rights.

[0020] Specific structural or functional descriptions of the embodiments are disclosed for illustrative purposes only and may be modified and implemented in various forms. Accordingly, the embodiments are not limited to the specific disclosed forms, and the scope of this specification includes modifications, equivalents, or substitutions that fall within the technical concept.

[0021] Terms such as "first" or "second" may be used to describe various components, but these terms should be interpreted solely for the purpose of distinguishing one component from another. For example, the first component may be named the second component, and similarly, the second component may be named the first component.

[0022] When it is stated that a component is "connected" to another component, it should be understood that it may be directly connected to or coupled with that other component, or that there may be other components in between.

[0023] The terms used in the embodiments are for illustrative purposes only and should not be interpreted as intended to be limiting. Singular expressions include plural expressions unless the context clearly indicates otherwise. In this specification, terms such as "comprising" or "having" are intended to indicate the existence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0024] In particular, where a 'step' in this specification is described as 'comprising' one or more detailed steps or sub-steps, said 'step' may be interpreted as including its own basic processing step while simultaneously performing the described detailed steps as well.

[0025] For example, if it is stated that 'a step of doing B to A' includes 'a step of doing D to C; a step of doing F to E; and a step of doing H to G,' the 'step of doing B to A' may be interpreted not merely as the basic operation of doing B to A, but as a configuration that performs detailed procedures together, such as a step of doing D to C, a step of doing F to E, and a step of doing H to G.

[0026] Accordingly, the above configuration does not exclude various sub-procedures included within the scope of execution of the corresponding step, and may be included within the scope of the present invention even if other procedures or means performing substantially the same or equivalent functions are substituted.

[0027] Expressions such as 'end part', 'both ends', 'one end', 'other end', and 'side end' of a component can be interpreted as referring to at least / any one of the end parts of that component.

[0028] In the description of the present invention, 'a method in which a device comprises a processor, a memory, a communication module, and a non-transient storage medium, and a program stored in the non-transient storage medium is executed by the processor,' the term 'method' may be interpreted as referring to the program stored in the non-transient storage medium itself or a part of the program.

[0029] The term 'Return' as used in the description of the present invention may refer to a result value being output, returned, or returned from a method, procedure, function, etc. used in a given program language / structure.

[0030] Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as generally understood by those skilled in the art to which the embodiments pertain. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an ideal or overly formal sense unless explicitly defined in this application. For example, the term 'artificial intelligence model' may be selected from one or more of known general artificial intelligence models.

[0031] In addition, when describing with reference to the attached drawings, identical components are assigned the same reference numeral regardless of drawing symbols, and redundant descriptions thereof are omitted. In describing the embodiments, if it is determined that a detailed description of related prior art could unnecessarily obscure the essence of the embodiments, such detailed description is omitted.

[0032] According to one embodiment, a method for detecting abnormal signs and optimizing the component replacement cycle of a digital twin-based subscription-type solar asset, wherein the device includes a processor, memory, a communication module, and a non-transient storage medium, and the processor executes a program stored in the non-transient storage medium, comprises: a step of collecting integrated operational data including weather environment data, power market indicators, and real-time power generation status data of the solar asset; a step of analyzing the integrated operational data to generate operational analysis information; and a step of generating operational execution information including at least one of an asset operation plan, a resource control plan, a settlement plan, a revenue distribution plan, a maintenance plan, and a contract management plan based on the operational analysis information; wherein the step of generating the operational analysis information comprises: a step of synchronizing a digital twin object that replicates the physical specifications and real-time degradation status of the solar asset in a virtual space based on the integrated operational data; The present invention provides a method for detecting anomalies and optimizing the component replacement cycle of a digital twin-based subscription-type solar asset, comprising: a step of performing an accelerated degradation simulation of a target component by chronologically inputting weather environment data and asset operation parameters into the digital twin object, and deriving the remaining useful life (RUL) and failure probability for each component by calculating the state error rate with a preset normal operation profile; and a step of generating operational execution information comprising: a step of calculating the physical replacement cost of the target component, the expected loss of power sales due to failure, and the remaining subscription contract period of the corresponding subscriber based on the remaining useful life and failure probability; and a step of determining the optimal replacement time for each component to maximize the economic return of the entire subscription cycle.

[0033] The step of collecting integrated operational data is the stage in which the device collects basic input information necessary for diagnosing the condition of subscription-based solar assets, predicting degradation, and optimizing component replacement cycles in conjunction with external and internal sources.

[0034] Here, integrated operational data refers to a data set in which data from different sources are aligned based on the same asset identification information and the same time axis to reflect the power generation performance, component degradation status, external environmental conditions, power sales conditions, and contract operation conditions of solar assets. For example, the integrated operational data may include meteorological environmental data such as solar radiation, ambient temperature, humidity, wind speed, precipitation status, and UV index received from a weather observation server or weather forecast server; power market indicators such as real-time system marginal price, hourly sales unit price, correction factor, and price indicators related to renewable energy certificates received from a power exchange or settlement server; and real-time power generation status data such as DC and AC voltage, current, output, output load factor, frequency, power factor, module surface temperature, vibration, and insulation status collected from inverters, junction boxes, power meters, panel sensors, or BMS-type sensors.

[0035] In the step of collecting integrated operational data, time reference unification, missing value correction, outlier removal, sampling period reordering, and asset unique identifier mapping can be performed on the raw data received from each of the above data sources. For example, since weather environment data may be received in 10-minute intervals and real-time power generation status data may be received in 1-second intervals, the device can resample the data based on a predetermined reference time grid, such as an operational analysis cycle of 5 minutes or 15 minutes, to reconstruct it into a single time-series input structure. Additionally, if the same solar asset consists of multiple panel strings, inverter channels, or sub-equipment modules, the device can generate structured data suitable for subsequent degradation analysis by aggregating or separately storing measurement values ​​for each sub-equipment at the asset or component level. The integrated operational data aligned in this way can be utilized as input data for subsequent digital twin synchronization and degradation simulation.

[0036] The step of generating operational analysis information is not merely to store the aforementioned integrated operational data, but to generate analysis result values ​​capable of estimating the current and future states of the solar assets.

[0037] Here, operational analysis information may be intermediate information including the synchronization results of digital twin objects, estimated values ​​of degradation progression of target parts, the state error rate relative to the normal operation profile, remaining lifespan per part, failure probability by time interval, and derived analysis values ​​necessary for calculating replacement priorities. In other words, the operational analysis information can function as judgment-based data for generating subsequent operational execution information, and can be distinguished from raw integrated operational data in that it is information reflecting interpretation, estimation, prediction, and comparison operations, rather than simple measurements.

[0038] The step of generating operational analysis information includes synchronizing a digital twin object that replicates the physical specifications and real-time degradation status of the solar assets in a virtual space based on the integrated operational data.

[0039] The step of synchronizing digital twin objects is a step of maintaining a virtual asset model within the device that corresponds to a physical solar power generation facility existing in the real world, and updating the structure, state, and operation parameters of the corresponding virtual asset to the latest state by reflecting actual sensor data and design information.

[0040] Here, physical specifications may refer to design and construction information such as panel rated capacity, cell array structure, string configuration, inverter rating, cooling structure, cable resistance, installation angle, azimuth angle, installation location, and component material characteristics. Additionally, real-time degradation status may refer to state values ​​indicating the current level of aging of the actual asset, such as the rate of output degradation over time, the increase in internal resistance, changes in junction temperature, insulation performance degradation, the trend of increasing switching losses, and the degree of capacitor capacity degradation. In other words, the digital twin object can be implemented as a virtual twin model that software-simulates not only the initial physical specifications of the target solar asset but also the accumulated damage or component degradation status from the start of operation to the present time.

[0041] In the step of synchronizing the digital twin object, the device can update the state variables of the virtual object by utilizing design drawings, commissioning history, initial installation environment information, and real-time power generation status data together. For example, the performance degradation factor of the panel module can be corrected by comparing the standard performance value at the beginning of installation with the current output curve, and the thermal stress state of the power semiconductor can be reflected as a state variable within the virtual object by using temperature sensor values ​​inside the inverter, output ripple data, and load rate fluctuation values. Here, synchronization may include a process of recalibrating the parameters of the virtual object to minimize the difference between actual measured values ​​and the predicted values ​​of the virtual model, rather than simple replication. Accordingly, the digital twin object can function as a virtual operator that simulates the current operating status of the actual solar asset in real-time or near-real-time.

[0042] The step of performing an accelerated degradation simulation of a target component by chronologically inputting weather environment data and asset operation parameters into the digital twin object, and deriving the remaining lifespan and failure probability for each component by calculating the state error rate with a preset normal operation profile, is a step of predicting how quickly a specific component will wear out or fail in the future by reflecting expected future weather and operating conditions in a synchronized virtual model.

[0043] Here, asset operating parameters may refer to variables representing the actual operating conditions of a photovoltaic asset, such as output load factor, switching frequency, operating time, partial load operation ratio, number of peak operation repetitions, cooling status, and reverse flow pattern. Additionally, a normal operating profile may be a reference operating model established based on the manufacturer's shipment specifications, reference performance measured at initial installation, expected output curve in a standard environment, normal internal resistance range, and allowable temperature range.

[0044] In the step of performing accelerated degradation simulation, the device can calculate the degradation trajectory for each component by setting the component state at the current point in time as the initial condition and reflecting the weather and operating conditions expected over a certain period in the future. For example, for panel modules, it can estimate how much potential-induced degradation or output reduction progresses when repeated high-temperature and high-humidity environments and UV exposure accumulate; for power semiconductors, it can predict how much thermal fatigue life decreases based on the cumulative effects of junction temperature swing and switching frequency. For DC link capacitors, the rate of capacitance reduction due to ripple current and internal temperature rise can also be simulated. In this way, the device can predict future states by performing accelerated degradation simulations that reflect different physical, chemical, or electrical degradation mechanisms for each target component.

[0045] The step of calculating the state error rate is a step of quantifying how much the target component has deviated from the normal state at the present or future point in time by comparing the future state value derived from the accelerated degradation simulation results with the normal operation profile. For example, if the expected power generation efficiency of a panel decreases by 8 percent compared to the normal standard and the internal resistance increases by 15 percent, the device can calculate the state error rate by aggregating these deviations. Here, the state error rate may be a difference for a single parameter or a composite deviation index calculated by assigning weights to multiple state variables. Based on the state error rate, the device can estimate the point in time when each component is expected to reach the performance guarantee limit or the failure critical condition, and calculate the time remaining until that point as the remaining lifespan. Additionally, the device can derive the failure probability by calculating the probability of failure occurring within a specific time interval from the same simulation results in the form of a probability.

[0046] The step of generating operational execution information is a step of generating execution decision information to be reflected in actual asset operations and contract operations based on the above operational analysis information.

[0047] Here, operational execution information may be information including at least one of an asset operation plan, a resource control plan, a settlement plan, a revenue distribution plan, a maintenance plan, and a contract management plan. That is, the operational execution information may not be a simple analysis result, but rather an execution instruction or candidate execution plan indicating what measures to take in the actual field, when, and on what criteria based on the analysis result. For example, the timing of replacement of a specific panel string, the timing of output limiting for a specific inverter channel, adjustment of settlement criteria, preparation of a maintenance budget, and service response policies for each subscriber may be included in the operational execution information.

[0048] The step of calculating the physical replacement cost of the target component, the estimated loss in power sales due to failure, and the remaining subscription contract period of the subscriber based on the remaining lifespan and failure probability is a step of quantifying cost, loss, and contract factors to determine the economic feasibility of the component replacement time.

[0049] Here, physical replacement costs may include the unit purchase price of parts, transportation costs, on-site dispatch fees, labor costs, installation and re-commissioning costs, and work loss costs resulting from downtime. Estimated power sales loss may refer to future losses calculated by applying hourly power sales rates to the reduced power generation resulting from the failure of specific parts or severe performance degradation. Additionally, the remaining subscription contract period refers to the time remaining until the contract expiration date for the subscriber associated with the solar asset, signifying the period during which revenue can be recovered through parts replacement. In particular, given the nature of service models where solar assets are leased or utilized via subscription for a fixed period, when contract expiration is imminent, financial response strategies such as delaying the failure date until after the contract ends through output restriction, rather than immediately replacing high-cost parts, may be considered.

[0050] For example, if a power semiconductor has a remaining lifespan of 4 months, a replacement cost of 1.5 million won, an estimated cumulative loss of 2.2 million won in power sales over the next 6 months in the event of failure, and a remaining subscription contract period of 14 months, the device may determine that the economic disadvantage resulting from delaying replacement is greater. Conversely, if the replacement cost of a DC link capacitor is 800,000 won but the estimated loss is only 200,000 won and the remaining contract period is 2 months, the device may determine that bulk maintenance after the contract ends is more advantageous than immediate replacement. In this way, the device can derive economic indicators for each component by considering costs, losses, and contract periods together.

[0051] The step of determining the optimal replacement time for each component to maximize economic returns over the entire subscription cycle is to comprehensively compare the calculated remaining lifespan, failure probability, physical replacement cost, expected power sales loss, and remaining subscription contract period, and to select the time that provides the highest expected return among the replacement time candidates for each target component.

[0052] Here, the optimal replacement time does not simply mean the earliest or latest replacement time, but rather the time when the balance between the expenditure of replacement costs and the benefits of avoiding failure risk is best. For example, the device can calculate the expected net profit or expected loss avoidance for multiple replacement candidate times on a monthly or weekly basis, and determine the time when the net present value is maximized as the optimal replacement time. For instance, if the probability of failure of a specific internal inverter component is expected to be 80 percent next month, and the expected loss in power sales if the failure occurs and the power plant operation is suspended for 3 days is 1 million won, and the physical replacement cost to replace it before the failure is 300,000 won, then since the expected loss of 800,000 won is greater than the replacement cost of 300,000 won, preemptive replacement can be determined as the optimal replacement time that maximizes overall economic profit.

[0053] In the step of determining the optimal replacement timing, interactions between multiple components within the same asset can also be considered. For example, as panel module degradation intensifies, it can affect the operating efficiency of the inverter, and as power semiconductor degradation progresses, the stress on the DC link capacitor may increase. Therefore, the device can compare not only the optimal timing for individual components but also the total costs and total benefits of replacing multiple components as a bundle. Furthermore, since the optimal replacement timing can be integrated with not only maintenance plans but also asset operation plans, resource control plans, settlement plans, revenue distribution plans, and contract management plans, the service stability and profitability of the entire subscription-based solar asset can be improved simultaneously.

[0054] In addition, the step of deriving the remaining lifespan and failure probability for each of the above components comprises: defining the solar panel module, power semiconductor, and DC link capacitor as target components; extracting meteorological factors from the meteorological environment data, including cumulative UV exposure, temperature variation range, humidity variation range, number of temperature and humidity cycles, wind speed variation, and wind load variation; extracting operational factors from the asset operation parameters and real-time power generation status data, including DC-side voltage / current (IV) trajectory, AC-side voltage / current waveform, output load rate, switching frequency, and conversion efficiency variation; and mapping the meteorological factors and operational factors as stress factors representing thermal stress, electrical stress, mechanical stress, and environmental stress for each target component with respect to the Electrical Equivalent Circuit Model of the solar panel module and the Multiphysics Thermal Model of the power conversion inverter constituting the digital twin object. Based on the above stress factors, a step of performing a physics-based first-order accelerated simulation for a solar panel module to calculate potential-induced degradation (PID), micro-crack progression rate, and series resistance increase; a step of performing a data-driven second-order simulation to estimate the cumulative residual of non-linear thermal fatigue of inverter internal power semiconductors (IGBTs) and DC-Link Capacitors by extracting high-frequency noise components, switching ripple components, harmonic distortion components, and instantaneous temperature fluctuation components from the real-time power generation status data and inputting them into a time-series deep learning model; and a step of calculating an ensemble future degradation trajectory for each target component by synchronizing the results of the first-order accelerated simulation and the second-order simulation on the same time axis and weightedly combining them.The method may include: a step of generating an expected power generation efficiency curve, an expected output deviation curve, and an expected internal resistance increase curve for each target component from the above-mentioned ensemble future degradation trajectory, and calculating a Dynamic SoH Error Rate by comparing them with the above-mentioned normal operation profile generated based on the manufacturer's shipment specifications and the initial installation environment of the solar asset; a step of correcting a performance curve corresponding to the degradation characteristics of each target component among the above-mentioned power generation efficiency curve, expected output deviation curve, and expected internal resistance increase curve for each target component based on the above-mentioned Dynamic SoH Error Rate; and a step of deriving a temporal confidence interval of the remaining lifespan for each target component and a failure probability for each time interval by integrating the probability density function (PDF) at the point in time when the performance curve corresponding to the degradation characteristics of each target component does not satisfy the performance guarantee threshold condition corresponding to each target component.

[0055] The step of deriving the remaining lifespan and failure probability for each component described above may be a step of selecting components among multiple core components constituting a subscription-type solar asset that differ in their degradation progression patterns, failure mechanisms, and maintenance cost structures, and more precisely predicting the future state by applying different degradation estimation models for each component type.

[0056] Here, the solar panel module is a core component that converts light energy into DC electrical energy and may be a component susceptible to output degradation, potential-induced degradation, and mechanical cracking due to long-term exposure to the external environment. The power semiconductor may include switching elements, such as IGBTs, that convert DC to AC within the inverter and may be a component susceptible to the accumulation of junction fatigue due to repetitive switching operations and thermal cycles. Additionally, the DC link capacitor is a capacitor that smooths DC voltage ripples within the inverter and may be a component susceptible to capacitance reduction and equivalent series resistance increase resulting from ripple current, internal temperature rise, and degradation of dielectric properties. Accordingly, by defining the solar panel module, the power semiconductor, and the DC link capacitor as target components, the device can intensively identify core components that can have a critical impact on power generation losses within the same asset as targets for life management, and can individually track and predict components with different degradation mechanisms.

[0057] The step of extracting meteorological factors including cumulative UV exposure, temperature variation range, humidity variation range, number of temperature and humidity repetition cycles, wind speed variation, and wind load variation from the above meteorological environment data may be a step of quantifying and extracting external stress factors that cause physical and chemical wear of the parts, going beyond simply collecting weather information.

[0058] Here, the above cumulative UV exposure amount may represent the total amount of UV energy irradiated onto the surface or backsheet of a solar panel module over a certain period, the above temperature variation range may be calculated based on the difference between the highest and lowest temperatures on a daily or hourly basis, and the above humidity variation range may refer to the range of fluctuation in relative humidity. Additionally, the above number of temperature and humidity repetition cycles may refer to the number of times the temperature and humidity repeatedly rise and fall above a predetermined threshold, for example, it may be calculated as the number of times the temperature fluctuates between 10 and 30 degrees and the phenomenon of dew forming and drying repeats during a day. The above wind speed variation and wind load variation may represent the amount of physical pressure change acting on the panel structure due to strong winds or gusts, and may be utilized as external factors that exacerbate mechanical fatigue. For example, the device may analyze weather data collected over the past 30 or 90 days in a window unit to calculate each of the above weather factors in the form of an average value, maximum value, cumulative value, or number of repetitions.

[0059] The step of extracting operating factors including the DC side voltage / current (IV) trajectory, AC side voltage / current waveform, output load rate, switching frequency, and conversion efficiency fluctuation amount from the above asset operating parameters and real-time power generation status data may be a step of quantifying an internal operating load indicator that indicates how severely the solar asset is actually operating under conditions other than external weather factors.

[0060] Here, the DC-side voltage and current trajectory may refer to the time-series change path of voltage and current measured in the panel string or DC link section, and the AC-side voltage and current waveform may be waveform data measured at the inverter output terminal. Additionally, the output load rate may be calculated as the ratio of actual output to rated capacity, the switching frequency may represent the driving frequency of the power semiconductor, and the conversion efficiency fluctuation amount may be a value that tracks the efficiency change of AC output relative to DC input over time. In this case, the device can extract not only the simple average value but also the peak value, ripple magnitude, waveform distortion rate, and frequency of occurrence of sudden fluctuation events, and, for example, if the output load rate is maintained at a state close to 100 percent for a long time, it can be used as an operating factor indicating that high heat has accumulated in the internal components.

[0061] The step of mapping the above weather factors and the above operating factors to stress factors representing thermal stress, electrical stress, mechanical stress, and environmental stress for each target component with respect to the electrical equivalent circuit model of the solar panel module constituting the digital twin object and the multiphysics thermal analysis model of the power conversion inverter may be a step of converting the collected weather information and operating information into damage input values ​​that can be directly computed in each component model within a virtual space.

[0062] Here, the electrical equivalent circuit model may be a model that simulates the operation of a photovoltaic panel module by substituting it with circuit elements such as resistors, current sources, and capacitors, and the multiphysics thermal analysis model may be a virtual thermal distribution model that comprehensively analyzes the current flow, power loss, and heat generation and dissipation processes within the inverter. For example, the cumulative UV exposure amount and humidity variation range may be mapped to the environmental stress coefficient of the panel module, the abnormal deviation of the IV trajectory and output load rate may be converted into an electrical stress coefficient, and the temperature variation range and junction temperature rise amount may be converted into a thermal stress coefficient. In addition, the wind load fluctuation value and repetitive vibration component may be reflected as a mechanical stress coefficient, and the switching frequency may be linked to the thermal and electrical stress of the power semiconductor and converted into an accurate computational input value for the simulation. The mapping process may be performed using linear transformation, nonlinear function transformation, lookup table reference, or a pre-trained regression model, and different weighting schemes may be applied depending on the type of target component.

[0063] The step of performing a physics-based first-order accelerated simulation to calculate potential-induced degradation, microcrack propagation rate, and series resistance increase for a solar panel module based on the above stress factors may be a step of rapidly predicting chemical and mechanical aging phenomena occurring internally by reflecting the environmental and mechanical stresses received by the panel module into a mathematical physics model.

[0064] Here, the potential-induced degradation may refer to a phenomenon in which insulation performance deteriorates and output decreases due to leakage current occurring between the panel surface and the frame when high-voltage bias, humidity, and temperature environments act in combination, and the microcrack may refer to microcracks occurring inside the cell due to repetitive thermal expansion, contraction, and wind load. Additionally, the increase in series resistance may be caused by reduced wiring contact, cell cracking, and deterioration of solder joints. The device can simulate changes in open-circuit voltage, short-circuit current, maximum power point, and resistance parameters by inputting stress factors into the electrical equivalent circuit model, and by accumulating these changes over a time axis, it can calculate future degradation trends by compressing the aging process of a panel module spanning decades into a short-term computation.

[0065] The step of performing a data-based secondary simulation to estimate the nonlinear thermal fatigue accumulation residual of the inverter's internal power semiconductor and DC link capacitor by extracting high-frequency noise components, switching ripple components, harmonic distortion components, and instantaneous temperature fluctuation components from the above real-time power generation status data and inputting them into a time-series deep learning model may be a step of predicting the minute fatigue accumulation patterns of the inverter's internal electronic components, which are difficult to explain solely by physical law equations, based on data patterns.

[0066] Here, the high-frequency noise component may refer to high-speed electrical disturbances occurring during the switching operation, the switching ripple component may be periodic fluctuations superimposed on the DC link or output waveform, and the harmonic distortion component may be a waveform distortion indicator associated with power quality degradation. Additionally, the instantaneous temperature fluctuation component may refer to a temperature rise or fall event occurring rapidly within a short period of time.

[0067] Due to complex switching operations, these micro-electric shocks continuously accumulate inside the inverter, and invisible thermal fatigue damage can accumulate non-linearly in the power semiconductor junctions or the insulators inside the DC link capacitors. Therefore, the device can input current noise and temperature data into a time-series deep learning model—such as an LSTM, GRU, Transformer series, or a combination thereof—that has been pre-trained on past failure data patterns, and estimate the non-linear thermal fatigue accumulation residual corresponding to the extent of damage already lost from the remaining capacity based on the data.

