Method and system for pay-per-performance cleaning service of photovoltaic power station based on power generation guarantee

Abstract

The application discloses a photovoltaic power station pay-per-performance cleaning service method and system based on power generation capacity guarantee, relates to the field of component manufacturing technology, and comprises the following steps: simulating a non-fouling ideal power generation state, collecting fouling distribution data and environmental working condition data of a photovoltaic panel surface in real time, calculating ideal power generation capacity, theoretical power generation capacity under a current fouling state and predicted power generation capacity after cleaning, respectively, designing and generating an intelligent service contract, generating a cleaning operation scheme and performing directional cleaning, calculating actual power generation capacity gain, verifying the effect by using a confidence interval statistical method, and performing income calculation and expense settlement. In the application, the cleaning service charge is directly bound with the verifiable power generation capacity improvement effect, an objective and credible pay-per-performance system is constructed by using digital twinning, an intelligent contract and risk assessment technology, and the problems that the traditional cleaning service charge is decoupled from power generation benefit, the owner lacks motivation, and the technical advantages of the service provider cannot be quantitatively reflected are solved.

Description

Technical Field

[0001] This invention relates to the field of component manufacturing technology, and in particular to a method and system for pay-per-performance cleaning services for photovoltaic power plants based on power generation guarantees. Background Technology

[0002] The core of the pay-per-performance cleaning service for photovoltaic power plants is that the service provider charges fees based on the actual increase in power generation after cleaning, rather than charging by fixed number of times or area. This model deeply binds the interests of the service provider and the owner, achieving risk-sharing and goal alignment. The service provider can only receive corresponding compensation if it ensures that the cleaning effectively increases power generation. For the owner, it can directly reduce the cost risk of ineffective cleaning and transform operation and maintenance expenses into clear power generation revenue.

[0003] Currently, the traditional pay-per-use or annual subscription model is completely decoupled from the effect of power generation improvement, resulting in owners lacking the motivation to clean and making it difficult to highlight high-quality services; the evaluation of cleaning effect mainly relies on subjective experience or rough comparison, which cannot accurately quantify the impact of dust accumulation and cleaning gains, and lacks a reliable performance benchmark; due to the lack of quantitative models for risks such as weather and equipment failure, it is difficult to establish a fair risk-sharing mechanism, which restricts the performance-based payment business model.

[0004] Therefore, a method and system for providing pay-per-performance cleaning services for photovoltaic power plants based on power generation guarantees is proposed to solve the above problems. Summary of the Invention

[0005] The main objective of this invention is to provide a method and system for photovoltaic power plant cleaning services based on power generation guarantees and pay-per-performance, in order to solve the problems mentioned in the background above.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is: a photovoltaic power plant performance-based cleaning service method based on power generation guarantees, the method comprising the following steps: By using digital twin modeling technology for photovoltaic modules, a virtual power station model that is synchronized with the physical power station in real time is constructed to simulate an ideal power generation state without dust accumulation. Using a multispectral imaging sensor mounted on a drone and a local environmental monitoring device at the power station, real-time data on the surface dust distribution of photovoltaic panels and environmental conditions are collected. Based on the digital twin model and real-time ash accumulation data, the ideal power generation, the theoretical power generation under the current ash accumulation state, and the predicted power generation after cleaning are calculated using a photovoltaic power generation performance simulation algorithm. Design and generate smart service contracts with core terms that guarantee the increase in power generation and the reduction in losses, and clarify performance targets, verification cycles and revenue sharing rules; Based on the contract terms and dust distribution data, a cleaning operation plan is generated through path planning algorithms, and cleaning robots and drones are dispatched to perform targeted cleaning. During the cleanup and verification period, the actual power generation gain is calculated again using the digital twin model and real-time data, and the effect is verified using the confidence interval statistical method. Based on the verified actual power generation gain, and combined with the revenue sharing rules and risk assessment model in the contract, revenue calculation and fee settlement are automatically executed through smart contract technology.

[0007] Preferably, the construction of the digital twin model of the photovoltaic module includes the following steps: The layout, tilt angle, orientation, and specifications of photovoltaic power plants are collected using 3D laser scanning and