[0068] The step of synchronizing the results of the first acceleration simulation and the results of the second simulation on the same time axis and weightedly combining them to calculate the ensemble future degradation trajectory for each target part may be a step of integrating the physics-based prediction and the data-based prediction in a complementary manner to organize them into a single comprehensive degradation map.

[0069] Here, synchronization refers to the process of rearranging simulation results on the same time grid and interpolating or resampling them if necessary, as each simulation result may have a different sampling period or a different reference point. Additionally, weighted combination refers to the process of generating an integrated degradation trend by assigning different weights based on the reliability, past prediction error, current state fit, or type of target component of each simulation. For example, a relatively high weight may be applied to the results of the physics-based first-order accelerated simulation for panel modules, while a relatively high weight may be applied to the results of the data-based second-order simulation for power semiconductors and DC link capacitors. Furthermore, an additionally high combined weight may be assigned to the two results during time periods with significant mutual influence, such as when panel aging increases the electrical burden on the inverter. The ensemble future degradation trajectory calculated in this manner can serve as a comprehensive aging map that integrally shows how the condition of individual components deteriorates over time.

[0070] The step of generating expected power generation efficiency curves, expected output deviation curves, and expected internal resistance increase curves for each target component from the above ensemble future degradation trajectory, and calculating a dynamic state error rate by comparing them with a normal operation profile generated based on the manufacturer's shipment specifications and the initial installation environment of the solar asset, may be a step of converting the above comprehensive aging map into an indicator that can be used to determine actual performance degradation.

[0071] Here, the predicted power generation efficiency curve may indicate a trend of decreasing power generation efficiency over time, the predicted output deviation curve may indicate an increasing trend of future output deviation relative to normal output, and the predicted internal resistance increase curve may indicate an increasing trend of electrical loss factors within the component. The normal operation profile may include an ideal power generation curve, output stability range, and allowable resistance range expected when the same equipment is in a new product state and receives the same sunlight and temperature conditions, and the device may calculate a dynamic state error rate by converting the difference between the predicted curves and the normal operation profile into a composite indicator. For example, if the predicted power generation efficiency curve at a specific future point in time represents a level of 90 percent compared to the normal profile, the device may calculate the difference of 10 percent as the dynamic state error rate, and the dynamic state error rate may be continuously updated over time according to real-time external environmental changes and values ​​reflecting the current state.

[0072] The step of correcting the performance curve corresponding to the degradation characteristics of each target component among the expected power generation efficiency curve, expected output deviation curve, and expected internal resistance increase curve for each target component based on the dynamic state error rate may be a step of selecting the most critical degradation performance indicator for each component and precisely correcting the prediction curve to better reflect reality by reflecting the difference between the current state and the future state prediction. For example, in the case of a solar panel module, the expected power generation efficiency curve may be used as the main performance curve; in the case of a power semiconductor, the expected output deviation curve or the internal resistance increase curve may be used as the main performance curve; and in the case of a DC link capacitor, both output deviation and internal resistance increase may be considered. The correction may be performed by accelerating the degradation slope when the dynamic state error rate is large, or conversely, by smoothing the slope of the prediction curve when the error rate is low and current state stability is high. For example, if it is determined that the rate of increase in the internal resistance of a DC link capacitor is faster in the actual operating environment than in the existing prediction, the device may correct the slope of the expected internal resistance increase curve to a steeper rising curve by reflecting the error rate.

[0073] The step of deriving the temporal confidence interval of the remaining life for each target component and the failure probability for each time interval by integrating the probability density function at the point in time when the performance curve corresponding to the degradation characteristics of each target component does not satisfy the performance guarantee threshold condition corresponding to each target component may be a step of converting the corrected future performance curve into probability information that can be directly used for maintenance decision-making.

[0074] Here, the performance guarantee threshold condition may include at least one of the manufacturer's warranty standard, contractual minimum performance maintenance condition, maximum output deviation acceptable for system operation, or internal resistance limit. The device may generate a distribution at the point in time when the performance curve falls below the threshold condition using Monte Carlo sampling, bootstrap estimation, or a stochastic transfer model, and may calculate the probability of failure within a given interval by integrating the probability density function of the distribution over specific time intervals. Additionally, the device may derive a temporal confidence interval for the remaining life of each target component using the median, upper and lower percentiles, or confidence level intervals of the distribution. For example, if the probability that the expected power generation efficiency curve of a solar panel module will fall below the performance guarantee threshold efficiency after 8 months is 60 percent and increases to 85 percent after 10 months, the device may suggest the 8 to 10-month interval as a priority inspection or replacement candidate interval for the panel module. In addition, if the probability that the internal resistance increase curve of a DC link capacitor will exceed the allowable resistance limit after 5 months is 40 percent and after 6 months is 70 percent, the device can estimate the remaining life of the component within a confidence range of 5 to 6 months. In this way, the device can provide a basis for quantitative judgment on which component among multiple components within the same asset should be inspected or replaced first at a specific time by utilizing the temporal confidence interval of the remaining life of each component and the failure probability for each time interval.

[0075] And, the step of calculating the ensemble future degradation trajectory comprises: generating a first degradation state vector including an expected power generation efficiency degradation rate, an expected series resistance increase rate, an expected cell damage area ratio, and an expected insulation performance degradation rate for each solar panel module from the results of the first acceleration simulation; generating a second degradation state vector including an expected junction temperature swing, an expected conduction loss increase rate, an expected switching loss increase rate for each power semiconductor, and an expected capacitance degradation rate, an expected equivalent series resistance increase rate, and an expected ripple tolerance degradation rate for each DC link capacitor from the results of the second simulation; rearranging the first degradation state vector and the second degradation state vector on the same time grid at preset sampling cycles and performing interpolation operations on missing time points to generate a time-series degradation matrix for each target component; and calculating reliability parameters including an error variance, a residual autocorrelation coefficient, a prediction uncertainty, and a recent time point response sensitivity for each target component based on the difference between the time-series degradation matrix and the normal operation profile. Based on the above reliability parameters, a step of calculating a first weight for the results of a first acceleration simulation for the solar panel module and a second weight for the results of a second simulation for the power semiconductor and DC link capacitor, and normalizing the first weight and the second weight to generate a weight vector for each target component; a step of generating an ensemble degradation state vector for each target component by linearly combining each degradation component of a time-series degradation matrix using the weight vector, and generating a degradation transition covariance matrix based on the time-series variance and covariance of the ensemble degradation state vector; a step of inputting the ensemble degradation state vector and the degradation transition covariance matrix into an ensemble Kalman filter to sequentially correct the time-series degradation state, and performing Monte Carlo sampling on the corrected degradation state to generate a plurality of future degradation scenarios;and may include the step of deriving an ensemble future degradation trajectory for each target component, including an expected power generation efficiency degradation curve, an expected output deviation curve, an expected internal resistance increase curve, and an expected thermal fatigue accumulation curve for each of the photovoltaic panel module, power semiconductor, and DC link capacitor, by superimposing the above future degradation scenarios on a time axis.;

[0076] The step of calculating the above-mentioned ensemble future degradation trajectory may be a step of integrating the results of physics-based degradation analysis and the results of data-based degradation estimation within the same analysis framework, and probabilistically predicting the future degradation behavior of each target component by reflecting the reliability and prediction uncertainty of each result.

[0077] Here, the above-mentioned ensemble future degradation trajectory is not a single degradation prediction value at a specific point in time, but may be a set of prediction paths that chronologically represent how the performance degradation, output fluctuation, internal resistance change, and thermal fatigue accumulation patterns of the target component will unfold over multiple future points in time. That is, the above-mentioned ensemble future degradation trajectory may be the result of organizing different degradation patterns for solar panel modules, power semiconductors, and DC link capacitors into a single integrated prediction frame, and may be utilized as a key input value for subsequent calculation of dynamic state error rates, performance curve correction, remaining life estimation, and failure probability calculation.

[0078] In the step of generating a first degradation state vector including the expected power generation efficiency degradation rate, expected series resistance increase rate, expected cell damage area ratio, and expected insulation performance degradation rate per solar panel module from the above first acceleration simulation results, the previously performed physics-based simulation results can be structured into an array form having mathematical direction and size.

[0079] Here, the first degradation state vector may refer to a data unit that groups physical aging indicators specific to a solar panel module into a single numerical group. The predicted power generation efficiency degradation rate may indicate how much the ability to convert light energy into electrical energy will decrease in the future, and the predicted series resistance increase rate may indicate the degree of increase in losses in the internal path through which current flows. Additionally, the predicted cell damage area ratio may indicate the proportion of the total cell area occupied by microcracks or inactive regions within the cell, and the predicted insulation performance degradation rate may be an indicator indicating how much insulation performance will weaken due to moisture penetration, UV exposure, or encapsulation material degradation. The device extracts these elements from the results of the first accelerated simulation and constructs them in the form of a vector, for example, with the predicted power generation efficiency degradation rate, the predicted series resistance increase rate, the predicted cell damage area ratio, and the predicted insulation performance degradation rate as elements, thereby expressing the degradation state of the panel module as a combination of multiple characteristics rather than a single indicator.

[0080] In the step of generating a second degradation state vector including the expected junction temperature swing, expected conduction loss growth rate, expected switching loss growth rate, and expected capacitance degradation rate, expected equivalent series resistance growth rate, and expected ripple tolerance degradation rate for each DC link capacitor from the above second simulation results, the electrical and thermal fatigue indicators of the core electronic components inside the inverter obtained through data-based deep learning simulation can be structured into a separate numerical array.

[0081] Here, the second degradation state vector may refer to a set of indicators representing the degradation of the state of the power semiconductor and DC link capacitor responsible for power conversion within the inverter. The predicted junction temperature swing of the power semiconductor may represent the range of the rise and fall in junction temperature that occurs repeatedly when the device is turned on and off, and the predicted conduction loss growth rate and the predicted switching loss growth rate may indicate how much the loss consumed as thermal energy increases when power passes through the device or is switched. Additionally, the predicted capacitance degradation rate of the DC link capacitor may indicate how much the basic ability to store and smooth charge will decrease, the predicted equivalent series resistance growth rate may indicate how much the parasitic resistance value causing internal heat generation will increase, and the predicted ripple tolerance degradation rate may indicate how much the allowable limit for withstanding microcurrent fluctuations will decrease. The device may use the results of the second simulation to generate vectors for the power semiconductor and the DC link capacitor separately, or integrate them into a single expanded vector if necessary.

[0082] In the step of rearranging the first degradation state vector and the second degradation state vector on the same time grid at each preset sampling period and performing interpolation operations on missing time points to generate a time-series degradation matrix for each target part, a synchronization operation can be performed to align physical-based data and data-based predicted values, which have different periods and result formats, onto a single common timetable. For example, since the physical-based simulation is calculated on a daily basis and the data-based simulation can be calculated on an hourly basis or based on the time of an event occurrence, the device can unify them on the same time grid of a preset sampling period, such as on a 1-hour basis, a 6-hour basis, or a 1-day basis. In addition, if some degradation component values ​​are missing at a specific time point, gaps in the data over time can be filled by performing linear interpolation, spline interpolation, Kalman predicted value interpolation, or adjacent time point average interpolation. As a result, the device can generate a time-series degradation matrix for each target component, where the row direction represents the time point and the column direction represents the degradation component, which can be utilized as a common representation for performing subsequent reliability evaluation and weighted combination.

[0083] In the step of calculating reliability parameters including error variance per target part, residual autocorrelation coefficient, prediction uncertainty, and response sensitivity at the most recent time point based on the difference between the above-mentioned time-series degradation matrix and the normal operation profile, it is possible to quantitatively evaluate how stable and reliable the completed time-axis-based prediction data is compared to the normal operation standard.

[0084] Here, the error variance can indicate how widely the predicted values ​​deviate from the normal operating profile, and the residual autocorrelation coefficient can indicate how continuously past prediction errors are linked to current and future errors. Additionally, the prediction uncertainty can be calculated based on the dispersion between multiple prediction paths, the width of the model output confidence interval, or the accumulated value of past validation errors, and the recent time point response sensitivity can indicate how sensitively the degradation state of the corresponding component responds to recent environmental or operational stress changes. For example, if the recent time point response sensitivity is excessively high and the error variance is also large, the prediction result can be judged as unstable; conversely, if the error variance is small, the residual autocorrelation coefficient is low, and the response sensitivity is within an appropriate range, the result can be judged as highly reliable.

[0085] Based on the reliability parameter above, a first weight is calculated for the results of a first accelerated simulation for solar panel modules, and a second weight is calculated for the results of a second simulation for power semiconductors and DC link capacitors. In the step of generating a weight vector for each target component by normalizing the first and second weights, the final fusion ratio can be determined by numerically evaluating which method of simulation results better reflects reality and is more reliable for each component. For example, a higher first weight can be assigned to the physics-based first accelerated simulation results for solar panel modules, which exhibit significant long-term environmental exposure and structural wear, and a higher second weight can be assigned to the data-based second simulation results for power semiconductors and DC link capacitors, which are sensitive to complex electrical noise and non-linear heat generation. Subsequently, by normalizing the sum of the first and second weights to be 1 or 100 percent, a weight vector for each target component can be generated to determine the ratio at which the two simulation results are combined. If necessary, the weight vector can be extended to include detailed weights at the degradation component level as well as at the component level.

[0086] In the step of generating an ensemble degradation state vector for each target part by linearly combining each degradation component of the time-series degradation matrix using the above weight vector, and generating a degradation transition covariance matrix based on the time-series variance and covariance of the ensemble degradation state vector, the physical-based results and data-based results are fused into a single integrated degradation state according to the previously determined reflection ratio, and a statistical matrix can be constructed that indicates how the fused data spreads and interacts with one another over time.

[0087] Here, a linear combination may refer to a mathematical operation that generates an ensemble degradation state vector, which is a single integrated result, by summing the value obtained by multiplying the first-order model result by a first weight and the value obtained by multiplying the second-order model result by a second weight. Additionally, the degradation transition covariance matrix may be a matrix that records together variance indicating how much today's degradation state can spread at a later point in time tomorrow or thereafter, and covariance indicating how much a specific degradation component mutually influences other degradation components. For example, the higher the correlation between an increase in series resistance and a decrease in power generation efficiency, the larger the corresponding covariance value may be calculated; this can serve as basic data to reflect the interlocking structure between each component when predicting future degradation states.

[0088] In the step of inputting the above ensemble degradation state vector and degradation transition covariance matrix into an ensemble Kalman filter to sequentially correct the degradation state at each time point and performing Monte Carlo sampling on the corrected degradation state to generate multiple future degradation scenarios, the future degradation trajectory can be refined more precisely and various possibilities can be virtually expanded by using a mathematical filter algorithm and a probability extraction technique.

[0089] Here, the ensemble Kalman filter may be a filter that sequentially estimates the most realistic state value while removing noise and uncertainty contained in past observation data and current prediction data. Accordingly, the device can continuously modify and correct the ensemble degradation state vector whenever additional real-time power generation state data is collected. Subsequently, the device can perform Monte Carlo sampling using the corrected degradation state as the initial state, thereby generating tens to thousands of future degradation scenarios that reflect stochastic disturbances such as heatwaves, sudden load increases, sudden weather changes, or changes in operating conditions. Each scenario may be a virtual path indicating the speed and trend at which the same component may degrade under future weather and operating variations.

[0090] In the step of deriving an ensemble future degradation trajectory for each target component, including an expected power generation efficiency degradation curve, an expected output deviation curve, an expected internal resistance increase curve, and an expected thermal fatigue accumulation curve for each of the solar panel module, power semiconductor, and DC link capacitor, by superimposing the above-mentioned future degradation scenarios on a time axis, a plurality of virtual future scenarios can be superimposed on a single time graph to derive the representative degradation path with the highest probability of occurrence and its distribution range. For example, if numerous scenario curves generated by Monte Carlo sampling are superimposed on the same time axis, such as a period of 5 or 10 years in the future, sections where multiple curves are dense and sections where they are relatively dispersed can be distinguished. The device can generate the expected power generation efficiency degradation curve, the expected output deviation curve, the expected internal resistance increase curve, and the expected thermal fatigue accumulation curve using the dense section as a central trend line, and may include upper and lower confidence bands as needed. The ensemble future degradation trajectory for each target component derived in this way can be directly utilized in subsequent steps by comparing it with a normal operating profile to calculate the dynamic state error rate, correct the performance curve, and calculate the temporal confidence interval of the remaining life and the failure probability for each time interval, thereby providing a quantitative basis for more reliably determining the maintenance priority and optimal replacement time for each target component.

[0091] According to one embodiment, a method for predicting real-time expected revenue of a rental solar power plant through the fusion of weather data by multiple building types, wherein the device includes a processor, memory, a communication module, and a non-transient storage medium, and the processor executes a program stored in the non-transient storage medium, comprises: a step of collecting integrated operational data including weather environment data, power market indicators, and real-time power generation status data of a solar asset; a step of analyzing the integrated operational data to generate operational analysis information; and a step of generating operational execution information including at least one of an asset operation plan, a resource control plan, a settlement plan, a revenue distribution plan, a maintenance plan, and a contract management plan based on the operational analysis information; wherein the step of generating operational analysis information comprises: a step of aligning building type information, weather environment data, and real-time power generation status data included in the integrated operational data to generate a weather-building fusion profile reflecting solar radiation acceptance characteristics, shading effect characteristics, load linkage characteristics, and power generation efficiency characteristics by time of day by building type. The present invention provides a method for predicting real-time expected profits of a rental solar power plant through the fusion of weather data by multiple building types, comprising: a step of calculating real-time expected profit information including expected power generation amount by time of day, expected power sales unit price, expected self-consumption savings, and expected power sales revenue by combining power market indicators with the weather-building fusion profile; and a step of generating operation execution information including: a step of generating operation standard information including operation priority, output control standard, settlement standard, and profit distribution standard for each solar asset based on the real-time expected profit information; and a step of determining an asset operation plan and a contract management plan by building type to maximize expected profits over a predetermined period based on the real-time expected profit information and the operation standard information.

[0092] The step of collecting integrated operation data is to collect external environmental information, power trading information, and facility operation information in a form that can be interconnected in order to analyze the profitability of rental solar power plants in real time.

[0093] Here, integrated operational data may refer to a data set in which data collected from different sources is sorted according to the same time standard, the same asset identification system, and the same building identification system. For example, the integrated operational data may include meteorological environmental data such as solar radiation, ambient temperature, relative humidity, cloud cover, wind speed, and wind direction received from a weather observation server or weather forecast server; power market indicators such as real-time system marginal price, time-of-day retail electricity rates, trading unit prices for renewable energy certificates, standard unit prices for peak power basic rates, and rate tables by contract type received from a power exchange or settlement server; and real-time power generation status data such as DC-side voltage, DC-side current, AC-side output, instantaneous generation, cumulative generation, backflow power, panel surface temperature, and inverter efficiency collected from inverters, junction boxes, power meters, panel sensors, or monitoring gateways. In addition, the integrated operation data can be mapped to building type information already stored in a database to generate a weather-building fusion profile in a subsequent step, and the building type information may include the use, roof structure, installation location, contract type, past power consumption history, and information related to surrounding obstacles of the target building.

[0094] In the step of collecting integrated operational data, time reference unification, missing value correction, outlier removal, sampling period reordering, and asset unique identifier mapping can be performed on the raw data received from each of the above data sources. For example, since weather environment data may be updated every 10 minutes and real-time power generation status data may be updated every 1 second or 1 minute, the device can resample the data based on an operational analysis cycle of a predetermined reference time grid, such as every 5 minutes, every 15 minutes, or every payment interval, to reconstruct it into a single time-series input structure. In addition, if the same solar asset consists of multiple panel strings, multiple inverter channels, or multiple sub-equipment modules, the device can generate structured data suitable for subsequent revenue analysis by aggregating or separately storing measurement values ​​for each sub-equipment at the solar asset unit, building unit, or contract unit level. The integrated operational data aligned in this way can be utilized as input values ​​for the subsequent operational analysis information generation step.

[0095] The step of generating operational analysis information is not merely to store the aforementioned integrated operational data, but to generate analysis results capable of determining what power generation performance and economic value the solar assets installed in the target building will have in the present and future.

[0096] Here, operational analysis information may refer to intermediate analysis information that includes a weather-building fusion profile reflecting solar radiation acceptance characteristics, shading effect characteristics, load linkage characteristics, and power generation efficiency characteristics by time of day, combined with real-time expected revenue information calculated by combining this profile with power market indicators. In other words, the operational analysis information can be distinguished from integrated operational data in that it is not the raw measured values ​​or raw weather values ​​themselves, but rather information serving as a basis for judgment generated through the alignment, model computation, prediction, comparison, and combination processing of multiple data.

[0097] The step of generating a weather-building fusion profile is to convert the structural characteristics of the target building, external environment characteristics, and actual solar asset operation characteristics into a single integrated feature set.

[0098] Here, the weather-building fusion profile refers to a profile that combines building-specific solar radiation acceptance characteristics, shading effect characteristics, load linkage characteristics, and time-of-day power generation efficiency characteristics in the form of a multidimensional feature vector. This profile is designed to reflect the fact that even under identical weather conditions, the actual potential for power generation and self-consumption may vary depending on the building type, roof structure, surrounding obstacles, and power consumption patterns. Building type information may include data such as roof shape, floor height, and the presence of surrounding high-rise structures, along with usage information such as whether the target building is a factory, commercial building, or apartment. Solar radiation acceptance characteristics may refer to how efficiently sunlight is received depending on the roof's slope or azimuth angle, while shading effect characteristics may be a numerical representation of the degree of shadowing cast on the panels by surrounding buildings, trees, or rooftop structures over time. Additionally, load linkage characteristics may indicate the extent to which the solar power generation time zone overlaps with the actual consumption time zone, depending on whether the building has high power consumption during the day, such as a factory, or high loads during the evening, such as a residential building.

[0099] In the step of generating a weather-building fusion profile, the target building's usage classification, roof shape information, installable area, historical power consumption patterns, and surrounding obstacle information included in the building type information can be matched with solar radiation, ambient temperature, humidity, cloud cover, wind speed, and wind direction information included in the weather environment data, and actual power generation, panel temperature, and conversion efficiency information included in the real-time power generation status data. For example, the device can correlate weather values ​​and power generation status values ​​for the same time period to extract patterns of actual power generation efficiency degradation, and by referencing historical consumption patterns for each building, determine whether a building has a high propensity for self-consumption or a high proportion of reverse power transmission. Accordingly, the weather-building fusion profile can be generated not merely as a simple power plant operation profile, but as an integrated profile for revenue analysis that comprehensively reflects building structure, load demand, and external weather changes, and can be utilized as a set of input features to predict hourly expected power generation and dynamic power demand curves in subsequent steps.

[0100] The step of calculating real-time expected profit information is a step of deriving real-time profit forecast results including hourly expected power generation, expected power sales unit price, expected self-consumption savings, and expected power sales revenue by combining power market indicators with the above-mentioned weather-building fusion profile.

[0101] Here, real-time expected revenue information may be information that quantifies the economic value expected to be generated by rental solar assets at the current time or in each time interval of a predetermined future period. The above-mentioned hourly expected power generation amount may be a power generation forecast value calculated based on a weather-building fusion profile, and the above-mentioned expected power sales unit price may be a sales unit price determined by combining the real-time system marginal price and the trading unit price of renewable energy certificates. In addition, the above-mentioned expected self-consumption savings may be a value reflecting the retail electricity costs saved and the peak basic charge avoidance effect resulting from replacing the building's internal load with solar power, and the above-mentioned expected power sales revenue may be a value calculated by applying the sales unit price to the amount of surplus electricity transmitted to the external grid without being consumed within the building.

[0102] In the step of calculating real-time expected profit information, the dynamic power demand of the target building is estimated based on load connection characteristics by building type and temperature variability by time period, and the self-consumed power and surplus transmitted power can be distinguished by overlapping the expected power generation amount and the dynamic power demand by payment interval. For example, in the case of a commercial building where the cooling load surges during the day, the self-consumed power amount may be calculated significantly even with the same power generation amount, and in the case of a logistics warehouse-type building with a low load on weekends or at night, the surplus transmitted power amount may be calculated relatively significantly. The device can calculate the hourly usage fee savings, the peak basic fee avoidance amount, and the power sales revenue, respectively, by reflecting these differences, and can determine the actual expected profit level more conservatively by subtracting a weather variability risk margin proportional to the solar irradiance prediction error rate of the composite prediction model as needed. For example, if the estimated power generation at a specific time is calculated to be 50 kWh, of which 20 kWh is consumed inside the building and 30 kWh is transmitted to the external grid, the device can calculate the retail electricity cost savings corresponding to 20 kWh as the estimated self-consumption savings and separately calculate the estimated electricity sales revenue by applying the sales unit price corresponding to 30 kWh. The real-time expected revenue information calculated in this way can be used as basic data to set operation priorities, output control criteria, settlement criteria, and revenue distribution criteria in subsequent stages.

[0103] The step of generating operational execution information is a step of generating execution decision information to be reflected in actual asset operation and contract operation based on the operational analysis information. Here, operational execution information may not be merely the result of a simple profit forecast itself, but may be execution information indicating how to operate solar assets, on what criteria to settle, and in what manner to distribute profits and manage contracts according to the real-time expected profit information and operational standard information derived therefrom. For example, operational execution information may include priority operation rankings for specific building-type assets, criteria for output limiting or output increase, criteria for rental fee settlement, profit distribution coefficients, a list of assets prioritized for maintenance, and information on targets for contract renewal or contract adjustment.

[0104] The step of generating operational standard information is to calculate the operation priority, output control standard, settlement standard, and profit distribution standard for each solar asset based on the real-time expected profit information.

[0105] Here, "operation priority" may refer to criteria indicating which asset to prioritize maintaining at maximum output or which asset to apply a more conservative operation policy to, in cases where multiple solar assets or multiple building groups exist. "Output control criteria" may refer to criteria indicating under what conditions output restriction or output increase will be performed in the event of weather fluctuations, changes in electricity unit prices, surges in internal building load, or changes in grid connection conditions. Additionally, "settlement criteria" may refer to standards regarding the calculation of rental fees, the reflection of savings from self-consumption, the attribution of sales revenue, and the cost-sharing structure, while "profit distribution criteria" may refer to the profit distribution ratio or order among the rental business operator, building operator, and other stakeholders. For example, the device may set output control criteria to prioritize external sales over self-consumption when the current time-based electricity sales unit price is high and the building's own electricity unit price is relatively low, and may assign a higher operation priority to assets in building groups with a higher expected rate of return.

[0106] The step of determining asset operation plans and contract management plans by building type is a step of finalizing specific operation and contract execution plans to maximize expected returns over a predetermined period based on the real-time expected return information and the operational standard information mentioned above.

[0107] Here, the asset operation plan may include operation schedules, output control schedules, priority asset groups for maintenance, maintenance priorities, and profitability-oriented operation strategies for solar assets by building type. Additionally, the contract management plan may include whether to maintain rental contracts, priority for contract renewals, adjustments to terms for buildings with declining profitability, recalculation of contract unit prices, or criteria for proposing new contracts. For example, in the case of industrial buildings where power demand decreases due to factory shutdowns on weekends, excessive surplus power may be generated; therefore, the device may set an operation schedule that adjusts inverter output or considers the connection of storage devices. Furthermore, if the combined total of savings from self-consumption and revenue from power sales is expected to consistently fall below the guaranteed return agreed upon at the time of the contract, procedures to adjust contract terms may be initiated to temporarily reduce rental fees or adjust the settlement cycle. In this way, the device can make final operational execution decisions to maximize the economic efficiency of the entire rental solar power plant portfolio by reflecting both the revenue structure and contract structure by building type, rather than simply maximizing power generation.

[0108] In addition, the step of generating the weather-building fusion profile comprises: a step of extracting the roof shape, roof slope angle, azimuth angle, floor height, and potential installation area of ​​the target building and solar assets from the building type information; a step of generating a Building Geometry Model by placing the envelope shape of the target building and the roof surface where the solar assets are installed on a 3D spatial coordinate system based on surrounding obstacle information included in the building type information; a step of calculating the solar elevation angle and solar azimuth angle for each time period, and tracking the solar projection path for the Building Geometry Model to calculate a Dynamic Shading Fraction matrix corresponding to the roof surface and determining it as a shading influence characteristic; and a step of calculating local microclimate variables around the target building by combining the roof surface reflectance (Albedo), thermal transmittance characteristics, and radiative heat emission coefficient of the target building with the outdoor temperature, relative humidity, cloud cover, solar radiation forecast value, wind speed, and wind direction information extracted from the weather environment data. The method may include: determining solar radiation acceptance characteristics by estimating the effective solar radiation and surface temperature at different times reaching the individual panel surfaces of the solar assets based on the local microclimate variables and dynamic shading fraction matrix; calculating the power generation efficiency characteristics at different times by inputting the effective solar radiation and surface temperature at different times into a power generation efficiency calculation model defined with the temperature coefficient of the solar assets, panel rated efficiency, DC-AC conversion efficiency curve, and partial shading correction coefficient; extracting load linkage characteristics by time-series clustering the past power consumption patterns of the target building derived from the real-time power generation status data, accumulated power generation history data, and the integrated operation data; and deriving the weather-building fusion profile by combining the shading influence characteristics, solar radiation acceptance characteristics, power generation efficiency characteristics at different times, and load linkage characteristics into a multidimensional feature vector.

[0109] The step of generating a weather-building fusion profile may be a step of generating a basic profile to predict the profitability of rental solar assets installed in the building over time by combining the structural characteristics of the target building, external weather conditions, the power generation characteristics of the actual solar assets, and the power consumption tendencies of the building into a single integrated analysis structure.

[0110] Here, the aforementioned weather-building convergence profile refers not to simple weather forecast values ​​or simple power generation history data, but to a multidimensional set of features that comprehensively reflects building shape, shading effects, local microclimate, panel light reception conditions, power generation efficiency, and load-linking characteristics. Therefore, the weather-building convergence profile can be utilized as key input information for calculating hourly projected power generation, projected self-consumption savings, and projected electricity sales revenue in subsequent stages.

[0111] The step of extracting the roof shape, roof slope angle, azimuth angle, floor height, and solar asset installation area of ​​the target building from the above building type information may be a step of quantifying and collecting the physical and structural characteristics of the actual building where the solar panels will be installed.

[0112] Here, the roof shape may refer to the geometric form of a roof, such as a flat roof, gable roof, pitched roof, or stepped roof; the roof slope angle may represent the angle formed by the roof surface with the horizontal plane; and the azimuth angle may indicate, as an angle value, which direction the roof or panel is facing, such as due south, southeast, or southwest. Additionally, the floor height may be structural information indicating how high the target building is located above the ground, and the solar asset installation area may refer to the actual effective installation area excluding ventilation fans, cooling towers, water tanks, maintenance passages, and structural support limits from the total rooftop area. The device may extract each of the above values ​​in numerical form using architectural drawings, BIM data, building management system information, or on-site scan data.

[0113] Based on the surrounding obstacle information included in the above building type information, the step of generating a building shape model in which the outer shell shape of the target building and the roof surface where the solar assets are installed are placed on a three-dimensional spatial coordinate system may be a step of reproducing the target building and the surrounding environment three-dimensionally in a virtual space on a computer.

[0114] Here, the aforementioned surrounding obstacle information may refer to location, height, and shape data of adjacent buildings, steel towers, large trees, advertising structures, antennas, railings, or rooftop facilities that are taller than the target building or have the potential to block sunlight. The aforementioned building shape model may be a digital three-dimensional model in which the exterior walls, roof surfaces, roof surfaces where panels are to be installed, and surrounding obstacles of the target building are arranged in proportion to their actual positions on a three-dimensional coordinate system composed of X, Y, and Z axes. Accordingly, the device can spatially interpret which roof areas receive direct light and which areas are obscured by shadows, depending on the position of the sun at a specific time.

[0115] The step of calculating the solar elevation angle and solar azimuth angle for each time period, and tracking the solar projection path for the building shape model to calculate a dynamic shading fraction matrix corresponding to the roof surface and determining it as a shading effect characteristic, may be a step of quantifying the shadow effect formed on the roof surface of the target building for each time period by reflecting the position of the sun which changes according to the Earth's rotation and revolution.

[0116] Here, the solar elevation angle may be an astronomical angle indicating how high the sun is above the horizon, and the solar azimuth angle may be an angle indicating the direction from which the sun shines. The device can calculate the solar elevation angle and solar azimuth angle based on the date, time, latitude, and longitude, and can track the shadow path formed by surrounding obstacles by projecting the calculated solar direction vector onto the building shape model. The dynamic shading fraction matrix may be in the form of a matrix consisting of a time axis and a roof surface division zone axis, and can store the shading share by time of day, for example, such as shadows forming on 20 percent of the roof area at 9 AM, 0 percent at noon, and 30 percent at 4 PM. Such a dynamic shading fraction matrix can be utilized as a key indicator representing the shading influence characteristics of the building.

[0117] The step of calculating local microclimate variables around a target building by combining outdoor temperature, relative humidity, cloud cover, solar radiation forecast values, wind speed, and wind direction information extracted from the above meteorological environment data with the roof surface reflectance, thermal transmittance characteristics, and radiant heat emission coefficient of the target building may be a step of precisely calculating local environmental conditions around the building roof that are difficult to explain with only wide-area meteorological information.

[0118] Here, the roof surface reflectance can indicate how much solar radiation is reflected depending on the roof material and color, the thermal transmittance characteristic can refer to the degree to which heat passes through the roof structure between the exterior and the interior, and the radiative heat emission coefficient can indicate the degree to which the heated roof surface releases heat to the outside. For example, even under the same ambient temperature conditions, a black metal roof can generate a high surface temperature, while a light-colored concrete roof may exhibit different temperature change patterns due to differences in reflectance and heat capacity. Accordingly, the device can calculate local microclimate variables, such as the effective ambient temperature near the roof surface, wind speed, heat accumulation on the back of the panel, and local humidity correction values, by combining the meteorological environment data and building material property information.

[0119] The step of determining solar radiation acceptance characteristics by estimating the effective solar radiation and surface temperature at different times reaching the individual panel surfaces of the solar asset based on the local microclimate variables and dynamic shading fraction matrix described above may be a step of quantifying the amount of solar energy actually received by the panel and the thermal state formed therefrom. Here, the effective solar radiation may refer not merely to solar radiation from the sky, but to the actual solar radiation reaching the panel surface that reflects shadow effects, roof reflection effects, correction for angle of incidence on the slope, and local atmospheric conditions. Additionally, the surface temperature may be the final panel surface temperature that reflects not only the ambient temperature but also the amount of light received on the panel surface, wind speed, roof surface radiant heat, and back surface heat release conditions. For example, it may be estimated that a specific panel string has low effective solar radiation in the morning due to the shadow effect of an adjacent building, and that in the afternoon, although the solar radiation is sufficient, the surface temperature rises excessively, resulting in a decrease in efficiency. Such results can be defined as solar radiation acceptance characteristics indicating how effectively the target building can receive solar energy at different times of the day.

[0120] The step of calculating the power generation efficiency characteristics by inputting the effective solar irradiance and surface temperature for each time period into a power generation efficiency calculation model defined with the temperature coefficient of the solar asset, panel rated efficiency, DC-AC conversion efficiency curve, and partial shading correction coefficient may be a step of calculating how much electricity can ultimately be efficiently produced by combining the panel's hardware specifications and actual external conditions.

[0121] Here, the temperature coefficient may be a coefficient indicating how much efficiency decreases when the panel temperature rises by 1 degree above the reference temperature, and the panel rated efficiency may be a basic efficiency value under standard test conditions. Additionally, the DC-AC conversion efficiency curve may represent the power loss characteristics that occur according to the load rate during the process in which the inverter converts DC power to AC power, and the partial shading correction coefficient may be a coefficient that corrects the effect of additional reduction in total output when a part of the panel is affected by shading. The device can calculate power generation efficiency characteristics by time period by combining the effective solar irradiance and surface temperature with each model parameter, and accordingly, can generate a precise power generation efficiency profile adapted to the actual building environment.

[0122] The step of extracting load linkage characteristics by time-series clustering the past power consumption patterns of the target building derived from the real-time power generation status data, time-series accumulated power generation history data, and the integrated operation data may be a step of analyzing how well the time patterns of power production and consumption in the building are aligned.

[0123] Here, the aforementioned power generation history data may be data recording how much power the solar asset has generated hourly over the past few months or years, and the aforementioned past power consumption pattern may be data indicating how the internal heating and cooling load, production facility load, lighting load, and other base loads are distributed hourly during the same period. The device may align the power generation history data and the consumption pattern on a time axis and then cluster time series segments that have similar self-consumption structures or similar reverse transmission structures. For example, in some buildings, the timing of power generation and consumption may strongly coincide due to the operation of production facilities during the daytime, while in other buildings, the proportion of afternoon or nighttime consumption may be high, resulting in a low self-consumption rate and a high surplus transmission rate. The load connection characteristics extracted in this way can be used as input values ​​to more precisely estimate the amount of self-consumed power and surplus transmitted power in a subsequent step.

[0124] The step of deriving the weather-building fusion profile by combining the above-mentioned shading influence characteristics, solar radiation acceptance characteristics, time-dependent power generation efficiency characteristics, and load linkage characteristics into a multidimensional feature vector may be a step of structuring each of the previously calculated individual characteristics into a single integrated profile.

[0125] Here, the multidimensional feature vector may be a numerical vector arranged along a time axis, or it may be in the form of a matrix or tensor that includes building type, panel zone, and time-specific features. For example, a feature vector for a specific time period may simultaneously include elements such as shading fraction, effective solar irradiance, surface temperature, expected power generation efficiency, self-consumption connection, and surplus transmission potential. The device can derive a weather-building fusion profile by accumulating such multidimensional feature vectors over time, and this profile can subsequently be used as an input value for a time-series-spatial attention-based composite prediction model to calculate expected power generation, expected self-consumption savings, and expected power sales revenue by time period. Therefore, the weather-building fusion profile can function as key reference information for real-time expected revenue analysis that comprehensively reflects building structure, weather fluctuations, power generation efficiency, and consumption patterns.

[0126] And, the step of calculating the real-time expected revenue information comprises: a step of calculating the expected power generation amount by time of day by inputting the weather-building fusion profile into a pre-trained time-series-spatial attention-based composite prediction model; a step of generating a weather-dependent dynamic power demand curve based on the thermal inertia of the target building's HVAC by combining the load connection characteristics and time-of-day temperature variability extracted from weather environment data; a step of aligning and superimposing the expected power generation amount by time of day and the dynamic power demand curve according to settlement intervals of the power trading market, and calculating the self-consumed power amount replacing the target building's internal power receiving load and the surplus transmitted power amount sent back to the external grid based on the superposition result; a step of extracting the peak-cut contribution power amount among the self-consumed power amount that reduces the grid load by synchronizing with the target building's maximum demand power calculation interval; a step of extracting a time-of-use (ToU) retail power rate table corresponding to the target building's contract type from the power market indicators; and a step of extracting the standard unit price for charging the peak power basic charge from the power market indicators. A step of extracting the real-time System Marginal Price (SMP) from the above-mentioned power market indicators; a step of extracting the trading unit price of the Renewable Energy Certificate (REC) of the target building from the above-mentioned power market indicators; a step of calculating the amount of savings in usage charges by applying the time-of-day retail electricity rate table to the above-mentioned self-consumption amount; a step of calculating the amount of peak basic charge avoided by applying the standard unit price for the peak basic charge to the above-mentioned peak cut contribution amount; a step of calculating the estimated self-consumption savings representing the actual benefit value of the target building by summing the amount of savings in usage charges and the amount of avoidance of the peak basic charge; and a step of determining the estimated electricity sales unit price by combining the above-mentioned real-time System Marginal Price and the Renewable Energy Certificate trading unit price.The method may include: a step of calculating expected power sales revenue by applying an expected power sales unit price to the above-mentioned surplus power amount; a step of calculating a volatility risk margin proportional to the forecast error rate of the composite forecasting model; and a step of finally determining real-time expected revenue information that serves as the basis for the actual rental fee settlement of the rental business operator by subtracting the volatility risk margin from the total sum of the expected self-consumption savings and the expected power sales revenue.

[0127] The step of calculating the above real-time expected revenue information may be a step of quantifying the economic value that rental solar assets installed on the target building can generate on an hourly basis by reflecting power generation forecasts, building load forecasts, and power market price structures together in the previously generated weather-building fusion profile.

[0128] Here, the aforementioned real-time expected profit information does not refer merely to a simple power generation forecast, but may refer to comprehensive profit information that reflects electricity costs saved through self-consumption, the effect of avoiding basic charges due to the reduction of peak demand power, revenue from surplus power sold to the external grid, and risk adjustment factors due to weather forecast errors. Therefore, the aforementioned real-time expected profit information can be utilized as a key basis for judgment to determine the rental business operator's settlement standards, asset operation priorities, and contract management standards in subsequent stages.

[0129] In the step of calculating the expected power generation amount by time period by inputting the above weather-building fusion profile into a pre-trained time-series-spatial attention-based composite prediction model, the previously derived multidimensional feature data of the building is input into an AI-based prediction model to predict future power production by time period.

[0130] Here, the aforementioned time-series-spatial attention-based composite prediction model may be a model that simultaneously analyzes two dimensions of data: weather changes over time and the three-dimensional structure of the building roof and the physical arrangement of the panels. Through the attention mechanism, the relative influence of local shading, roof slope, azimuth angle, panel installation location, and weather variability on the total power generation at specific times can be learned, thereby enabling the calculation of more precise hourly predicted power generation than simple proportional calculations of solar radiation. For example, even if the same solar radiation forecast is input, the actual amount of light received may differ for buildings with different roof slopes and azimuth angles, and even for the same building, power generation may vary depending on the difference in shadow formation patterns between the morning and afternoon. The device can calculate the predicted power generation for each payment interval or each standard time unit using the composite prediction model, and the predicted power generation can be utilized as a basis for calculating subsequent self-consumption and external sales revenue.

[0131] In the step of generating a weather-dependent dynamic power demand curve based on the thermal inertia of the target building's heating, cooling, and air conditioning by combining the above-mentioned load linkage characteristics and time-varying temperature variability extracted from weather environment data, it is possible to simulate how much electricity the building will use during the day in conjunction with the weather changes of the day or future.

[0132] Here, the aforementioned load linkage characteristic may be a load linkage indicator derived through the alignment relationship between past power consumption and generation patterns, and the aforementioned time-dependent temperature variability may be a value indicating how much the outside temperature rises or falls over time. Furthermore, the aforementioned heating, cooling, and air conditioning thermal inertia may refer to the physical characteristic of a building envelope and interior space maintaining their state for a certain period once heated or cooled. Accordingly, even if the outside temperature changes rapidly, the power demand for heating and cooling inside the building may not react to a maximum immediately but may change with a certain delay. By combining this thermal inertia delay effect with weather variability, the device can generate a dynamic power demand curve that fluctuates time-dependently in a weather-dependent manner, rather than a fixed past average load.

[0133] In the step of aligning and superimposing the above-mentioned hourly projected power generation and dynamic power demand curves by settlement intervals of the power trading market, and calculating the amount of self-consumed power replacing the internal power receiving load of the target building and the amount of surplus transmitted power sent back to the external grid based on the superposition result, the amount of electricity produced and consumed by the building can be compared on the same time axis.

[0134] Here, the settlement interval may be a time unit in which trading and settlement are performed in the power trading market, for example, a 15-minute or 1-hour unit. The device compares the estimated generation amount with the dynamic power demand for each settlement interval; if the estimated generation amount is less than or equal to the power demand, the entire or most of the generation amount may be calculated as self-consumed power, and if the estimated generation amount exceeds the power demand, the portion exceeding the demand may be calculated as surplus transmitted power. For example, if the estimated generation amount in a specific settlement interval is 100 kWh and the dynamic power demand is 70 kWh, 70 kWh may be classified as self-consumed power used for internal load replacement, and 30 kWh as surplus transmitted power transmitted to the external grid. Such classification may be used as preliminary data for separately calculating internal savings revenue and external sales revenue.

[0135] The step of extracting the peak cut contribution amount that reduces the system load by synchronizing with the maximum demand power calculation interval of the target building among the above self-consumed power amounts may be a step of selecting the portion that directly contributes to the reduction of the basic charge among the self-consumed power amounts.

[0136] Here, the aforementioned maximum demand power calculation interval may be a standard time period used for calculating contracted power or basic charges, and the aforementioned peak cut contribution power amount may refer to the amount of self-consumed power during that time period that substantially contributes to reducing maximum demand power. If necessary, the aforementioned peak cut contribution power amount may be converted from a standard power value into an average power or converted power form and applied in accordance with the maximum demand power calculation rules. For example, if a building's cooling load surges during the afternoon peak hours, the maximum power received from the grid may decrease if solar power produced during the same period replaces the internal load; therefore, the device may extract only the portion of self-consumed power that is synchronized with the peak hours and use it as an input value for calculating the peak basic charge avoidance amount.

[0137] In the step of extracting a time-based retail electricity rate table corresponding to the contract type of the target building from the above-mentioned power market indicators, a retail electricity rate structure that can be reduced through self-consumption can be determined. Here, the above-mentioned contract type may refer to a type of electricity rate system such as industrial, general, educational, or residential use, and the time-based retail electricity rate table may be a table in which unit prices are distinguished by off-peak, mid-peak, and peak load time periods. After verifying the contracted power information and rate contract type of the target building, the device can extract the corresponding time-based retail electricity rate table from a database or an external settlement server.

[0138] The step of extracting the standard unit price for charging the peak power basic charge from the above-mentioned power market indicators may be a step of securing a reference value for calculating the additional savings value resulting from the reduction of maximum demand power. Here, the standard unit price for charging the peak power basic charge may be the standard unit price of a fixed charge charged in conjunction with the maximum demand power or contracted power recorded during a specific period. The device may extract the unit price according to the contract type of the target building and the basic charge calculation rules, and this can be directly used to calculate the subsequent peak basic charge avoidance amount.

[0139] The step of extracting the real-time system marginal price from the above-mentioned power market indicators may be a step for calculating the wholesale power price component of the market value of surplus power sold to the external grid. Here, the real-time system marginal price may be a reference price traded per 1 kWh of power in the power market at a specific time period, and may fluctuate in real time depending on the system supply and demand situation, power demand by time period, and the composition of power generation sources. The device may extract the real-time system marginal price from power market indicators collected from a power exchange server or a market information provision system.

[0140] The step of extracting the trading unit price of the renewable energy supply certificate for the target building from the above-mentioned power market indicators may be a step for calculating the value of the renewable premium added to the surplus transmitted power. Here, the above-mentioned trading unit price of the renewable energy supply certificate may be the market trading unit price of a certificate granted based on the performance of renewable energy generation. The device may extract the above-mentioned trading unit price by reflecting the facility type of the target building, the eligibility for certificate issuance, and the trading market standards, and, if necessary, may sort the value converted into a unit of electricity and reflect it as an additional component of external sales revenue.

[0141] The step of calculating the savings in usage fees by applying a time-based retail electricity rate table to the above self-consumed electricity amount may be a step of calculating the amount saved as solar power consumed directly inside the building replaces the purchase of external electricity. The device can calculate the savings in usage fees by multiplying the self-consumed electricity amount for each billing interval by the retail electricity unit price corresponding to that time period. For example, if the self-consumed electricity amount during peak load hours is 10 kWh and the retail electricity unit price for that time period is 200 won / kWh, the savings in usage fees may be calculated as 2,000 won.

[0142] The step of calculating the peak basic charge avoidance amount by applying the standard unit price for peak power basic charges to the above-mentioned peak cut contribution amount of power may be a step of calculating the fixed charge effect saved due to the reduction in maximum demand power separately from the simple energy saving effect.

[0143] Here, the peak basic charge avoidance amount may be the amount saved when the base rate for calculating the basic charge is lowered as the solar asset partially replaces the power demand during peak hours. The device may calculate the peak basic charge avoidance amount by converting the above-mentioned peak cut contribution power amount into a standard for the maximum demand power reduction amount, and then applying the above-mentioned standard unit price for the peak power basic charge or substituting it into a formula based on the maximum demand power reduction amount. For example, if the peak reduction contribution amount is 5kW and the standard unit price for the basic charge is 7,000 won / kW, a peak basic charge avoidance amount of 35,000 won may be calculated.

[0144] The step of calculating the estimated self-consumption savings, which represent the actual benefit value of the target building, by summing the above usage fee savings and peak basic fee avoidance amounts, may be a step of organizing the internal consumption-based revenue effect into a single integrated savings indicator.

[0145] Here, the aforementioned estimated self-consumption savings may include not only simple electricity purchase savings but also the reduction in basic charges resulting from the decrease in peak power demand; thus, it can more accurately reflect the internal profitability that the target building actually obtains through solar assets. In other words, the estimated self-consumption savings can function as a direct economic effect indicator showing how much the building owner or consumer has protected from actual expenses.

[0146] The step of determining the expected power sales unit price by combining the real-time system marginal price and the transaction unit price of renewable energy supply certificates may be a step of determining the unit price of power sold to the external system in the form of the sum of the wholesale power price and the certificate value. Here, the expected power sales unit price may be the external sales unit price that a rental business operator or an asset operator can expect per unit of surplus transmitted power. The device may determine the expected power sales unit price by simply adding the certificate transaction unit price to the real-time system marginal price, or by reflecting facility weighting factors, correction coefficients, or subtraction factors according to market systems.

[0147] The step of calculating the expected electricity sales revenue by applying the expected electricity sales unit price to the above-mentioned surplus transmitted electricity amount may be a step of calculating the expected revenue from electricity sold to an external grid that is not used for internal consumption. For example, if the surplus transmitted electricity amount in a specific payment range is 30 kWh and the expected electricity sales unit price is 150 won / kWh, the device may calculate the expected electricity sales revenue for that range as 4,500 won. This expected electricity sales revenue may be maintained separately from the above-mentioned expected self-consumption savings amount, or may be added when calculating the subsequent total expected revenue.

[0148] The step of calculating a weather variability risk margin proportional to the solar radiation prediction error rate of the above-described composite prediction model may be a step intended to correct for the possibility that expected returns may be overestimated due to the uncertainty inherent in the power generation prediction itself. Here, the solar radiation prediction error rate may be a prediction accuracy indicator calculated based on the difference between past predicted values ​​and actual measured values, and the weather variability risk margin may be a risk adjustment amount to be deducted from expected returns in proportion to the error rate. For example, since significant deviations in actual power generation may occur even with the same predicted power generation amount in seasons or regions with severe changes in cloud cover, the device may assign a larger risk margin as the prediction error rate increases. This prevents real-time expected return information from being calculated in an overly optimistic manner.

[0149] The step of finally determining real-time expected profit information, which serves as the basis for the actual rental fee settlement of the rental business operator, by subtracting the weather volatility risk margin from the total sum of the above-mentioned expected self-consumption savings and expected power sales revenue, may be a step of determining the final settlement basis amount by adjusting the predicted risk from the total expected profit that reflects both internal savings value and external sales revenue.

[0150] Here, the real-time expected revenue information may represent the net expected value that a rental business operator can actually expect during a specific time period or specific operating period, and can be directly used in subsequent steps to set operation priorities, establish settlement criteria, adjust contract terms, and establish profit distribution criteria. Therefore, the real-time expected revenue information can function not merely as a projected revenue, but as an indicator of actual operational value that reflects internal building savings, external sales effects, and weather forecast risks.

[0151] According to one embodiment, a distributed subscription resource clustering and virtual power plant (VPP) integrated control method for responding to a renewable energy bidding system, wherein the device includes a processor, memory, a communication module, and a non-transient storage medium, and the method is performed by the processor by executing a program stored in the non-transient storage medium, comprises: a step of collecting integrated operational data including weather environment data, power market indicators, and real-time power generation status data of solar assets; a step of analyzing the integrated operational data to generate operational analysis information; and a step of generating operational execution information including at least one of an asset operation plan, a resource control plan, a settlement plan, a profit distribution plan, a maintenance plan, and a contract management plan based on the operational analysis information; wherein the step of generating operational analysis information comprises: a step of extracting resource characteristics encompassing the installed capacity, solar irradiance sensitivity, and past power generation patterns of individual solar assets based on the integrated operational data, and grouping a plurality of solar assets into a plurality of logical resource clusters according to a preset virtual power plant (VPP) configuration standard. and a step of integrating the above weather environment data and cluster-specific resource characteristics to calculate the expected integrated power generation amount for each cluster in the settlement period according to the above power market indicators, and deriving a cluster-unit bidding proposal model and a power generation prediction error rate that meet the bidding requirements of the power market; and the step of generating the above operation execution information includes: a step of combining the cluster-unit bidding proposal model and the real-time system marginal price of the above power market indicators to calculate the expected profit and power generation error penalty resulting from participation in the virtual power plant bidding, and determining a resource control plan including the output control amount and operation schedule of individual solar assets within the cluster to maximize the expected profit;The present invention provides a distributed subscription resource clustering and virtual power plant (VPP) integrated control method for responding to a renewable energy bidding system, comprising the step of: calculating the integrated revenue of the virtual power plant by evaluating the bid execution results and the integrated power generation achievement rate according to the resource control plan; and determining the settlement plan and revenue distribution plan by proportionally applying the contribution of individual subscribers to the logical resource cluster configuration and the performance record of defending against the power generation prediction error rate.

[0152] The step of collecting integrated operation data is a step of securing external environment information, market information, and facility operation information by aligning them on the same time axis and the same asset identification system in order to perform integrated control of a virtual power plant in response to the renewable energy bidding system.

[0153] Here, integrated operational data refers to a data set structured to enable the interlinking of meteorological environment data, power market indicators, and real-time power generation status data of solar assets. For example, meteorological environment data may include real-time solar irradiance, ambient temperature, relative humidity, cloud cover, wind speed, and wind direction for each region where solar assets are located; power market indicators may include settlement prices by settlement interval, minimum bid units, price quotation units, real-time system marginal prices, trading unit prices for renewable energy certificates, and reference values ​​related to forecasting system imbalances; and real-time power generation status data may include voltage, current, output, cumulative power generation, inverter status, and communication status information for each solar asset. The device can generate an integrated operational dataset capable of subsequent analysis by sorting each of the above data by reference time and performing missing value correction, outlier removal, asset identification number mapping, and resampling based on settlement intervals.

[0154] The step of generating operational analysis information is not merely about storing the aforementioned integrated operational data, but rather about producing analysis results to determine the composition of resource groups eligible for bidding and the potential generation capacity for each payment interval. Here, operational analysis information may refer to intermediate analysis results that include the facility characteristics of individual solar assets, weather response characteristics, past generation patterns, complementarity between assets, estimated integrated generation capacity per cluster, bidding proposal models, and generation prediction error rates. In other words, the aforementioned operational analysis information may serve as a basis for judgment that acts as a standard for subsequent resource control and settlement decisions, and it can be distinguished from the integrated operational data in that it is an analysis result reflecting statistical processing, prediction model computations, and optimization computations, rather than raw data.

[0155] The step of extracting resource characteristics and grouping multiple solar assets into multiple logical resource clusters is a step of reorganizing individual solar assets into aggregated resources suitable for virtual power plant bidding units, rather than treating individual solar assets independently.

[0156] Here, resource characteristics may refer to characteristic values ​​including the installed capacity of individual solar assets, power generation responsiveness to changes in solar irradiance, historical power generation patterns by time of day, power generation forecast error characteristics, and correlations between assets. Solar irradiance sensitivity may be an indicator representing how sensitively the power output of a corresponding solar asset fluctuates in response to changes in the amount of sunlight. Furthermore, a logical resource cluster may refer to a unit logically grouped to operate multiple solar assets installed in physically different locations as a single collective resource for bidding and control purposes. For example, it can be configured to group east-facing resources with high power generation in the morning and west-facing resources with high power generation in the afternoon into a single group to flatten overall power generation variability.

[0157] In the above step, after extracting resource characteristics encompassing the installed capacity, solar irradiance sensitivity, and past power generation patterns of individual solar assets, multiple solar assets can be arranged into multiple logical resource clusters according to pre-set virtual power plant configuration criteria.

[0158] Here, the criteria for configuring a virtual power plant may include minimum participation capacity, grid connection constraints, acceptable forecast error levels, responsiveness within the same payment interval, and the allowable range of characteristic deviation within the asset group. For example, even if a specific solar asset does not meet the minimum bid capacity on its own, it may be included in the same cluster if it satisfies the bid requirements for each payment interval when combined with other assets. Furthermore, since grouping only assets with excessively different irradiance sensitivities can lead to increased variability in power generation under specific weather conditions, the device may adjust the grouping to ensure that the characteristic deviation within the cluster remains within the allowable range. In other words, the aforementioned clustering is not merely a criterion of simple geographical proximity, but rather a process of finding combinations that can mutually compensate for forecast errors and enhance market bid stability.

[0159] The step of integrating meteorological environment data and cluster-specific resource characteristics to calculate the expected integrated power generation amount for each cluster in the payment section, and deriving a cluster-unit bidding proposal model and a power generation prediction error rate that meet the bidding requirements of the power market, is a step of predicting how stably the aforementioned logical resource cluster can supply a certain amount of power generation in the actual bidding market.

[0160] Here, the settlement interval may refer to the time unit during which electricity trading volume is settled in the electricity trading market, and the bid proposal model may refer to a combination of the projected generation volume and bid price that a specific cluster will submit to the electricity market within that settlement interval. Additionally, the generation forecast error rate may be a value that quantifies the range of probabilistic error that may occur between the predicted generation volume and the actual generation volume due to factors such as actual weather fluctuations.

[0161] In the above step, multiple weather forecast scenarios and resource characteristics by cluster can be input into a time-series forecasting model to calculate an expected integrated power generation distribution based on probability density, and the stability of the forecast can be evaluated using the median, lower percentile, and the deviation between the upper percentile of the distribution. Subsequently, the device derives a combination of bid capacity and bid price that maximizes risk-adjusted expected return by reflecting settlement prices by payment interval, imbalance penalty rates, and non-compliance criteria, and can finalize this as a bid proposal model at the cluster level. That is, the above step may not stop at predicting “how much can be generated” but may also be a step that derives “how much to bid at what price to be most advantageous.”

[0162] The step of generating operational execution information is a step of converting the aforementioned operational analysis information into actionable planning information that can be reflected in actual operation and settlement. Here, operational execution information may be execution information that includes at least one of an asset operation plan, a resource control plan, a settlement plan, a revenue distribution plan, a maintenance plan, and a contract management plan. In other words, operational execution information can be distinguished from operational analysis information in that it is not merely a step of analyzing how much power will be generated, but rather a specific action and financial plan regarding how to actually control the power plant and how to distribute revenue to subscribers.

[0163] The step of calculating the expected profit and power generation error penalty resulting from participation in the virtual power plant bidding by combining a cluster-unit bidding proposal model and a real-time system marginal price, and finalizing a resource control plan including the output control amount and operation schedule of individual solar assets within the cluster to maximize the expected profit, is a step of finalizing equipment control commands to preemptively prevent financial losses that may occur due to the difference between the power generation bid in the power market and the actual power generation.

[0164] Here, the expected revenue can be calculated based on the optimal bid capacity, the optimal bid price, the real-time system marginal price, and the settlement price for each settlement interval, and the generation forecast error penalty may refer to the cost imposed by market regulations for generating less or more power than the bid amount. For example, if a specific cluster bids 100 kWh but generation of 120 kWh is expected due to a surge in solar irradiance, and the penalty for excess generation resulting from this is calculated to be greater than the additional sales revenue, the device may issue a reduction command to the inverter output of a specific solar asset within the cluster to determine the output control amount and the operation schedule for that time period to match the actual generation amount to the target level.

[0165] To this end, the device can estimate the expected cumulative imbalance trajectory for the remaining time using a state-space model-based model predictive control algorithm by comparing the cumulative trajectory of the real-time integrated power generation with the cumulative reference trajectory of the target power generation within the payment interval. Then, if an overgeneration state is predicted, the device determines the target reduction amount by comparing the expected penalty for power generation prediction error with the loss of power sales opportunity due to output control. Subsequently, the device can dynamically allocate the output control amount for each individual asset by considering the irradiance sensitivity, real-time output margin, inverter response speed, recent output stability, and communication latency for each solar asset. Afterward, the device finalizes the operation schedule by defining the triggering point, reduction slope, target limit output value, and control duration for issuing an effective power limit command to the smart inverter, and registers this as a resource control plan.

[0166] The step of calculating the integrated revenue of the virtual power plant by evaluating the bid execution results and the integrated power generation achievement rate according to the resource control plan, and determining the settlement plan and revenue distribution plan by proportionally applying the individual subscriber's contribution to the logical resource cluster configuration and the performance record in defending the power generation prediction error rate, is a step of calculating the total revenue and the portions attributable to individual subscribers based on the actual bid execution results.

[0167] Here, the integrated revenue from the virtual power plant may refer to net or actual profit reflecting generation forecast error penalties, control costs, and other settlement adjustment items from realized revenue across the entire settlement period. Additionally, the contribution of an individual subscriber may refer to the extent to which the solar assets provided by that subscriber contributed to the formation of the cluster's bid capacity, the achievement of actual generation, or the mitigation of volatility, and the performance in defending the generation forecast error rate may refer to the extent to which a specific subscriber's assets contributed to suppressing the expansion of errors across the entire cluster or offsetting weather volatility. For example, among two subscribers who provided the same installed capacity, the device may grant a higher settlement ratio to the subscriber who demonstrated higher output stability during the actual settlement period and contributed more significantly to reducing the overall cluster forecast error rate. Accordingly, the settlement plan may be established to include the settlement amount, settlement timing, and settlement threshold value for each individual subscriber, and the revenue distribution plan may be determined as a structure that differentially distributes the integrated revenue from the virtual power plant based on each subscriber's contribution and error defense performance.

[0168] Ultimately, the distributed subscription resource clustering and virtual power plant integrated control method for responding to the aforementioned renewable energy bidding system may be a method that enables a series of operational procedures to be executed within a single integrated framework. These procedures include reorganizing distributed solar assets into logical resource clusters based on integrated operational data, optimizing bidding capacity and prices by payment interval, suppressing imbalances through real-time control after winning the bid, and finally settling and distributing integrated profits. Accordingly, bidding participation requirements that are difficult for individual solar assets to meet on their own can be supplemented at the cluster level, and a reasonable settlement structure proportional to the contribution of each subscriber can be implemented while mitigating risks associated with weather variability and forecasting errors.

[0169] In addition, the step of grouping the solar assets into multiple logical resource clusters comprises: a step of calculating the difference between the hourly predicted power generation calculated by a previously trained power generation prediction model using the past power generation pattern and weather environment data as input values ​​and the actual power generation for the corresponding time period as the hourly power generation prediction error history for each solar asset; a step of calculating the solar irradiance sensitivity for each solar asset, which indicates the power generation response to changes in solar irradiance, based on the correlation between the hourly solar irradiance change amount included in the weather environment data and the hourly actual power generation change amount included in the past power generation pattern; and a step of generating a prediction error covariance matrix for the power generation prediction errors among the multiple solar assets based on the hourly power generation prediction error history for each solar asset, and calculating a Portfolio Variance indicator, which indicates the effect of offsetting weather volatility between assets, as a resource characteristic from the prediction error covariance matrix. A step of obtaining virtual power plant configuration criteria including a minimum participation capacity condition for the virtual power plant, physical grid node constraints based on line connection points, a maximum allowable prediction error limit per cluster, a controllable response time condition within the same payment interval, and a condition allowing deviation in solar irradiance sensitivity for each solar asset within the cluster; and a step of configuring a Mixed-Integer Linear Programming (MILP) optimization problem using the portfolio variance indicator, installed capacity for each solar asset, solar irradiance sensitivity for each solar asset, and virtual power plant configuration criteria as input values, and finally determining a plurality of the solar assets into a plurality of logical resource clusters such that the total sum of portfolio variances within the cluster is minimized and the deviation in solar irradiance sensitivity within the cluster is within a preset range;The step of deriving the cluster-unit bidding proposal model and the power generation prediction error rate, which includes the above, comprises: inputting ensemble weather forecast data composed of multiple weather forecast scenarios and cluster-unit resource characteristics into a time-series forecasting model for each logical resource cluster to compute a cluster-unit probability density-based expected integrated power generation distribution for the settlement interval; calculating the lower percentile, median percentile, and upper percentile for each confidence interval of the expected integrated power generation distribution, and calculating a first deviation rate of the lower percentile relative to the median percentile and a second deviation rate of the upper percentile relative to the median percentile; calculating the cluster-unit power generation prediction error rate based on the first deviation rate and the second deviation rate; and extracting the minimum bidding unit, price quotation unit, settlement price per settlement interval, forecasting system imbalance penalty rate, and non-performance settlement criteria included in the power market indicators. The method may include: a step of constructing a Risk-Adjusted Expected Revenue objective function by combining the expected wholesale electricity revenue when a specific amount of power generation on the above-mentioned expected integrated power generation distribution is assumed as the bid capacity and the bid price corresponding to the specific amount of power generation is set to match the price quotation unit, and a forecasting system imbalance penalty calculated in proportion to the power generation prediction error rate relative to the bid capacity; and a step of deriving an optimal bid capacity and optimal bid price pair (Price-Quantity Pair) that maximizes the output value of the above-mentioned objective function and confirming it as a bid proposal model for the cluster unit.

[0170] The step of grouping the above solar assets into multiple logical resource clusters may be a step of reorganizing individual solar assets into a pool of resources eligible for bidding at the virtual power plant level by comprehensively considering the power generation characteristics, weather responsiveness, forecast error structure, and suitability for market participation of each asset, rather than simply grouping individual solar assets based on their physical installation locations.

[0171] Here, the aforementioned logical resource cluster may refer to a virtual set of resources that are actually installed in different locations but are grouped together to operate as a single bidding unit based on the expected integrated power generation, prediction stability, and control responsiveness according to the payment interval. Therefore, the above step may not be a simple bundle of facilities, but rather a process of designing an asset combination to simultaneously achieve weather variability mitigation and bidding stability.

[0172] In the step of calculating the difference between the hourly predicted power generation amount produced by a previously trained power generation prediction model using the aforementioned past power generation pattern and weather environment data as input values, and the actual power generation amount for that time period, as the hourly power generation prediction error history for each solar asset, it is possible to quantify in a time series the extent of the prediction deviation that individual solar assets have shown in the past.

[0173] Here, the aforementioned power generation prediction model may be a model that calculates the expected power generation for a specific time period by learning past weather and power generation history, and may be implemented as any one of a regression model, a time-series neural network model, or a tree-based prediction model. The aforementioned predicted power generation by time period may refer to the amount of electricity calculated in advance to be produced at a specific time under the same or similar past weather conditions, and the aforementioned actual power generation may refer to the actual amount of electricity measured by a meter or inverter monitoring device at that time. For example, if a specific solar asset was predicted to generate 100 kWh at 10:00 AM but actually generated only 90 kWh due to the influx of clouds, the device may calculate the difference of 10 kWh or a corresponding ratio value as the power generation prediction error for that time period and accumulate and store it in a database. The history of prediction errors calculated in this way can be utilized as basic data to determine how unstably a specific asset responds to weather changes.

[0174] Based on the correlation between the hourly variation in solar radiation included in the above meteorological environment data and the hourly actual variation in power generation included in the above past power generation pattern, in the step of calculating the solar radiation sensitivity of each solar asset, which indicates the power generation responsiveness to changes in solar radiation, it is possible to evaluate how immediately and significantly a solar asset responds to changes in sunlight intensity.

[0175] Here, the aforementioned solar irradiance sensitivity may be a response indicator representing how significantly the power generation of a specific solar asset increases or decreases when the same change in solar irradiance occurs. For example, there may be assets whose power generation decreases by nearly the same proportion of 10 percent when solar irradiance decreases by 10 percent, while there may be assets whose power generation plummets by more than 15 percent depending on the panel type, installation angle, partial shading structure, or inverter efficiency characteristics. Additionally, topographical characteristics such as the presence of nearby mountains, high-rise buildings, or structures, which cause a sharp decrease in power generation only within specific solar elevation angle ranges, can also be reflected in the aforementioned solar irradiance sensitivity. The device can quantify the solar irradiance sensitivity of each solar asset by calculating regression coefficients, correlation coefficients, or sensitivity coefficients between the hourly change in solar irradiance and the actual change in power generation, thereby determining whether a specific asset exhibits high output variability during sudden weather changes.

[0176] In the step of generating a prediction error covariance matrix for the prediction errors of power generation among multiple solar assets based on the history of prediction errors of power generation by time period for each solar asset, and calculating a portfolio variance indicator representing the effect of offsetting weather volatility between assets as a resource characteristic from the prediction error covariance matrix, it is possible to analyze not only the magnitude of the error of individual assets but also the error linkage relationship between assets.

[0177] Here, the aforementioned forecast error covariance matrix may be a statistical matrix in tabular form representing the extent to which the forecast error of each asset increases or decreases in conjunction with the forecast error of other assets for multiple solar assets. For example, two assets installed in the same region may exhibit a large positive covariance because their power generation decreases simultaneously upon a sudden influx of clouds, whereas assets installed in different weather zones may show opposing error patterns during certain time periods, resulting in low or negative covariance values. Based on this covariance structure, the aforementioned portfolio variance indicator may be a value representing how much the overall forecast uncertainty of a group is reduced when multiple solar assets are grouped into a single group; a smaller value indicates a superior offsetting effect against weather volatility between assets. Therefore, the aforementioned portfolio variance indicator can be utilized as a key resource characteristic to evaluate which combination provides more stable bidding performance, even among assets of the same capacity, when forming a cluster.

[0178] In the step of obtaining virtual power plant configuration criteria, which include minimum participation capacity conditions for virtual power plants, physical grid node constraints based on line connection points, maximum allowable prediction error limits per cluster, controllable response time conditions within the same payment interval, and allowable deviation conditions for solar irradiance sensitivity per solar asset within a cluster, constraints can be secured to satisfy the legal rules of the power market and the limitations of the physical power grid, rather than arbitrarily grouping power plants.

[0179] Here, the aforementioned minimum participation capacity condition may be the minimum bidding capacity that a virtual power plant must secure to participate in power market bidding, and the aforementioned physical grid node constraint may be a restriction that allows only assets connected to a specific line connection point or grid node to be grouped into the same control unit. Additionally, the aforementioned maximum allowable prediction error limit may be a stability criterion requiring that the generation prediction error rate at the cluster unit level not exceed a certain level, and the aforementioned controllable response time condition within the same settlement interval may refer to the temporal response limit within the settlement interval where output can be adjusted according to commands. The aforementioned condition allowing deviation in solar irradiance sensitivity may be a condition intended to prevent assets with excessively different response characteristics from being included together within a specific cluster. For example, since real-time control may become impossible if the outputs of power plants within a group fluctuate excessively differently when clouds pass, the device may apply the aforementioned configuration criteria to ensure that assets with similar solar irradiance sensitivities are combined within a certain range.

[0180] In the step of constructing a mixed-integer linear programming optimization problem using the above-mentioned portfolio variance indicator, installed capacity per solar asset, solar irradiance sensitivity per solar asset, and virtual power plant configuration criteria as input values, and finally determining multiple solar assets into multiple logical resource clusters such that the total sum of portfolio variances within the cluster is minimized and the deviation of solar irradiance sensitivity within the cluster is within a preset range, the asset combination problem can be formulated as a mathematical optimization problem to search for an optimal solution.

[0181] Here, mixed-integer linear programming may be an optimization model that simultaneously uses a binary decision variable indicating which asset is included in which cluster, and a continuous variable that considers cluster-specific capacity, error, and sensitivity constraints. The device solves the optimization problem to determine asset placement such that the total sum of portfolio variances is minimized for the entire set of clusters, while simultaneously satisfying minimum participation capacity, system node consistency, tolerance limits for prediction error, and sensitivity deviation conditions. Multiple logical resource clusters thus finally determined can be utilized as independent market participation units in the subsequent bidding proposal stage.

[0182] The step of deriving the cluster-unit bidding proposal model and the power generation prediction error rate described above may be a step of calculating the expected integrated power generation and its uncertainty for each payment interval for an already configured logical resource cluster, and calculating the optimal capacity-price combination to be used in the actual bidding by reflecting both the market price structure and the penalty structure. In other words, the above step may be a step of predicting how much a cluster can generate while simultaneously determining how much to bid at what price to be most advantageous.

[0183] In the step of inputting ensemble weather forecast data consisting of multiple weather forecast scenarios and resource characteristics for each cluster into a time series prediction model for each logical resource cluster, and calculating the expected integrated power generation distribution based on probability density at the cluster level for the payment interval, the power generation distribution of the cluster can be predicted in the form of a probability by reflecting multiple forecast scenarios simultaneously rather than a single weather forecast value.

[0184] Here, the ensemble weather forecast data may be a set of multiple weather forecast scenarios generated by different numerical forecasting models, initial conditions, or correction methods. The time series forecasting model may calculate the expected integrated power generation distribution by payment interval in the form of a probability density function by reflecting the installed capacity, solar irradiance sensitivity, portfolio dispersion characteristics, and weather forecast scenarios of each asset constituting the cluster. This may not simply predict that "100 kWh will be generated tomorrow," but rather calculate the power generation in the form of a probability distribution, such as "there is a 70 percent probability of being in the 95–105 kWh range and a 20 percent probability of being in the 90–95 kWh range." In this case, the device can simultaneously verify the central tendency and the range of uncertainty of the power generation forecast by obtaining not only a single average value but also the entire distribution for a specific payment interval.

[0185] In the step of calculating the lower percentile, median percentile, and upper percentile for each confidence interval of the above-mentioned integrated power generation distribution, and calculating the first deviation rate of the lower percentile relative to the median percentile and the second deviation rate of the upper percentile relative to the median percentile, the degree of spread of the above-mentioned probability distribution curve can be quantified to distinguish the downward and upward ranges of the prediction uncertainty.

[0186] Here, the median percentile may be the most probabilistically representative level of power generation, e.g., the 50th percentile; the lower percentile may be the low level of power generation expected in a pessimistic scenario, e.g., the 10th percentile; and the upper percentile may be the high level of power generation in an optimistic scenario, e.g., the 90th percentile. The first deviation rate may be a ratio indicating how far the lower value has fallen relative to the median, and the second deviation rate may be a ratio indicating how far the upper value has risen relative to the median. The larger these deviation rates are, the greater the uncertainty regarding tomorrow's power generation of the cluster.

[0187] In the step of calculating the power generation prediction error rate for each cluster based on the above first deviation rate and second deviation rate, the above bidirectional deviation rates can be combined to calculate a representative error rate value indicating the extent of error the cluster contains during actual bid execution.

[0188] Here, the above-mentioned power generation forecast error rate may be an error indicator that reflects both risks of under-generation and over-generation, and may be calculated by assigning different weights to conservative deviation rates and optimistic deviation rates. For example, if the first deviation rate is -5 percent and the second deviation rate is +7 percent, the device may calculate a representative error rate of approximately 6 percent on average, depending on the market structure and penalty system. In particular, if the structure has a larger market penalty for under-generation, a higher weight may be assigned to the first deviation rate, which represents a downward deviation, and accordingly, the power generation forecast error rate per cluster may be directly utilized in the subsequent risk-adjusted expected return calculation.

[0189] In the step of extracting the minimum bid unit, price quotation unit, settlement price by settlement interval, forecasting system imbalance penalty rate, and non-compliance settlement criteria included in the above-mentioned power market indicators, external constraints and economic conditions can be secured to configure the bid proposal model in accordance with actual market rules.

[0190] Here, the aforementioned minimum bid unit may be the minimum bid capacity allowed in the market, and the aforementioned price quotation unit may be a rule indicating the unit in which bid prices can be submitted. The aforementioned settlement price by settlement interval may be a standard unit price used for calculating profit after the actual winning bid, the aforementioned forecasting system imbalance penalty rate may be a penalty rate imposed based on the error in predicting power generation, and the aforementioned non-performance settlement criteria may be a reduction or correction rule applied when the target power generation amount is not met. The device may retrieve the aforementioned power market rules from a database or a power exchange linkage system and utilize them as constraints for bidding and profit calculation.

[0191] In the step of constructing a risk-adjusted expected return objective function by combining the expected wholesale electricity revenue when a specific amount of power generation on the above-mentioned integrated power generation distribution is assumed as the bid capacity and the bid price corresponding to that specific amount is set to match the price quotation unit, with the forecasting system imbalance penalty calculated in proportion to the power generation prediction error rate relative to the bid capacity, a mathematical objective function can be constructed to simulate how much net profit a specific bidding strategy ultimately brings.

[0192] Here, the above-mentioned expected electricity wholesale revenue may be a positive factor that can be obtained when electricity is sold normally in the market, and the above-mentioned forecasting imbalance penalty may be a negative factor that may be returned to or deducted from the market depending on the forecasting error. The above-mentioned risk-adjusted expected return objective function may be a function representing net profit obtained by subtracting the negative factor from this positive factor, and may additionally reflect items such as default settlement risk or operating costs as needed. For example, if 100 kWh, which is the median of the above-mentioned integrated generation distribution, is bid, the expected electricity wholesale revenue increases, but the risk of the imbalance penalty due to the forecasting error rate may also increase; conversely, if 90 kWh is bid conservatively, the revenue may decrease somewhat, but the penalty risk may decrease.

[0193] In the step of deriving the optimal bid capacity and optimal bid price pairs that maximize the output value of the above objective function and finalizing them as the cluster-unit bid proposal model, the most advantageous market participation method among various combinations of bid candidates can be finally determined. The device can generate possible bid capacity and price candidates for each payment interval, calculate the risk-adjusted expected return for each combination, and then select the bid capacity and bid price pair that maximizes that value. For example, if bidding the median value of 100 kWh results in a net profit of 80,000 won, but bidding the more conservative value of 90 kWh reduces the imbalance risk and increases the net profit to 85,000 won, the device can select 90 kWh and the corresponding price as the optimal bid capacity and optimal bid price pair. The cluster-unit bid proposal model finalized in this way can be combined with the real-time system marginal price in a subsequent step to calculate the expected profit and generation forecast error penalty, and can be utilized as reference information to finalize the resource control plan.

[0194] And, the step of finalizing the resource control plan comprises: a step of calculating the target generation amount for each settlement interval corresponding to the optimal bid capacity, assuming that the optimal bid capacity and optimal bid price included in the cluster-unit bid proposal model have been awarded in the power trading market, and calculating the expected profit for each settlement interval by applying the real-time system marginal price, the optimal bid price, and the settlement price for each settlement interval; a step of calculating the real-time integrated generation amount for each settlement interval by aggregating the real-time generation amount of individual solar assets within the cluster based on real-time generation status data collected during the settlement interval, and calculating a generation prediction error penalty by applying the prediction system imbalance penalty rate if the generation amount deviation between the target generation amount for each settlement interval and the real-time integrated generation amount for each settlement interval exceeds a preset allowable error range; and a step of monitoring the cumulative trajectory of the real-time integrated generation amount and the cumulative reference trajectory of the target generation amount for each settlement interval at preset control cycles within the settlement interval, and estimating the expected cumulative imbalance trajectory during the remaining time of the settlement interval by running a state-space model-based Model Predictive Control (MPC) algorithm. If the above-mentioned predicted cumulative imbalance trajectory is predicted to exceed the allowable error range and enter an over-generation state, a step of determining a cluster-unit target curtailment amount required to resolve the over-generation state by comparing the estimated amount of the generation prediction error penalty with the amount of lost power sales opportunities resulting from the execution of output control; and a step of calculating a distributed control optimization function with input variables such as solar irradiance sensitivity, real-time output margin, inverter active power response speed, recent output stability, and communication delay time for each solar asset, and dynamically allocating output control amounts for each solar asset based on the calculated value of the distributed control optimization function to distribute the target curtailment amount to individual solar assets within the cluster.and, based on the output control amount, a step of determining an operation schedule by defining a time-series triggering point, ramp rate, target limit output value, and control duration for transmitting an effective power limit command to the smart inverter of the solar asset via a network, and registering the determined operation schedule and output control amount as a resource control plan; may be included.

[0195] The step of finalizing the above resource control plan may be a step of minimizing economic losses due to power generation prediction errors by continuously tracking the difference between the target power generation amount for each settlement interval and the real-time integrated power generation amount, and controlling the output of individual solar assets when necessary, in order to implement the cluster-unit bidding proposal model already awarded in the power trading market to the actual power generation site.

[0196] Here, the aforementioned resource control plan does not refer to a simple reduction command itself, but may be a comprehensive control plan that includes target generation volume per payment interval, expected revenue, penalty for generation forecast error, target reduction amount, output control amount per individual asset, smart inverter control timing, reduction slope, target limit output value, and control duration. Therefore, the above step can be understood not as a simple prediction step, but as an execution step that dynamically aligns the actual power plant operating status after bidding with market target values.

[0197] Assuming that the optimal bid capacity and optimal bid price included in the above cluster-unit bid proposal model have been awarded in the power trading market, in the step of calculating the target generation amount for each settlement interval corresponding to the above optimal bid capacity and calculating the expected profit for each settlement interval by applying the above real-time system marginal price, optimal bid price, and settlement price for each settlement interval, a target generation standard value that the cluster must achieve in each settlement interval can be set based on the bidding conditions accepted in the market.

[0198] Here, the optimal bid capacity may be the power supply amount submitted to and adopted by the market for the entire settlement period, and the target generation amount for each settlement period may be a target value obtained by distributing the optimal bid capacity by settlement period or subdividing it by time period according to the winning bid conditions. For example, if a specific cluster is awarded a bid on the condition of supplying 100 MWh from 1:00 PM to 2:00 PM, the 100 MWh may be the target generation amount to be achieved in that settlement period. Additionally, the real-time system marginal price may be the actual power market price at that time, the optimal bid price may be the price submitted at the time of bidding, and the settlement price for each settlement period may be the price applied to the calculation of actual profit after winning the bid according to market rules. The device can calculate the expected profit for each settlement period by reflecting each of the above price elements, and through this, can determine whether it is economically advantageous to maintain the target generation amount in the current period or to reduce the penalty risk through reduction.

[0199] In the step of calculating the real-time integrated power generation amount for each payment period by aggregating the real-time power generation amounts of individual solar assets within the cluster based on real-time power generation status data collected during the aforementioned payment period, and calculating the power generation prediction error penalty by applying the aforementioned prediction system imbalance penalty rate when the power generation deviation between the target power generation amount for each payment period and the real-time integrated power generation amount for each payment period exceeds a preset allowable error range, it is possible to evaluate in real-time how well the actual power generation performance satisfies the target fulfillment conditions.

[0200] Here, the real-time power generation status data may include the instantaneous output, cumulative power generation, inverter operation status, string unit output, and communication delay information of each solar asset. The device can calculate the real-time integrated power generation of the entire cluster by summing the real-time power generation of each individual asset within the payment interval, and can calculate the power generation deviation by comparing this with the target power generation for each payment interval. For example, if the target power generation for a specific payment interval is 100 MWh and the real-time integrated power generation is calculated to be 115 MWh, the device can calculate a ratio deviation of 15 MWh or an equivalent deviation, and if the deviation exceeds the allowable error range, it can calculate the power generation prediction error penalty to be borne by applying the prediction system imbalance penalty rate. In this way, the device can evaluate in real-time the extent of actual economic disadvantage beyond a simple difference in power generation.

[0201] In the step of monitoring the cumulative trajectory of real-time integrated power generation and the cumulative reference trajectory of target power generation per payment period at each preset control cycle within the above payment period, and running a state-space model-based model prediction control algorithm to estimate the expected cumulative imbalance trajectory during the remaining time of the payment period, it is possible to preemptively predict not only the instantaneous error at the current time but also the performance results that will be accumulated until the end of the payment period.

[0202] Here, the cumulative trajectory of the real-time integrated power generation may refer to a curve formed by the accumulation of actual power generation on the time axis as the settlement period progresses, and the cumulative reference trajectory of the target power generation may refer to a reference curve formed by accumulating the target power generation over time. For example, if the weather perfectly matches the forecast, the actual cumulative curve may proceed in a shape almost identical to the reference curve; however, if sudden cloud dissipation or a surge in solar radiation occurs, the actual cumulative curve may rapidly exceed the reference curve. The state-space model-based model predictive control algorithm may be a control technique that predicts the output for a certain future period and calculates the optimal control amount by considering the current state, control input, disturbances, and constraints together, and the device can use this to estimate the expected cumulative imbalance trajectory during the remaining time of the settlement period. In other words, even at a point when the settlement period has not yet ended, it is possible to determine in advance whether the final allowable error will be exceeded if the current trend continues.

[0203] In the step of determining the target reduction amount per cluster unit required to resolve the overgeneration state by comparing the estimated amount of the generation prediction error penalty with the amount of lost power sales opportunity resulting from the execution of output control when the above-mentioned estimated cumulative imbalance trajectory is predicted to enter an overgeneration state by exceeding the allowable error range, the level of economic response to the overgeneration state can be quantitatively determined.

[0204] Here, the aforementioned target reduction amount may refer to the total sum of power generation that must be reduced for the entire cluster to return to within the allowable error range. An overgeneration state may refer to a condition where electricity is produced excessively more than the amount promised to the market, potentially causing settlement errors and grid burdens. In this case, failure to reduce power generation may result in a penalty for power generation forecasting errors; conversely, reducing power generation may result in the forfeiture of power that could have originally been sold, leading to a loss of opportunity for power sales. For example, if the estimated penalty for not reducing power is 500,000 won and the forfeited revenue from power sales is 200,000 won, the device may determine the target reduction amount by judging that reducing power generation is economically advantageous. Conversely, if the opportunity loss resulting from reduction is greater than the estimated penalty, the device may respond by mitigating the intensity of the reduction or delaying the timing of the reduction.

[0205] In order to distribute the above target reduction amount to individual solar assets within the cluster, a distributed control optimization function is computed using the solar irradiance sensitivity, real-time output margin, inverter active power response speed, recent output stability, and communication delay time for each solar asset as input variables, and in the step of dynamically allocating the output control amount for each solar asset based on the calculated value of the distributed control optimization function, the target reduction amount determined at the cluster level can be decomposed into specific control commands at the individual asset level.

[0206] Here, the aforementioned solar irradiance sensitivity may refer to the power generation responsiveness of each asset to weather changes, and the aforementioned real-time output margin may refer to the difference between the current output and the inverter rated output or the controllable lower limit output. Additionally, the aforementioned inverter active power response speed may be a value indicating how quickly the actual output decreases or increases after the input of a control command, the aforementioned recent output stability may be a value indicating how small the output variability is over a short period, and the aforementioned communication delay time may be the time required for a control command to be transmitted and actually reflected. The device configures a distributed control optimization function using the above variables as input values ​​so that, rather than simply distributing the target reduction amount evenly, it preferentially allocates a larger reduction amount to assets with fast and stable responses and large current reserve output, and conversely allocates a relatively smaller reduction amount to assets with slow responses or large communication delays. Accordingly, it is possible to stably achieve the control goal of the entire cluster while preventing the concentration of an excessive control burden on specific assets.

[0207] Based on the above output control amount, the operation schedule is determined by defining a time-series triggering point, reduction slope, target limit output value, and control duration for transmitting an effective power limit command to the smart inverter of the solar asset via the network, and in the step of registering the determined operation schedule and output control amount as a resource control plan, the previously calculated control amount can be concretized into an operation command that is executable in the actual field.

[0208] Here, the time-series triggering point may be a point in time indicating when output limiting begins for each solar asset, the reduction slope may be a control slope that causes output to decrease at a constant rate rather than dropping all at once, and the target limit output value may be a target output level to be maintained after reduction. Additionally, the control duration may indicate how long the reduction state is maintained. For example, for an inverter currently operating at 100 percent output, a schedule can be defined to start output limiting after 5 minutes, gradually reduce it to 70 percent over 2 minutes, and then maintain the output limit state for 15 minutes. The device can finalize an operation schedule including these elements for each smart inverter and issue commands via a network in a sequential or parallel manner. Subsequently, by registering the finalized operation schedule and output control amount as a resource control plan, bid execution during the settlement period can be dynamically maintained in accordance with market conditions and real-time power generation status. Ultimately, the step of finalizing the aforementioned resource control plan can function as a real-time control execution step to align the winning bid conditions with the actual operating status of the power plant, minimize economic disadvantages caused by power generation prediction errors, and enhance the market implementation stability of the entire virtual power plant.

[0209] According to one embodiment, a device includes a processor, memory, a communication module, and a non-transient storage medium, and a method for managing and settling power peak loads for a V2G (Vehicle to Grid) and ESS-linked solar subscription service, which is performed by the processor by executing a program stored in the non-transient storage medium, comprises: a step of collecting integrated operational data including weather environment data, power market indicators, and real-time power generation status data of solar assets; and a step of analyzing the integrated operational data to generate operational analysis information. and, based on the above-mentioned operational analysis information, a step of generating operational execution information including at least one of an asset operation plan, a resource control plan, a settlement plan, a revenue distribution plan, a maintenance plan, and a contract management plan; wherein the step of generating the operational analysis information comprises: a step of aligning real-time power generation status data of solar assets included in the integrated operational data, cumulative charging history data by vehicle, discharge participation history data by vehicle, charging power amount data by time of day by vehicle, charging status information of the ESS, and weather environment data to generate an integrated resource operation profile including surplus solar power amount by time of day, expected connection possible time of day by vehicle group, expected discharge possible capacity by vehicle group, charge / discharge possible capacity by ESS, and peak time response possible capacity; and a step of calculating peak response analysis information including expected load reduction amount by peak time, expected settlement unit price by time, expected discharge revenue by time, and expected charge / discharge cost by time by combining power market indicators with the above-mentioned linked resource operation profile; wherein the step of generating the operation execution information comprises: a step of generating linked resource operation standard information including discharge participation priority by vehicle group, ESS charge / discharge control standard, peak time response standard, and settlement standard based on the above-mentioned peak response analysis information;The present invention provides a method for managing and settling power peak load for a V2G (Vehicle to Grid) and ESS-linked solar subscription service, comprising the step of: determining a resource control plan including an expected discharge participation schedule by vehicle group and a charge / discharge schedule by ESS to reduce grid load during peak hours based on the peak response analysis information and linked resource operation standard information; and determining the settlement plan and revenue distribution plan by reflecting the actual power provision amount and peak reduction contribution amount calculated according to the implementation results of the resource control plan.

[0210] The step of collecting integrated operational data is a step in which the power generation status of solar assets, external environmental information, and market information are collected in a form that can be interconnected in order to interpret the operational status of a solar subscription service linked with V2G and ESS in real time and reflect it in subsequent control and settlement.

[0211] Here, integrated operational data may refer to a data set in which weather environment data, power market indicators, and real-time power generation status data of solar assets are sorted according to the same time standard, the same asset identification system, and the same settlement standard system. For example, weather environment data may include solar irradiance, ambient temperature, humidity, wind speed, and wind direction by time of day; power market indicators may include real-time system marginal price, demand response settlement standards, settlement unit price, power trading time intervals, commission standards, and market rule information; and real-time power generation status data of solar assets may include inverter output, DC side voltage and current, AC output, cumulative power generation, string unit output, and communication status. The device can construct an integrated operational dataset capable of subsequent calculations by sorting each of the above data by reference time point and performing missing value correction, outlier removal, asset identification number mapping, and resampling based on the analysis cycle standard.

[0212] The step of generating operational analysis information is not simply to store the integrated operational data, but to produce analysis results that can determine the responsiveness to peak hours and economic benefits in a combined state of solar assets, vehicle groups, and ESS.

[0213] Here, operational analysis information may refer to intermediate analysis results that include a connected resource operation profile generated by combining solar power generation, vehicle charging and discharging history, and ESS status, and peak response analysis information calculated by reflecting power market indicators thereon. In other words, the operational analysis information can be distinguished from integrated operational data in that it is not merely measurement data, but rather information serving as a basis for judgment that reflects the combined capacity of multiple resources, available connection times, potential load reduction, and expected revenue and cost structure.

[0214] The step of generating a linked resource operation profile is to structure the scale and time of resources that can actually be mobilized during peak hours when solar assets, electric vehicle assemblies, and ESS are combined.

[0215] Here, the linked resource operation profile may be multidimensional resource status information including the amount of solar power surplus by time of day, the expected time of day for connection by vehicle group, the expected discharge capacity by vehicle group, the charge / discharge capacity by ESS, and the capacity capable of responding to peak hours. For example, the device can calculate the expected solar power generation amount by time of day based on real-time power generation status data of solar assets and weather environment data, calculate the expected charging demand amount by time of day using charging power amount data by vehicle by time of day, and then derive the amount of solar power surplus by time of day based on the difference between the two. In addition, by analyzing the cumulative charging history, discharge participation history, and charging power amount data by time of day for each vehicle, the device can calculate the average connection start time, average connection end time, connection frequency by time of day, average connection duration, discharge participation frequency by time of day, average discharge duration, and average discharge contribution amount by vehicle. Subsequently, the device can cluster vehicles with similar usage patterns into multiple vehicle groups and calculate the expected time of day for connection and the expected discharge capacity for each vehicle group. For ESS, the charge and discharge capacity for each ESS can be derived by calculating the current charge amount, minimum maintenance charge amount, maximum allowable charge amount, chargeable capacity, and dischargeable capacity, and the capacity capable of responding to peak times can be calculated by aligning these resources based on peak times. Each element derived in this way can be combined into a multidimensional state vector based on the time axis to form a linked resource operation profile.

[0216] The step of calculating peak response analysis information is a step of quantifying the actual peak load reduction capability and the resulting economic value by combining power market indicators with the aforementioned linked resource operation profile. Here, peak response analysis information may include analysis information such as the expected load reduction amount by peak time period, the expected settlement unit price by time period, the expected discharge revenue by time period, and the expected charge / discharge costs by time period. For example, the device can calculate the amount of power that can actually be input for grid load reduction by summing the solar surplus power amount, the expected dischargeable capacity by vehicle group, and the dischargeable capacity by ESS during peak hours, and this can be defined as the expected load reduction amount by peak time period. In addition, the device can calculate the expected settlement unit price by time period by reflecting the real-time grid marginal price, demand response compensation unit price, settlement criteria, and service fee criteria, and can calculate the expected discharge revenue by applying the discharge participation amount of the vehicle group and ESS to this. Meanwhile, the device can calculate the expected charge / discharge costs by time period by reflecting charging costs, charge / discharge loss costs, degradation costs, or operating fees incurred during the charge / discharge process of the vehicle group and ESS. The peak response analysis information derived in this way can be used as a key input value to determine resource priority and charge / discharge schedules in subsequent steps.

[0217] The step of generating operational execution information is a step of generating actionable criteria and plans to be reflected in actual asset operation and settlement based on the operational analysis information. Here, operational execution information may be execution information that includes at least one of an asset operation plan, a resource control plan, a settlement plan, a profit distribution plan, a maintenance plan, and a contract management plan. That is, the operational execution information may not merely determine how much power can be supplied, but may include execution information that includes which vehicle group to prioritize for discharge, how to charge and discharge the ESS, and to whom and on what criteria the results of peak reduction will be distributed.

[0218] The step of generating information on operational standards for linked resources involves setting priority for discharge participation by vehicle group, ESS charge / discharge control standards, peak time zone response standards, and settlement standards based on the peak response analysis information. Here, priority for discharge participation by vehicle group may be a standard for determining which vehicle group among multiple vehicle groups should be prioritized for participation in peak response. For example, a vehicle group with a high recent participation compliance rate, a long average discharge duration, and a large amount of power contributed to discharge may be assigned a high priority. ESS charge / discharge control standards may be criteria regarding whether to maintain the ESS in a charged state before the peak time zone, operate it with priority for discharge during the peak time zone, or recharge it using surplus solar power after the peak time zone. Peak time zone response standards may include determining which time intervals are considered as peak response target intervals, what level of load reduction must be achieved, and what the allowable range of fluctuation is. Additionally, settlement standards may include rules for calculating the amount to be paid or the cost to be deducted from each participating entity based on peak reduction performance, actual power provision amount, participation compliance rate, and response speed.

[0219] The step of finalizing the resource control plan is to determine a resource control plan including an expected discharge participation schedule by vehicle group and a charge / discharge schedule by ESS so as to reduce the grid load during peak hours based on the peak response analysis information and linked resource operation standard information. Here, the expected discharge participation schedule by vehicle group may be a plan indicating how much power a vehicle group will discharge during a specific time period and until when it will participate, and the charge / discharge schedule by ESS may be a plan indicating at what point an ESS will perform charging or discharging. The device can extract the expected load reduction amount by peak time period, the expected settlement unit price by time period, the expected discharge revenue by time period, and the expected charge / discharge cost by time period included in the peak response analysis information, and construct a net profit objective function by setting the discharge participation priority by vehicle group and the ESS charge / discharge control standard as constraints. For example, by applying mixed-integer nonlinear programming to maximize a net profit objective function that sums the expected demand response settlement amount and expected discharge revenue and subtracts the charge / discharge costs and battery degradation loss costs, a resource control plan can be derived in which the ESS is allocated as a base peak reduction resource and the vehicle group is allocated as a grid variability response resource. In this case, the resource control plan may be a pre-schedule prior to the actual initiation of the peak control event and may be readjusted according to real-time state changes as needed.

[0220] The step of determining the settlement plan and the profit distribution plan involves settling the integrated revenue by reflecting the actual power supply amount and peak reduction contribution amount calculated based on the results of implementing the resource control plan, and distributing it to each participating entity. Here, the actual power supply amount may refer to the amount of power actually provided to the power grid by individual vehicles or ESS during the peak control event period, and the peak reduction contribution amount may refer to the reduction amount calculated at the system level based on the difference between the customer reference load of the target receiving facility and the actual grid power supply amount. For example, the device can determine the actual power supply amount of individual resources based on power measurement data collected from bidirectional power meters after the peak control event ends, and can calculate the peak reduction contribution amount through the difference between the customer reference load and the actual grid power supply amount. Subsequently, the device can calculate the integrated revenue by deducting the platform operation fee from the total settlement amount received from the power market, and can determine the V2G compensation for each individual subscription account by calculating performance indicators based on the discharge participation capacity, power response speed, and schedule fulfillment rate for each individual vehicle. In addition, a settlement plan can be established so that the V2G compensation for each individual subscription account is deducted from the next month's rental fee of the solar subscription service, and a profit distribution plan can be determined so that the surplus profit calculated by deducting the total amount of V2G compensation for each individual subscription account from the integrated profit is divided and allocated according to the ESS owner's shareholding ratio.

[0221] Ultimately, the power peak load management and settlement method for the aforementioned V2G and ESS-linked solar subscription service may be a method that analyzes the power generation status of solar assets and the availability of vehicle groups and ESS in real time to calculate the capacity capable of responding to peak hours, generates peak response analysis information reflecting the market price structure, sets discharge participation priorities and ESS charging / discharging standards for each vehicle group based on this, finalizes resource control plans, and enables consistent execution of settlement plans and revenue distribution plans based on actual implementation results. Accordingly, a resource set integrating solar assets, electric vehicle batteries, and ESS can be utilized not merely as a power storage resource, but as a subscription service asset that simultaneously achieves peak load reduction and settlement revenue generation.

[0222] In addition, the step of generating the linked resource operation profile comprises: a step of calculating the estimated solar power generation amount by time period based on the real-time power generation status data and weather environment data; a step of calculating the estimated charging demand power amount by time period based on the charging power amount data by vehicle by time period, and calculating the solar surplus power amount by time period by subtracting the estimated charging demand power amount by time period from the estimated solar power generation amount by time period; a step of calculating the average connection start time, average connection end time, connection frequency by time period, and average connection duration by vehicle based on the cumulative charging history data by vehicle and the charging power amount data by vehicle by time period; a step of calculating the discharge participation frequency by time period, average discharge duration, and average discharge contribution power amount by vehicle based on the discharge participation history data by vehicle; a step of clustering vehicles having similar usage patterns into a plurality of vehicle groups based on the average connection start time by vehicle, the average connection end time by vehicle, the connection frequency by time period, and the average connection duration; and a step of deriving the estimated connection possible time period for each vehicle group based on the clustering results. A step of calculating the expected dischargeable capacity for each vehicle group based on the discharge participation frequency, average discharge duration, and average discharge contribution power of the vehicles included in the vehicle group by time period; a step of calculating the current charge amount, minimum maintenance charge amount, maximum allowable charge amount, chargeable capacity, and dischargeable capacity for each ESS based on the charge status information of the ESS, and calculating the charge / dischargeable capacity for each ESS based on the current charge amount, minimum maintenance charge amount, maximum allowable charge amount, chargeable capacity, and dischargeable capacity for each ESS; a step of calculating the capacity capable of responding to peak times by arranging and summing the expected dischargeable capacity for each vehicle group and the charge / dischargeable capacity for each ESS, which are mapped to the expected connectionable time period for each vehicle group, according to the peak time period of the power system;and may include the step of deriving an integrated resource operation profile by combining the above-mentioned solar surplus power amount by time period, the expected connection possible time period by vehicle group, the expected discharge possible capacity by vehicle group, the charge / discharge possible capacity by ESS, and the peak time response possible capacity into a multidimensional state vector based on the time axis.;

[0223] The step of generating the above-mentioned linked resource operation profile may be a step of quantitatively structuring how solar assets, electric vehicle groups, and ESS can operate complementarily with each other on the same time axis. Here, the above-mentioned linked resource operation profile may not be simple power generation information or simple charging history information, but rather multidimensional resource status information that includes the amount of solar surplus power by time of day, the expected connection time slots for each vehicle group, the expected discharge capacity for each vehicle group, the charge / discharge capacity for each ESS, and the capacity capable of responding to peak times. Therefore, the above-mentioned linked resource operation profile can be utilized as basic data for generating peak response analysis information, linked resource operation standard information, and resource control plans in subsequent steps.

[0224] In the step of calculating the estimated solar power generation amount by time period based on the above real-time power generation status data and weather environment data, the amount of electricity that can be produced in each time interval in the future can be predicted by combining the current operating status of the solar power plant and external weather conditions.

[0225] Here, the real-time power generation status data may include the DC-side voltage, DC-side current, AC output, cumulative power generation, panel temperature, inverter efficiency, and equipment operation status information of the solar asset, and the weather environment data may include solar irradiance, ambient temperature, humidity, cloud cover, wind speed, and wind direction. The hourly estimated solar power generation refers to a predicted value of power production derived continuously along a time axis, such as, for example, that 50 kWh will be produced at 10:00 AM tomorrow and 100 kWh at noon. This can be calculated by comprehensively reflecting cloud movement, panel efficiency degradation due to temperature rise, and current output trends, rather than a simple calculation proportional to solar irradiance. The hourly estimated solar power generation can be used as a reference value to calculate the solar surplus power by comparing it with the estimated charging demand power amount in a subsequent step.

[0226] In the step of calculating the estimated charging demand power amount by time period based on the above vehicle-specific charging power amount data by time period, and calculating the surplus solar power amount by time period by subtracting the estimated charging demand power amount by time period from the estimated solar power generation amount by time period, it is possible to distinguish between the portion of the produced solar power that is preferentially used for electric vehicle charging and the remaining portion.

[0227] Here, the aforementioned data on charging power consumption by vehicle and time of day may refer to the amount of power consumed by an individual electric vehicle at a charger or the amount of charging power required during a specific time interval. The estimated charging demand power consumption by time of day may refer to the total sum of charging power expected to be needed during a specific time period in the future, based on how much power electric vehicles have typically consumed during that time period. For example, if the estimated solar power generation during a specific time period is 120 kWh and the vehicle charging demand power consumption during the same time period is 80 kWh, 40 kWh can be calculated as surplus solar power. This calculated surplus power consumption can be used as a leading indicator for ESS charging or for securing resources to respond to peak hours.

[0228] In the step of calculating the average connection start time, average connection end time, connection frequency by time period, and average connection duration by vehicle based on the cumulative charging history data and charging power amount data by time period for each vehicle, it is possible to quantitatively analyze when each vehicle is connected to the charger and how long it stays.

[0229] Here, the aforementioned cumulative charging history data may include past connection history, charging start time, charging end time, charging duration, and charging energy amount for each vehicle. The average connection start time and average connection end time may be representative times indicating when a specific electric vehicle typically plugs in and disconnects the charger; the connection frequency by time of day may be a probabilistic indicator representing the proportion of times the charger was connected during a specific time period; and the average connection duration may be a physical connection guarantee time indicating how long the connection state is typically maintained once connected. For example, if a specific vehicle exhibits a pattern of repeatedly connecting around 8:00 AM and disconnecting around 6:00 PM on weekdays, the device may calculate this as a representative connection pattern. Such information can be utilized as a basis for deriving the expected connection time zones for each vehicle group.

[0230] In the step of calculating the discharge participation frequency, average discharge duration, and average discharge contribution power amount by time of day for each vehicle based on the above-mentioned discharge participation history data for each vehicle, it is possible to evaluate whether the vehicle is simply parked and performing only charging, or whether it is a vehicle that is highly likely to actively participate in V2G services and provide power to the actual power grid.

[0231] Here, the aforementioned discharge participation history data may include the number of times a vehicle actually participated in discharge during specific past peak hours or demand response event periods, the discharge start time, the discharge end time, the discharge duration, and the discharge power amount. The aforementioned discharge participation frequency by time period may refer to the ratio of times a vehicle accepted a discharge request and actually participated in reverse discharge when a request was received in the past, and the aforementioned average discharge contribution power amount may refer to the average amount of power provided to the grid when participating in a single discharge. For example, some vehicles may frequently participate in discharge during evening peak hours but with short durations, while others may participate less frequently but provide more power per participation. These differences can serve as important criteria when calculating the expected discharge capacity for each vehicle group in a subsequent step.

[0232] In the step of clustering vehicles having similar usage patterns into multiple vehicle groups based on the average connection start time per vehicle, the average connection end time per vehicle, the connection frequency by time period, and the average connection duration, the irregular usage behaviors of numerous individual electric vehicles can be grouped into several statistically similar groups.

[0233] Here, the aforementioned vehicle group may refer to a group of vehicles with similar charger connection times, dwell times, charging behaviors, and potential for discharge participation. For example, they may be classified into a group of daytime office workers who go to work at 9:00 AM and leave at 6:00 PM, a group of residential vehicles that return home in the evening and connect at night, or a group of commercial facility visitors who visit irregularly during the day. The device can classify the aforementioned vehicles into multiple vehicle groups using a clustering algorithm, thereby reducing variability at the individual vehicle level while ensuring controllability at the vehicle group level. In other words, by using a vehicle group-level model, the uncertainty of individual vehicle predictions can be statistically offset.

[0234] In the step of deriving the expected connection time zones for each vehicle group based on the clustering results, it is possible to derive, in the form of time series segments, at which time of day each vehicle group is highly likely to be connected to the charger, enabling communication with the power grid and power trading.

[0235] Here, the aforementioned expected connection time zone may refer to a time period when a large number of vehicles belonging to a vehicle group are expected to be connected to the charger with a high probability. For example, for a residential vehicle group at night, the period between 9 PM and 6 AM the following day may be identified as a highly reliable expected connection time zone, while for a commercial parking vehicle group, the period between 9 AM and 5 PM may be identified as the primary connection time zone. This time zone information can be utilized as key information to determine when a specific vehicle group can be mobilized as an actual discharge resource.

[0236] In the step of calculating the expected discharge capacity for each vehicle group based on the discharge participation frequency, average discharge duration, and average discharge contribution power of the vehicles included in the vehicle group, it is possible to quantify how much of a discharge resource the vehicle group can be utilized in actual peak response.

[0237] Here, the estimated dischargeable capacity per vehicle group refers to the amount of dischargeable power or energy that the vehicle group is expected to supply to the grid during the expected connection time period. The device can conservatively calculate the dischargeable capacity at the vehicle group level by combining the discharge participation frequency, average discharge duration, and average discharge contribution power of the vehicles included in the same vehicle group. For example, if a specific vehicle group consists of 10 vehicles, and each vehicle can provide an average of 2 kWh of power but has a discharge participation frequency of 50 percent, the device can calculate the estimated dischargeable capacity of that vehicle group to be approximately 10 kWh. Accordingly, the estimated dischargeable capacity per vehicle group can be used as a key resource variable to determine the capacity capable of responding to peak times.

[0238] In the step of calculating the current charge amount, minimum maintenance charge amount, maximum allowable charge amount, chargeable capacity, and dischargeable capacity for each ESS based on the charge status information of the above ESS, and calculating the chargeable / dischargeable capacity for each ESS based on the current charge amount, minimum maintenance charge amount, maximum allowable charge amount, chargeable capacity, and dischargeable capacity for each ESS, the available energy range of the ESS operating as a stationary type, unlike a moving electric vehicle, can be accurately calculated.

[0239] Here, the current charge amount may refer to the amount of energy currently stored inside the ESS, the minimum maintenance charge amount may be a lower limit that must be maintained for battery protection and securing emergency reserve capacity, and the maximum allowable charge amount may be an upper limit allowed within the safe operation range. For example, if the safe operation range of the ESS is set between 20 percent and 80 percent and the current charge amount is 50 percent, the ESS may be evaluated as having a margin to be charged up to an additional 30 percent and, conversely, discharged up to 30 percent. Accordingly, the device can determine the chargeable and dischargeable capacities for each ESS by calculating the chargeable and dischargeable capacities for each ESS and combining them.

[0240] In the step of calculating the capacity capable of responding to peak times by arranging and summing the expected discharge capacity for each vehicle group and the charge / discharge capacity for each ESS, which are mapped to the expected connection time slots for each vehicle group, according to the peak time slots of the power system, the total resource size available for use at the time of responding to the actual peak load can be calculated.

[0241] Here, the aforementioned peak time response capacity may refer to the total amount of active power that can actually be supplied or absorbed by mobilizing vehicle groups and ESS during times when demand in the power system is concentrated. For example, if there are two vehicle groups that can be connected during a specific peak time, and the sum of the expected dischargeable capacities of those vehicle groups is 150 kWh, and the dischargeable capacity of the ESS is 80 kWh, the device can calculate the peak time response capacity as approximately 230 kWh. Such a value can be used to generate peak response analysis information in a subsequent step.

[0242] In the step of deriving a linked resource operation profile by combining the above-mentioned solar surplus power amount by time period, the expected connection time period by vehicle group, the expected discharge capacity by vehicle group, the charge / discharge capacity by ESS, and the peak time response capacity into a multidimensional state vector based on the time axis, the previously calculated individual elements can be packaged into a single integrated operation information structure.

[0243] Here, the aforementioned multidimensional state vector may refer to a structured data representation in which solar surplus power, vehicle group connection status, vehicle group discharge capacity, ESS charge / discharge capacity, and peak response capacity are arranged together at each point in time along the time axis. The device can generate a linked resource operation profile by aligning each of the above elements along the same time axis and then combining them into a state vector for each time interval. The linked resource operation profile derived in this way can serve as a basic blueprint for a future smart contract or artificial intelligence control system to determine when to store solar surplus power in the ESS and when to deploy the vehicle group's battery or ESS to the power grid. Furthermore, in subsequent steps, it can be utilized as an input value for generating peak response analysis information that calculates the expected load reduction amount by peak time, the expected settlement unit price by time, the expected discharge revenue, and the expected charge / discharge cost.

[0244] And, the step of determining the settlement plan and profit distribution plan comprises: a step of extracting the discharge participation priority by vehicle group, ESS charge / discharge control criteria, peak time period response criteria, and settlement criteria included in the linked resource operation standard information, and identifying peak control event intervals according to the peak time period response criteria; a step of extracting the estimated load reduction amount by peak time period, estimated settlement unit price by time period, estimated discharge revenue by time period, and estimated charge / discharge cost by time period included in the peak response analysis information, and setting the discharge participation priority by vehicle group and ESS charge / discharge control criteria as constraints; a step of constructing a Net Profit Objective Function by summing the estimated Demand Response (DR) settlement amount corresponding to the estimated load reduction amount by peak time period and the estimated discharge revenue by time period, and subtracting the estimated charge / discharge cost by time period and battery degradation loss cost. A step of applying Mixed-Integer Nonlinear Programming (MINLP) to maximize the output value of the above net profit objective function to derive a charging / discharging schedule per ESS that allocates the ESS, a deterministic resource, as a base peak reduction resource, and an expected discharge participation schedule per vehicle group that allocates the vehicle group, a probabilistic resource, as a grid variability response resource, and to finalize a resource control plan including the derived charging / discharging schedule per ESS and the expected discharge participation schedule per vehicle group; a step of calculating the system-level peak reduction contribution amount by calculating the difference between the Customer Baseline Load (CBL) of the target power receiving facility and the actual grid power received amount after the peak control event according to the above resource control plan ends; and a step of finalizing the actual power provided by individual resources that were sent back to the power grid during the peak control event period based on power measurement data collected from bidirectional meters connected to individual electric vehicles and ESS included in the above vehicle group.The method may include: a step of calculating a total settlement amount received from the power market based on the above-mentioned peak reduction contribution amount; a step of calculating an integrated profit by deducting a platform operation fee from the above-mentioned total settlement amount; a step of calculating performance indicators for each individual vehicle based on the discharge participation capacity, power response speed, and schedule fulfillment rate for each individual vehicle; a step of calculating V2G compensation for each individual subscription account in proportion to the above-mentioned performance indicators for each individual vehicle; a step of establishing a settlement plan so that the V2G compensation for each individual subscription account is deducted from the next month's rental fee of the solar subscription service; a step of calculating surplus profit by deducting the total amount of V2G compensation for each individual subscription account from the above-mentioned integrated profit; and a step of determining a profit distribution plan so that the surplus profit is divided and allocated according to the ESS owner's equity ratio.

[0245] The step of determining the above settlement plan and profit distribution plan may be a step of calculating the total amount of settlement funds secured from the power market by reflecting together the forecast information derived from the preliminary planning stage of peak response resources and the implementation results of actual peak control events, and reasonably distributing it to each participating entity.

[0246] Here, the settlement plan may refer to a financial processing plan that includes amounts received from the power market, compensation to be attributed to individual participants, fees to be deducted, and details for reflecting the settlement of the next rental fee, and the revenue distribution plan may refer to a plan that determines the share of the integrated revenue to be attributed to each stakeholder, such as individual subscription accounts and ESS owners. Therefore, the above step may not simply calculate peak reduction performance, but may be a step that determines the final economic attribution structure by linking the predicted revenue structure before peak control with the actual performance after peak control.

[0247] In the step of extracting the priority of discharge participation by vehicle group, ESS charge / discharge control criteria, peak time period response criteria, and settlement criteria included in the above linked resource operation standard information, and identifying the peak control event interval according to the above peak time period response criteria, the time interval subject to actual control and settlement and the operation standard to be applied first can be specified.

[0248] Here, the priority of discharge participation by vehicle group may be a value indicating which vehicle group among multiple vehicle groups will be utilized as a discharge resource first, and the ESS charge / discharge control criterion may be a criterion for determining whether to utilize the ESS as a base peak reduction resource or maintain it as a reserve resource. In addition, the peak time period response criterion may be a criterion for identifying a time period when the grid load exceeds a preset threshold, a time period when a demand response event is requested, or a time period when the power unit price rises rapidly as a peak control event period. For example, the device may identify the time period from 2:00 PM to 5:00 PM as a peak control event period by referring to the criterion information, and subsequent net profit evaluation and resource schedule determination may be performed based on this period.

[0249] In the step of extracting the expected load reduction amount by peak time, the expected settlement unit price by time, the expected discharge revenue by time, and the expected charge / discharge cost by time included in the peak response analysis information, and setting the discharge participation priority by vehicle group and ESS charge / discharge control criteria as constraints, forecast-based economic information and operational criteria necessary for establishing actual resource control and settlement plans can be organized together. Here, the expected load reduction amount by peak time may refer to the magnitude of the power grid load expected to be reduced by utilizing vehicle groups and ESS during a specific peak control event period, and the expected settlement unit price by time may be the expected compensation unit price to be applied in the corresponding time period according to the demand response settlement system or power trading settlement criteria. The expected discharge revenue by time may refer to the expected compensation amount obtainable by vehicle groups and ESS performing discharge, and the expected charge / discharge cost by time may include ESS charge / discharge loss costs, battery operation costs, and system operation costs. For example, as a standard for ESS charge / discharge control, a rule may be set to limit the remaining ESS capacity so that it does not fall below 20 percent, and this can function as a constraint to prevent excessive discharge instructions from being issued during the subsequent optimization process.

[0250] In the step of constructing a net profit objective function by summing the expected demand response settlement amount corresponding to the expected load reduction amount for each peak time period and the expected discharge revenue for each time period, and subtracting the expected charge / discharge costs and battery degradation loss costs for each time period, a criterion function can be generated that can mathematically determine whether a specific peak control strategy actually brings economic benefits.

[0251] Here, the above-mentioned estimated demand response settlement amount may be an amount expected to be paid by the power market or grid operator corresponding to the load reduced during peak hours, and the above-mentioned estimated discharge revenue per hour may be market revenue calculated based on the discharged power amount. On the other hand, the above-mentioned estimated charge / discharge cost per hour may include charging energy costs, conversion loss costs, and operating costs, and the above-mentioned battery degradation loss cost may be a value converted into a monetary amount representing the decrease in asset value due to the reduction in the lifespan of the vehicle battery or ESS battery. The device may construct a net profit objective function by adding and subtracting each of the above items, and this may be a function to be maximized in a subsequent optimization problem. For example, even if there exist two schedules capable of achieving the same peak reduction amount, if one has a high ESS degradation cost and the other has a low vehicle group participation cost, the device may assign a higher net profit value to the latter schedule.

[0252] In the step of applying a mixed integer nonlinear programming method to maximize the output value of the above net profit objective function to derive a charging and discharging schedule for each ESS that allocates a deterministic resource, the ESS, as a base peak reduction resource, and an expected discharge participation schedule for each vehicle group that allocates a probabilistic resource, the vehicle group, as a grid variability response resource, and determining a resource control plan including the derived charging and discharging schedule for each ESS and the expected discharge participation schedule for each vehicle group, an optimal peak response scenario that simultaneously considers economic efficiency and resource characteristics can be derived.

[0253] Here, the ESS can be treated as a deterministic resource because its installation location and controllable state are fixed, while the vehicle group can be treated as a probabilistic resource because its actual connection and participation depend on a certain probability. By applying mixed-integer nonlinear programming, the device can determine at which time period to prioritize discharging the ESS and at which time to induce the vehicle group's participation in discharge, thereby finding the combination that maximizes the net profit objective function. For example, a resource control plan can be derived in which the ESS, which has a fast response and high controllability, is prioritized during the early part of the peak time, and the vehicle group, which has a sufficiently high probability of participation, is deployed as an additional discharge resource after the mid-point. This finalized resource control plan can serve as a standard for executing subsequent peak control events and measuring actual performance.

[0254] In the step of calculating the system-level peak reduction contribution by calculating the difference between the customer reference load of the target receiving facility and the actual grid power received amount after the peak control event according to the resource control plan above is completed, the extent to which peak control was actually performed can be quantified based on the grid power received amount. Here, the customer reference load may refer to the reference load that the target receiving facility is estimated to have originally consumed had peak control not been performed, and the actual grid power received amount may refer to the amount or value of power actually received from the grid during the peak control event period. The device can calculate the system-level peak reduction contribution by calculating the difference between the customer reference load and the actual grid power received amount. For example, if the reference load was estimated to be 500kW during the peak control event period but the actual power received amount was measured as 380kW, the device may determine that a peak reduction contribution equivalent to 120kW has been achieved. This peak reduction contribution can serve as a direct basis for calculating the total settlement amount to be received from the power market.

[0255] In the step of determining the actual amount of power supplied by individual resources to the power grid during a peak control event period based on power measurement data collected from bidirectional power meters connected to individual electric vehicles and ESS included in the above vehicle group, it is possible to determine, based on measurement, how much power each individual resource actually supplied, separately from the total peak reduction contribution of the system.

[0256] Here, the bidirectional power meter may be a metering device capable of measuring both the amount of power discharged by an electric vehicle or ESS to the external power grid and the amount of power received from the external power grid. The device can calculate the amount of power actually supplied by individual resources back to the power grid during peak control event periods by analyzing measurement data collected from the bidirectional power meters corresponding to each electric vehicle and ESS. For example, if a specific vehicle supplies 8 kWh, another vehicle supplies 5 kWh, and an ESS supplies 30 kWh to the power grid, the device can separately determine the actual amount of power supplied by each resource. This actual amount of power supplied can be directly utilized for calculating subsequent performance indicators and determining compensation for each individual subscription account.

[0257] In the step of calculating the total settlement amount received from the power market based on the above-mentioned peak reduction contribution, the actual peak reduction performance at the system level can be converted into a monetary amount according to market settlement rules. Here, the above-mentioned total settlement amount may refer to the total amount of compensation or settlement payments made according to the power market or the demand response settlement system. The device may calculate the total settlement amount by applying the demand response settlement unit price, the peak response unit price, or the market compensation unit price to the previously calculated peak reduction contribution.

[0258] In the step of calculating the integrated profit by deducting platform operation fees from the total settlement amount, the actual distributable integrated profit can be calculated by first subtracting the costs required for the operation of the service platform from the total profit received from the market. Here, the platform operation fees may include data collection costs, communication costs, settlement system operation costs, control platform usage fees, and other service fees. The device can calculate the integrated profit by deducting these platform operation fees from the total settlement amount, and the integrated profit can be used as a base amount for individual participant compensation and the distribution of remaining profits.

[0259] In the step of calculating performance indicators for individual vehicles based on the discharge participation capacity, power response speed, and schedule fulfillment rate for each individual vehicle, it is possible to quantify how faithfully and effectively each vehicle participated in the peak control event. Here, the discharge participation capacity may be the amount of power actually provided by a specific vehicle, the power response speed may refer to the time from the control command until the actual discharge begins or the speed of reaching the output, and the schedule fulfillment rate may refer to the ratio of actual fulfillment to the planned discharge participation schedule. The device can calculate performance indicators for individual vehicles by combining these elements, and, for example, may assign higher performance indicators to vehicles that provide a large amount of actual power, have a fast response, and a high schedule fulfillment rate.

[0260] In the step of calculating V2G rewards for each individual subscription account in proportion to the performance indicators for each individual vehicle mentioned above, the amount of the integrated revenue to be allocated to the vehicle owner can be determined based on the results of the previously assessed performance evaluation. Here, the V2G reward refers to the amount received as compensation for the electric vehicle owned or used by the individual subscription account participating in a peak control event. Vehicles or subscription accounts with higher performance indicators may be allocated more rewards, while conversely, those with low participation or unstable responses may be allocated relatively lower rewards.

[0261] In the stage of establishing a settlement plan so that the V2G compensation for each individual subscription account is deducted from the following month's rental fee of the solar subscription service, the compensation to be paid to individual subscribers can be reflected in the form of a rental fee reduction rather than a cash payment. This method simplifies the settlement process for platform operations and allows subscribers to directly experience the benefits in the rental fee billed the following month. For example, if the V2G compensation for a specific subscription account is calculated at 30,000 won, the device can establish a settlement plan so that 30,000 won is deducted from the following month's rental fee. This method can have the effect of preventing subscribers from leaving the service and encouraging continuous participation in discharge.

[0262] In the step of calculating surplus profit by deducting the total amount of V2G compensation per individual subscription account from the above-mentioned integrated revenue, the final economic value remaining on the platform after excluding the compensation to be attributed to individual vehicle participants can be calculated. Here, the above-mentioned surplus profit may be the amount remaining after deducting the total compensation distributed to subscribers from the integrated revenue, and this may serve as a basis for subsequent distribution to ESS owners or profits attributed to other operators.

[0263] In the step of determining the profit distribution plan so that the above surplus profit is divided and allocated according to the ESS owner's shareholding ratio, the profit remaining after V2G participation rewards may be allocated to the entity that provided the ESS according to the shareholding ratio.

[0264] Here, the above ESS owner share ratio may refer to the economic or contractual share that each owner holds in the integrated ESS resource when there are multiple ESS owners. The device may calculate the allocated amount per ESS owner by applying the above share ratio to the surplus profit and finalize this as a profit distribution plan. The step of determining the settlement plan and profit distribution plan in this manner may be a step that enables the determination of market settlements, platform revenue, individual subscriber rewards, and ESS owner distributions within a consistent framework based on the actual performance of peak control events.

[0265] According to one embodiment, a method for operating a blockchain platform for the division of ownership and subscription revenue distribution of a solar power plant based on Security Token Offering (STO), wherein the device includes a processor, memory, a communication module, and a non-transient storage medium, and the processor executes a program stored in the non-transient storage medium, comprises: a step of collecting integrated operational data including weather environment data, power market indicators, and real-time power generation status data of a solar asset; a step of analyzing the integrated operational data to generate operational analysis information; and a step of generating operational execution information including at least one of an asset operation plan, a resource control plan, a settlement plan, a revenue distribution plan, a maintenance plan, and a contract management plan based on the operational analysis information; wherein the step of generating operational analysis information comprises: a step of aligning the power generation capacity, construction cost, expected lifespan, and real-time power generation status data of the solar asset included in the integrated operational data to generate a real asset tokenization profile reflecting the issuance size of a Security Token Offering (STO) that divides the total ownership of the solar asset into predetermined units, the share of rights per token, the value depreciation rate, and the characteristics of the asset return rate by time period. and a step of calculating asset return analysis information including total power sales revenue based on real-time power generation and power unit price, deduction of platform operation fees, asset maintenance reserves, and expected dividends per unit of token security by combining weather environment data and power market indicators with the real asset tokenization profile; wherein the step of generating the operation execution information comprises: generating token asset operation standard information including dividend cycles, conditions for automatic transfer of proceeds, token transaction approval criteria, and tax withholding criteria of a smart contract running on a blockchain distributed ledger, based on the asset return analysis information;The present invention provides a method for operating a blockchain platform for the division of ownership and distribution of subscription revenue of a security to token (STO)-based solar power plant, comprising the step of: inputting actual generated power sales revenue and subscription fees into smart contract logic based on the asset revenue analysis information and token asset operation standard information to calculate dividends proportional to the token share ratio held by each subscriber, and hashing and recording the distribution details on a blockchain network to finalize the settlement plan and revenue distribution plan.

[0266] The step of collecting integrated operational data is to secure external environmental information, power market information, and power generation facility operation information in an interconnected form to calculate the asset value, profitability, and dividend potential of tokenized securities-based solar power plants in real time.

[0267] Here, integrated operational data refers to a data set in which meteorological environment data, power market indicators, and real-time power generation status data of solar assets are sorted according to the same time standard, the same asset identification system, and the same settlement standard system. For example, meteorological environment data may include hourly solar irradiance, ambient temperature, humidity, wind speed, wind direction, and cloud cover; power market indicators may include real-time System Marginal Price, Renewable Energy Certificate trading unit price, settlement standards, market fee structure, and price fluctuation information; and real-time power generation status data may include DC-side voltage, DC-side current, AC output, cumulative power generation, panel temperature, inverter operating status, and communication status of solar assets. The device can construct an integrated operational dataset capable of subsequent value calculation by sorting each of the above data by reference time and performing missing value correction, outlier removal, asset identification number mapping, and resampling based on the analysis cycle standard.

[0268] The step of generating operational analysis information is not merely to store the aforementioned integrated operational data, but to produce analysis results for determining the ownership division structure and time-based profitability of the solar assets.

[0269] Here, operational analysis information may refer to intermediate analysis results that include a physical asset tokenization profile generated based on the solar asset's power generation capacity, construction costs, expected lifespan, and real-time power generation status, as well as asset return analysis information calculated by reflecting meteorological data and power market indicators. In other words, the aforementioned operational analysis information can be distinguished from integrated operational data in that it is not raw data, but rather analysis results with financial and operational significance, such as the divisibility structure of the underlying asset, value depreciation structure, hourly rate of return characteristics, power sales revenue, platform fees, and expected dividends.

[0270] The step of creating a real asset tokenization profile is to divide a physical asset, such as an actual solar power plant, into tradable token securities on the blockchain and quantitatively define the rights represented by each token and the structure of value fluctuation.

[0271] Here, the real asset tokenization profile may refer to asset structure information including the issuance size of token securities divided into predetermined units of the total ownership of the solar asset, the share of rights per token, the value depreciation rate, and characteristics of asset return by time period. For example, the device may align the power generation capacity, construction cost, expected lifespan, and real-time power generation status data of the solar asset included in the integrated operation data to determine the total number of issued tokens based on the total economic value of the asset, and calculate the ratio of individual tokens to the total ownership and the collateral facility capacity per unit token. Additionally, the device may calculate the value depreciation rate by combining the straight-line depreciation rate based on the expected lifespan and elapsed operating period with the physical cumulative degradation rate estimated from the real-time power generation status data, and determine the characteristics of asset return by time period by deriving the life-cycle levelized cost of generation and the dynamic internal rate of return curve based on past time-series power generation patterns and the predicted power generation amount of the expected lifespan interval. This information can ultimately serve as a “blueprint for breaking down real-world solar power plants into digital units that can be held by small investors,” and can be structured as off-chain data and packaged into a form that can be referenced by smart contracts on the blockchain through hash values ​​and metadata for oracle integration.

[0272] The step of calculating asset return analysis information is to combine weather environment data and power market indicators with the above-mentioned real asset tokenization profile to calculate the actual profit structure that the solar asset can generate during the current time and future time intervals.

[0273] Here, asset return analysis information may refer to the results of a return analysis that include total power sales revenue based on real-time power generation and unit prices, deductions for platform operation fees, asset maintenance reserves, and estimated dividends per unit of token securities. The device can load the value depreciation rate, the share of rights per token, and the characteristics of asset return by time period after verifying the integrity of off-chain data corresponding to the real asset tokenization profile using a hash value. Subsequently, the device can calculate the effective power generation efficiency per unit hour by combining real-time solar radiation forecasts, ambient temperature, and time period information from weather environment data with the DC side voltage drop and module temperature variability included in real-time power generation status data, and calculate the real-time power generation amount by time period by combining this with the power generation capacity of the solar asset. Additionally, the device can calculate the unit price of electricity by time period by combining the real-time system marginal price and the trading unit price of renewable energy certificates, and then apply this to the real-time power generation amount by time period to derive the power sales revenue by time period and the total power sales revenue.

[0274] The device goes a step further to estimate marginal maintenance costs based on the deviation between the straight-line depreciation rate and the physical cumulative degradation rate, calculate asset maintenance reserves by applying risk hedging ratios, and calculate platform operation fee deductions by reflecting gas costs, oracle integration processing fees, and entrusted operation rates. Subsequently, it calculates distributable net profit by subtracting the asset maintenance reserves and platform operation fee deductions from total power sales revenue, and calculates the estimated dividend per unit of token security by applying a dividend reflection coefficient based on the rights share per token and the characteristics of asset returns by time period. In other words, the above step can be understood as a step that calculates “how much money is actually earned today or over a specific future period, how much of that is set aside as expenses, and consequently how much can be distributed per unit of token” on top of the previously created tokenization structure. These values ​​can be structured into interconnected revenue analysis records to generate asset return analysis information.

[0275] The step of generating operational execution information is to convert the above operational analysis information into actionable standard information and settlement results that can be reflected in the criteria for actual dividend execution and transaction approval.

[0276] Here, operational execution information may be execution information including at least one of an asset operation plan, a resource control plan, a settlement plan, a profit distribution plan, a maintenance plan, and a contract management plan; in this embodiment, in particular, token asset operation standard information, a settlement plan, and a profit distribution plan may be primarily generated. That is, operational execution information can be distinguished from operational analysis information in that it is an execution step that goes beyond merely analyzing how much profit an asset generates, and includes determining the frequency and conditions for distributing that profit and the criteria for approving token transactions.

[0277] The step of generating token asset operation standard information is a step of setting dividend and transaction rules that a smart contract running on a blockchain distributed ledger must follow based on the asset return analysis information.

[0278] Here, the token asset operation standard information may be standard information including dividend cycles, conditions for automatic profit transfers, token transaction approval criteria, and tax withholding criteria. For example, the device may set dividend cycles on a monthly, quarterly, or specific generation profit accumulation interval basis, and may define conditions for automatic profit transfers that trigger an automatic transfer when the distributable net profit exceeds a preset amount. Additionally, token transaction approval criteria may include a minimum holding period, KYC verification status, a range of transferable quantities, and whitelist conditions on the smart contract; and tax withholding criteria may include tax rates by tax bracket and rules for designating taxpayers. Ultimately, this may be a set of execution rules that allow the code to automatically settle money, approve token transactions, and deduct taxes without human intervention.

[0279] The step of finalizing the settlement and profit distribution plan involves inputting the actual generated electricity sales revenue and subscription fees into the smart contract logic based on the aforementioned asset profit analysis information and token asset operation standard information, calculating dividends proportional to the token shareholding ratio of each subscriber, and recording this on the blockchain network to finally confirm the settlement and profit distribution. Here, the actual generated electricity sales revenue may refer to revenue determined by the actual generation performance of solar assets and the actual market price during a fixed settlement cycle, and the subscription fee may refer to regular charges received during the service operation process. The device can input the actual electricity sales revenue and subscription fees into the smart contract logic to recalculate the distributable net profit or compare it with the scheduled dividend, and then calculate the dividends proportional to the token shareholding ratio of each subscriber. Subsequently, the distribution details, including the calculated dividends, recipients, settlement time, and transaction identifiers, can be hashed and recorded on the blockchain network, thereby enabling the implementation of a transparent settlement structure that prevents subsequent tampering. In other words, it can be understood as a step in which, when actual money is deposited according to previously established rules, it is automatically distributed to each subscriber's wallet or account, and the record is permanently stored in an tamper-proof form.

[0280] Ultimately, the method for operating a blockchain platform for the division of ownership and distribution of subscription revenue for solar power plants based on the aforementioned token securities may be a method that enables a series of procedures to be executed within a single integrated framework: structuring physical solar assets into token securities units based on integrated operational data; calculating asset profit analysis information reflecting weather, market, and power generation conditions; and finalizing smart contract-based operational standards along with actual settlement and revenue distribution results on a distributed ledger. Accordingly, the revenue structure of the actual solar power plant can be converted into tradable rights units on the blockchain, while simultaneously ensuring the integrity of dividend calculations and settlement records for each token holder.

[0281] In addition, the step of generating the real asset tokenization profile comprises: configuring underlying asset calculation data by extracting the total construction cost, power generation capacity, expected lifespan, and real-time power generation status data of the solar asset from the integrated operational data; calculating the total number of tokens to be issued to liquidate the total ownership of the solar asset by applying a pre-established fiat currency-based face value split unit to the total construction cost, and determining the total number of tokens to be issued as the issuance size of the Security Token Offering (STO); determining the ratio that one unit token occupies in the total ownership of the solar asset as the right share per token based on the total number of tokens to be issued, and calculating the collateral facility capacity per unit token corresponding to the right share per token; calculating the straight-line depreciation rate based on financial accounting standards based on the expected lifespan and the elapsed operating period up to the present time; and estimating the physical cumulative degradation rate of the solar panel based on the DC voltage drop, the time-dependent output degradation rate, and module temperature variability included in the real-time power generation status data. A step of dynamically combining the above straight-line depreciation rate and physical cumulative degradation rate according to a time-weighted function to calculate a value depreciation rate as a time-series function that tracks the decline in the actual value of the underlying asset of the token security in a time-series manner; a step of calculating the Levelized Cost of Energy (LCOE) of the solar asset based on the past time-series power generation pattern generated by the accumulation of the above real-time power generation status data and the predicted power generation amount during the expected lifespan interval;The method may include the step of calculating an expected revenue flow by time period by reflecting a pre-established standard power unit price for asset valuation to the seasonal solar radiation deviation and hourly power production efficiency variability derived from the past time-series power generation pattern, and determining the characteristics of the asset return by time period by deriving a dynamic Internal Rate of Return (IRR) curve based on the expected revenue flow by time period and the life-cycle levelized cost of generation; the step of structuring the token security issuance scale, the right share ratio per token, the collateral facility capacity per unit token, the value depreciation rate, and the characteristics of the asset return by time period into off-chain data outside the blockchain platform; and the step of generating a hash value for data integrity verification regarding the off-chain data and deriving a real asset tokenization profile by packaging it into a metadata format for Oracle integration containing the hash value so that a smart contract to be deployed on a distributed ledger within the blockchain platform can reference it.

[0282] The step of generating the above-mentioned real asset tokenization profile may be a step of generating tokenization reference information by comprehensively analyzing the asset's cost structure, facility capacity, expected lifespan, actual operating status, value depreciation structure, and yield structure in order to structure the real physical asset, such as a solar power plant, into token securities that can be traded and distributed on the blockchain.

[0283] Here, the aforementioned physical asset tokenization profile may refer to asset structuring information including the issuance size of token securities, the share of rights per token, the collateral facility capacity per unit token, the value depreciation rate, and characteristics of asset return rates over time. Therefore, the above step can be understood not as a simple nominal equity split, but as a step of quantitatively converting the economic value and operational status of physical solar assets into digital rights units.

[0284] In the step of constructing basic asset calculation data by extracting data on the total construction cost, power generation capacity, expected lifespan, and real-time power generation status of solar assets from the aforementioned integrated operational data, the most basic financial information and physical facility information for issuing solar power plants in the form of token securities can be organized.

[0285] Here, the aforementioned total construction cost may be the total investment cost including land acquisition costs, panel and inverter purchase costs, structure construction costs, wiring and junction box installation costs, supervision costs, and grid connection costs. The aforementioned power generation capacity may refer to the maximum power scale that the power plant can produce under rated conditions, and the aforementioned expected lifespan may refer to the total operating period during which the facility is expected to be operated economically or technically. Additionally, real-time power generation status data may include DC-side voltage, DC-side current, AC output, cumulative power generation, panel temperature, inverter efficiency, and alarm history. By extracting each of the above values ​​to construct basic asset calculation data, the device can utilize this as reference data for determining the token issuance scale and asset value in subsequent stages.

[0286] In the step of calculating the total number of tokens to be issued to liquidate the entire ownership of solar assets by applying a face value division unit based on a pre-established fiat currency standard to the total construction cost, and determining the total number of tokens to be issued as the size of the token security issuance, a standard can be established to divide the total economic value of the real assets into small rights units that can be traded by the general public.

[0287] Here, the aforementioned par value division unit may be a fixed division standard, such as a standard fiat currency face value per token, such as 1,000 won, 10,000 won, or 100,000 won. The device may calculate the total number of issued tokens by dividing the total construction cost by the aforementioned par value division unit, and the total number of issued tokens may represent the issuance scale indicating how many token security units the total ownership of the solar asset is divided into. For example, if the total construction cost is 1 billion won and the par value division unit is 5,000 won, the total number of issued tokens may be calculated as 200,000. The issuance scale calculated in this way may be used as a basis for calculating subsequent dividends and equity share ratios.

[0288] In the step of determining the ratio of one unit token to the total ownership of solar assets based on the above total number of issued tokens as the right share per token and calculating the collateral facility capacity per unit token corresponding to the right share per token, the size of the ownership represented by each token and the corresponding amount of physical assets can be determined.

[0289] Here, the ownership stake per token may represent the percentage of ownership in the total solar asset occupied by a single token, and is generally calculated as the reciprocal of the total number of issued tokens. Additionally, the collateralized facility capacity per unit token may be a value obtained by dividing the total generation capacity by the total number of issued tokens, indicating what scale of facility performance one token effectively corresponds to. For example, if the total generation capacity is 1 MW and the total number of issued tokens is 200,000, one unit token can be calculated as representing a facility capacity equivalent to 5 W. Through this, investors can clearly understand that the tokens they hold are not merely numbers, but are linked to a specific physical capacity and economic rights of the actual solar facility.

[0290] In the step of calculating the straight-line depreciation rate under financial accounting standards based on the above-mentioned expected lifespan and the elapsed operating period up to the present time, it is possible to accountably evaluate how the book value of an asset decreases over time.

[0291] Here, the straight-line depreciation rate may be a depreciation rate converted to an annual, monthly, or hourly basis, assuming that the total cost of building the asset decreases at a uniform rate over its expected life. For example, an asset with an expected life of 20 years may be subject to a straight-line depreciation rate of 5 percent per year, and the longer the elapsed operating period to date, the lower the residual book value may be. The device can calculate an accounting-based depreciation structure based on the expected life and elapsed operating period and utilize this as a basis for calculating the subsequent value depreciation rate.

[0292] In the step of estimating the physical cumulative degradation rate of the solar panel based on the DC side voltage drop, time-dependent output degradation rate, and module temperature variability included in the above real-time power generation status data, it is possible to technically evaluate how much the actual facility has aged, separate from the decline in accounting value on paper.

[0293] Here, the DC-side voltage drop may be an indicator reflecting the electrical performance degradation of the panel or string, the time-dependent output degradation rate may be a value indicating how much the actual output has decreased compared to the same weather conditions or expected output, and the module temperature variability may be a value reflecting the impact of repetitive thermal stress on the panel's lifespan. For example, even if only about 10 percent has been depreciated on the books, if the DC-side voltage drop and output degradation are greater than expected, the actual physical degradation rate may be assessed as 15 percent or more. The device can calculate the physical cumulative degradation rate of the panel by analyzing each of the above factors, which can serve as a technical value depreciation indicator showing how much the actual asset is experiencing performance degradation, separate from simple accounting depreciation.

[0294] In the step of dynamically combining the above straight-line depreciation rate and physical cumulative deterioration rate according to a time-weighted function to calculate a value depreciation rate as a time-series function that tracks the decline in the actual value of the underlying asset of the token security over time, an integrated value decline model that reflects both accounting value decline and physical value decline can be generated.

[0295] Here, the aforementioned time-weighted function may be a function designed to assign different weights to the accounting depreciation rate and the physical cumulative deterioration rate depending on the early, mid, or late stages of asset operation. For example, in the early stages of operation, the straight-line depreciation rate under financial accounting standards may have a relatively significant impact on the assessment of asset value decline, while as the asset enters the long-term operation stage, the influence of the physical cumulative deterioration rate, which represents the actual degree of machine aging, may be reflected more significantly. Through this time-weighted combination, the device can calculate a value depreciation rate that tracks the decline in the asset's actual value over time, which can be used as a key parameter for subsequent token dividends and valuation.

[0296] In the step of calculating the levelized cost of generation for the life cycle of a solar asset based on the historical time-series generation pattern generated by accumulating the above real-time generation status data and the predicted generation amount during the expected lifespan, it is possible to evaluate the average cost per unit of power produced by the asset over its entire lifespan.

[0297] Here, the aforementioned life cycle levelized cost of electricity generation may be a cost indicator calculated by reflecting construction costs, maintenance costs, and long-term power generation forecasts. The device can calculate the life cycle levelized cost of electricity generation by estimating the total expected power generation by combining past power generation patterns and the predicted power generation over the expected lifespan, and by allocating construction costs and other costs to it. The lower this value, the more the power plant can be evaluated as a high-efficiency asset that produces electricity at a relatively low cost, and it can be used as a reference value for calculating subsequent time-series expected revenue streams and internal rate of return curves.

[0298] In the step of determining the characteristics of the asset return rate by time by calculating the expected revenue flow by time by reflecting the pre-set standard power unit price for asset evaluation on the seasonal solar radiation deviation and the variability in power production efficiency by time derived from the past time series power generation pattern, and deriving a dynamic internal rate of return curve based on the expected revenue flow by time and the life cycle levelized cost of generation, it is possible to dynamically predict how much economic profitability the asset will have at each point in time.

[0299] Here, the aforementioned reference electricity unit price may be the electricity sales unit price, the settlement reference unit price, or a reference price applied for evaluation purposes. The device can derive an expected revenue stream by time of day by reflecting the characteristics of high power generation in certain seasons and low generation in others due to seasonal variations in sunlight, as well as differences in power generation efficiency by time of day. Subsequently, a dynamic internal rate of return curve can be calculated by comparing the above expected revenue stream with the levelized cost of generation over the life cycle, thereby allowing for the determination of whether the profitability of the asset increases or decreases during a specific time of day or season. This dynamic internal rate of return curve may be the result of quantifying the characteristics of the asset's return by time of day.

[0300] In the step of structuring the above-mentioned token security issuance scale, right share ratio per token, collateral facility capacity per unit token, value depreciation rate, and time-dependent asset return characteristics into off-chain data outside the blockchain platform, the various asset structuring parameters calculated earlier can be organized into external data in a format that can be referenced by smart contracts.

[0301] Here, the aforementioned off-chain data may be structured data stored in a data repository outside the blockchain but verifiable and referenceable by smart contracts within the blockchain. The device can organize each of the aforementioned parameters into a specific field structure, such as JSON, table records, or metadata schemas, thereby enabling the logic for token issuance, dividends, and transaction approval to reference the same data structure. In particular, since directly recording vast amounts of data—such as the real-time degradation rate of solar power plants or complex yield curves—onto the blockchain every time can lead to excessively high processing costs and reduced system processing speed, it may be more efficient to structure and store this data as off-chain data.

[0302] In the step of generating a hash value for data integrity verification regarding the above-mentioned off-chain data and packaging it into a metadata format for oracle integration containing the hash value so that a smart contract to be deployed on a distributed ledger within a blockchain platform can reference it, thereby deriving a real asset tokenization profile, the vast amount of asset data stored outside the blockchain and the smart contract inside the blockchain can be securely connected.

[0303] Here, the aforementioned hash value may be a summary value for integrity verification calculated for the entire off-chain data or a key record, and since the hash value may change if even a single item of the off-chain data is altered, the falsification or tampering of the data can be verified immediately. The above-mentioned metadata format for oracle integration may be an integration structure configured to allow a smart contract to retrieve and verify necessary data by referencing the above-mentioned hash value and the location of external data. The device can finally derive a real asset tokenization profile by generating metadata including the above-mentioned hash value and packaging it in a form that can be registered in a distributed ledger or referenced by a smart contract. The real asset tokenization profile derived in this way can function as key reference information for converting the physical value and revenue structure of an actual solar power plant into the rights structure of a digital token security.

[0304] And, the step of calculating the asset return analysis information comprises: verifying forgery or tampering of off-chain data corresponding to the real asset tokenization profile using hash values ​​included in the oracle integration metadata format, and obtaining characteristics of the value depreciation rate, the right share per token, and the asset return rate by time period from the verified off-chain data; calculating the effective power generation efficiency per unit hour by combining the real-time solar radiation forecast, ambient temperature, and time period information included in the weather environment data, the DC voltage drop and module temperature variability included in the real-time power generation status data, and the physical cumulative degradation rate included in the value depreciation rate; calculating the real-time power generation amount by time period based on the power generation capacity and effective power generation efficiency per unit hour of the solar asset; extracting the real-time System Marginal Price (SMP) and Renewable Energy Certificate (REC) trading unit price included in the power market indicator, and calculating the power unit price by time period by combining the real-time System Marginal Price and the REC trading unit price; A step of calculating hourly power sales revenue by applying hourly power unit prices to the real-time power generation amount for each hour, and calculating total power sales revenue by accumulating the calculated hourly power sales revenues; a step of calculating the deviation between the straight-line depreciation rate and the physical cumulative degradation rate, and estimating the probability of component failure within the remaining lifespan and the marginal maintenance cost based on the deviation; a step of calculating an asset maintenance reserve by applying a pre-set risk hedge ratio to the total power sales revenue to depreciate the marginal maintenance cost; and a step of calculating a platform operation fee deduction amount by summing the gas fee representing the network transaction processing cost of the blockchain platform, the oracle integration processing fee, and the operation cost based on the solar power plant entrusted operation rate.The method may include: a step of calculating a distributable net profit by subtracting an asset maintenance reserve and a platform operation fee deduction from the total power sales revenue; a step of calculating an estimated dividend per unit of token security by applying a dividend reflection coefficient based on the rights share ratio per token and the characteristics of asset return rates by time period to the distributable net profit; and a step of generating asset return analysis information by structuring the real-time power generation amount by time period, the power unit price by time period, the total power sales revenue, the platform operation fee deduction, the asset maintenance reserve, the distributable net profit, and the estimated dividend per unit of token security into an interconnected profit analysis record.

[0305] The step of calculating the above asset return analysis information may be a step of quantifying the revenue structure that a solar power plant can actually generate at a specific point in time or during a specific settlement cycle by combining the previously generated real asset tokenization profile with external environment information and power market information.

[0306] Here, the aforementioned asset return analysis information does not refer merely to calculated electricity sales revenue, but may be comprehensive financial analysis information structured by linking real-time power generation by time of day, electricity unit prices by time of day, total electricity sales revenue, platform operation fee deductions, asset maintenance reserves, distributable net profit, and estimated dividends per unit of token security. Therefore, the aforementioned asset return analysis information can be utilized as a key input value in subsequent stages to determine the dividend cycle, automatic transfer conditions, transaction approval criteria, and tax withholding standards of the smart contract.

[0307] Regarding off-chain data corresponding to the above-mentioned real asset tokenization profile, the authenticity of the data is verified using hash values ​​included in the oracle integration metadata format. In the step of obtaining characteristics such as the value depreciation rate, the share of rights per token, and the asset return rate by time period from the verified off-chain data, the key parameters required for analysis can be retrieved after first confirming whether the asset structure information stored outside the blockchain is identical to the original.

[0308] Here, the above hash value may be a summary value for integrity verification generated for the entire off-chain data or a core record, and the above oracle integration metadata format may be a structure defining the reference relationship between the smart contract and the external data storage. The device can recalculate the current hash value of the off-chain data stored on an external server and compare it with the hash value that was securely recorded on the blockchain to verify whether the two values ​​match. If even a single number or item in the data is changed, the hash value will change; therefore, the device can determine the data as invalid or contaminated and exclude it from analysis. By loading characteristics such as the value depreciation rate, the share of rights per token, and the asset return rate by time period only from off-chain data whose integrity has been verified in this way, the reliability of subsequent profit calculation and dividend calculations can be ensured.

[0309] In the step of calculating the effective power generation efficiency per unit time by combining the real-time solar radiation forecast, ambient temperature, and time zone information included in the above weather environment data, the DC side voltage drop and module temperature variability included in the real-time power generation status data, and the physical cumulative degradation rate included in the above value depreciation rate, it is possible to derive a “actual power generation efficiency” that reflects the actual operating environment and asset aging status, rather than a simple rated efficiency.

[0310] Here, the real-time solar irradiance forecast mentioned above may refer to the amount of solar radiation energy expected to reach the panel surface during a specific time period, and the ambient temperature and time period information may be environmental factors explaining the temperature rise and efficiency fluctuations of the panel. Additionally, the DC voltage drop may reflect performance degradation due to electrical losses, module temperature variability may reflect a decrease in output stability due to thermal stress, and the physical cumulative degradation rate may be a parameter indicating the degree of long-term performance degradation of the panel. For example, even if a power generation efficiency of 20 percent is expected due to sufficient theoretical solar irradiance, the effective power generation efficiency may drop to 15 percent if the module temperature rises excessively or the voltage drop is significant. The device can calculate the effective power generation efficiency for a specific time unit, such as 15 minutes or 1 hour, by combining each of the above information, and this may be a value that more realistically represents the power production capacity actually expected during that time period.

[0311] In the step of calculating real-time power generation by time period based on the power generation capacity and effective power generation efficiency per unit time of the solar asset, the amount of electricity that the solar power plant can actually produce during a specific time interval can be calculated. Here, the power generation capacity may be a value representing the rated power generation capacity of the solar asset, and the effective power generation efficiency per unit time may be an efficiency value achievable under actual operating conditions during that time period. The device can calculate real-time power generation by time period by reflecting the power generation capacity, the effective power generation efficiency, and, if necessary, solar irradiance information together. For example, even if a solar power plant has a power generation capacity of 1 MW, if the effective power generation efficiency during a specific time period is maintained at a level of 90 percent, the real-time power generation during that time period can be converted to a level of 0.9 MWh. The real-time power generation by time period calculated in this way can be used as a direct input value for calculating subsequent electricity sales revenue.

[0312] In the step of extracting the real-time System Marginal Price and the trading unit price of Renewable Energy Certificates included in the aforementioned power market indicators, and calculating the hourly power unit price by combining the real-time System Marginal Price and the Renewable Energy Certificate trading unit price, the economic value of one unit of power generation in the market can be determined for each time period. Here, the real-time System Marginal Price may refer to the wholesale power unit price at a specific point in time in the power trading market, and the Renewable Energy Certificate trading unit price may refer to the market price of certificates granted in correspondence with renewable energy generation performance. The device can calculate the hourly power unit price by combining the two price information, and at this time, a simple summation method, a weighted summation method, or a correction method based on policy criteria may be applied. For example, if the real-time System Marginal Price is high and the REC trading unit price also rises during a specific time period, the power unit price for that time period may be calculated to be relatively higher than that of other time periods. The hourly power unit price derived in this way can serve as a reference value for determining how much revenue the power generation for each time period actually generates.

[0313] In the step of calculating hourly power sales revenue by applying hourly power unit prices to the real-time power generation amount for each time period and calculating total power sales revenue by cumulatively summing the calculated hourly power sales revenues, both the revenue for individual time intervals and the cumulative revenue over the entire operation period can be calculated.

[0314] Here, hourly electricity sales revenue may be an amount calculated by multiplying the amount of electricity generated in a specific time interval by the electricity unit price of that interval, and total electricity sales revenue may be the total amount obtained by accumulating and summing the revenues from multiple time intervals. For example, the revenue for the first time interval can be calculated by multiplying the amount of electricity generated at 1:00 PM by the unit price at that time, and the revenue for the second time interval can be calculated by reflecting the amount of electricity generated and the unit price at 2:00 PM; then, the total electricity sales revenue can be calculated by sequentially accumulating these on a daily, monthly, or specific dividend period. The total electricity sales revenue calculated in this way can be utilized as the total revenue value that serves as the basis for calculating maintenance reserves, operating fee deductions, and expected dividends in subsequent stages.

[0315] In the step of calculating the deviation between the above straight-line depreciation rate and the physical cumulative degradation rate, and estimating the probability of component failure and the marginal maintenance cost within the remaining lifespan based on the deviation, the future maintenance burden can be predicted by utilizing the difference between the decrease in financial value and the actual physical performance degradation.

[0316] Here, the aforementioned divergence may be a value representing the difference between the accounting rate of depreciation and the actual rate of equipment deterioration. For example, even if the book value is assessed to be decreasing gradually financially, the divergence may be calculated to be significant if the actual rate of physical cumulative deterioration is increasing rapidly. Based on this divergence, the device can estimate the probability of a specific component failing during its remaining lifespan and the corresponding marginal maintenance cost, which can be used as a reference value for calculating subsequent maintenance provisions. In other words, the risk of future repair costs can be quantified by measuring the gap between the book asset value and the actual equipment aging.

[0317] In the step of calculating the asset maintenance reserve by applying a pre-set risk hedging ratio to the total electricity sales revenue to amortize the aforementioned maintenance marginal costs, a reserve can be established to cover equipment failures and repair costs that may occur in the future.

[0318] Here, the aforementioned risk hedging ratio may represent the percentage of total power sales revenue to be separately set aside to cover future maintenance risks. The device can calculate the risk hedging ratio by referencing the previously estimated marginal maintenance costs and calculate an asset maintenance reserve by applying it to the total power sales revenue. For example, even if total power sales revenue is high during a specific settlement period, a higher maintenance reserve may be established if there is a significant risk of future panel replacement or inverter repair. Accordingly, when calculating subsequent dividends, a conservative profit structure reflecting the risk reserve, rather than just the total revenue, can be implemented.

[0319] In the step of calculating the platform operation fee deduction amount by summing the gas fee representing the network transaction processing cost of the above-mentioned blockchain platform, the oracle integration processing fee, and the operation cost based on the solar power plant consignment operation rate, platform operation-related costs that must be deducted from the electricity sales revenue can be calculated.

[0320] Here, gas fees may be processing costs incurred to execute and record transactions on the blockchain network, oracle integration processing fees may be costs incurred for retrieving and verifying external data, and commissioned operation rates may be remuneration rates for power plant maintenance and operation agency services. The device may calculate the platform operation fee deduction amount by summing each of the above cost items, and this may be an item deducted from total revenue when calculating subsequent distributable net profit. As a result, the actual distributable amount may be adjusted to a value reflecting actual costs associated with blockchain and asset operations, rather than simple sales revenue.

[0321] In the step of calculating the distributable net profit by subtracting the asset maintenance reserve and platform operation fee deductions from the total power sales revenue mentioned above, the size of the profit actually distributable to token holders can be determined.

[0322] Here, the aforementioned distributable net profit may refer to the actual profit remaining after deducting essential expense items from total revenue. The device can calculate the distributable net profit by subtracting asset maintenance provisions and platform operation fee deductions from total power sales revenue, which serves as the direct basis for the subsequent calculation of expected dividends. In other words, the aforementioned distributable net profit may be a concept of net profit after risk reserves and operating expenses have already been reflected, rather than simple revenue.

[0323] In the step of calculating the expected dividend per unit of token security by applying a dividend reflection coefficient based on the rights share ratio per token and the characteristics of asset returns by time period to the above-mentioned distributable net profit, the distributable profit can be converted into a form that can be distributed to individual token units.

[0324] Here, the entitlement share per token may be the ownership ratio that an individual token holds within the total assets, and the dividend reflection coefficient may be an adjustment coefficient designed to reflect asset return characteristics over time or a specific dividend policy. The device can calculate the estimated dividend per unit of token security by applying the entitlement share per token and the dividend reflection coefficient to the distributable net profit. For example, even among tokens holding the same entitlement share, the estimated dividend per unit token may be calculated to be higher if the return characteristics of a specific period reflect favorably to dividends. Such a value can be used as an automatic dividend reference value during subsequent smart contract execution.

[0325] In the step of generating the asset profit analysis information by structuring the above-mentioned real-time power generation amount by time period, power unit price by time period, total power sales revenue, platform operation fee deduction amount, asset maintenance reserve, distributable net profit, and estimated dividend per unit of token security into interconnected profit analysis records, all previously calculated financial and operational parameters can be integrated into a consistent data structure that can be utilized by smart contracts and settlement systems.

[0326] Here, the aforementioned profit analysis record may refer to a structured data record in which input and output values, deduction items, and dividend calculation results are linked and stored by time period. The device can generate asset profit analysis information by grouping each of the above values ​​into a record format, thereby allowing power generation and price information at a specific point in time, total revenue, cost deduction structure, and estimated dividend per unit token to be managed as a single set of analysis information. The asset profit analysis information generated in this way can be utilized as direct input data to generate token asset operation standard information, including dividend cycles, automatic transfer conditions, token transaction approval criteria, and tax withholding criteria, in a subsequent stage.

[0327] According to one embodiment, a device includes a processor, memory, a communication module, and a non-transient storage medium, and a method for calculating differential subscription fees and managing contracts through credit risk scoring of solar subscribers, which is performed by the processor by executing a program stored in the non-transient storage medium, comprises: a step of collecting integrated operational data including weather environment data, power market indicators, and real-time power generation status data of solar assets; a step of analyzing the integrated operational data to generate operational analysis information; and a step of generating operational execution information including at least one of an asset operation plan, a resource control plan, a settlement plan, a profit distribution plan, a maintenance plan, and a contract management plan based on the operational analysis information; wherein the...

Claims

Claim 1 A method for detecting anomalies and optimizing component replacement cycles of a digital twin-based subscription-type solar asset, wherein the device includes a processor, memory, a communication module, and a non-transient storage medium, and the processor executes a program stored in the non-transient storage medium, comprising: a step of collecting integrated operational data including weather environment data, power market indicators, and real-time power generation status data of the solar asset; a step of analyzing the integrated operational data to generate operational analysis information; and a step of generating operational execution information including at least one of an asset operation plan, a resource control plan, a settlement plan, a revenue distribution plan, a maintenance plan, and a contract management plan based on the operational analysis information; wherein the step of generating the operational analysis information comprises: a step of synchronizing a digital twin object that replicates the physical specifications and real-time degradation status of the solar asset in a virtual space based on the integrated operational data; The method comprises the step of performing an accelerated degradation simulation of a target component by inputting weather environment data and asset operation parameters in a time-series manner into the digital twin object, and deriving the remaining useful life (RUL) and failure probability for each component by calculating the state error rate with a preset normal operation profile; and the step of generating the operation execution information comprises: the step of calculating the physical replacement cost of the target component, the expected loss of power sales due to failure, and the remaining subscription contract period of the subscriber based on the remaining useful life and failure probability; and the step of determining the optimal replacement time for each component to maximize the economic profit of the entire subscription cycle; and the step of deriving the remaining useful life and failure probability for each component comprises: the step of defining a solar panel module, a power semiconductor, and a DC link capacitor as target components; and the step of extracting weather factors including the cumulative UV exposure amount, temperature variation range, humidity variation range, number of temperature and humidity repetition cycles, wind speed variation, and wind load variation from the weather environment data.A step of extracting operating factors including DC-side voltage / current (IV) trajectories, AC-side voltage / current waveforms, output load rate, switching frequency, and conversion efficiency fluctuation amount from the above asset operating parameters and real-time power generation status data; a step of mapping the above weather factors and operating factors into stress factors representing thermal stress, electrical stress, mechanical stress, and environmental stress for each target component with respect to the Electrical Equivalent Circuit Model of the solar panel module constituting the digital twin object and the Multiphysics Thermal Model of the power conversion inverter; a step of performing a physics-based first-order acceleration simulation for the solar panel module based on the above stress factors to calculate Potential Induced Degradation (PID), micro-crack progression rate, and series resistance increase amount; and a step of extracting high-frequency noise components, switching ripple components, harmonic distortion components, and instantaneous temperature fluctuation components from the above real-time power generation status data and inputting them into a time-series deep learning model to the inverter internal power semiconductor (IGBT) and DC-Link capacitor (DC-Link A step of performing a data-driven secondary simulation to estimate the cumulative residual of nonlinear thermal fatigue of a capacitor; a step of synchronizing the results of the primary accelerated simulation and the secondary simulation on the same time axis and weightedly combining them to calculate an ensemble future degradation trajectory for each target component; a step of generating an expected power generation efficiency curve, an expected output deviation curve, and an expected internal resistance increase curve for each target component from the ensemble future degradation trajectory, and calculating a Dynamic SoH Error Rate by comparing them with the normal operation profile generated based on the manufacturer's shipment specifications and the initial installation environment of the solar asset;A method for detecting anomalies and optimizing component replacement cycles of digital twin-based subscription-type solar assets, comprising: a step of correcting a performance curve corresponding to the degradation characteristics of each target component among the expected power generation efficiency curve, expected output deviation curve, and expected internal resistance increase curve for each target component based on the dynamic state error rate; and a step of deriving a temporal confidence interval of the remaining lifespan for each target component and a failure probability for each time interval by integrating the probability density function (PDF) at the point in time when the performance curve corresponding to the degradation characteristics of each target component does not satisfy the performance guarantee threshold condition corresponding to each target component; Claim 2 In claim 1, the step of calculating the ensemble future degradation trajectory comprises: generating a first degradation state vector including an expected power generation efficiency degradation rate, an expected series resistance increase rate, an expected cell damage area ratio, and an expected insulation performance degradation rate for each solar panel module from the results of the first acceleration simulation; generating a second degradation state vector including an expected junction temperature swing, an expected conduction loss increase rate, an expected switching loss increase rate for each power semiconductor, and an expected capacitance degradation rate, an expected equivalent series resistance increase rate, and an expected ripple tolerance degradation rate for each DC link capacitor from the results of the second simulation; rearranging the first degradation state vector and the second degradation state vector on the same time grid at preset sampling periods and performing interpolation operations on missing time points to generate a time-series degradation matrix for each target component; and calculating reliability parameters including an error variance, a residual autocorrelation coefficient, a prediction uncertainty, and a recent time point response sensitivity for each target component based on the difference between the time-series degradation matrix and the normal operation profile. Based on the above reliability parameters, a step of calculating a first weight for the results of a first acceleration simulation for the solar panel module and a second weight for the results of a second simulation for the power semiconductor and DC link capacitor, and normalizing the first weight and the second weight to generate a weight vector for each target component; a step of generating an ensemble degradation state vector for each target component by linearly combining each degradation component of a time-series degradation matrix using the weight vector, and generating a degradation transition covariance matrix based on the time-series variance and covariance of the ensemble degradation state vector; a step of inputting the ensemble degradation state vector and the degradation transition covariance matrix into an ensemble Kalman filter to sequentially correct the time-series degradation state, and performing Monte Carlo sampling on the corrected degradation state to generate a plurality of future degradation scenarios;A method for detecting anomalies and optimizing component replacement cycles of a digital twin-based subscription-type solar asset, comprising the step of deriving an ensemble future degradation trajectory for each target component, including an expected power generation efficiency degradation curve, an expected output deviation curve, an expected internal resistance increase curve, and an expected thermal fatigue accumulation curve for each of the solar panel module, power semiconductor, and DC link capacitor, by superimposing the above future degradation scenarios on a time axis; Claim 3 delete

Images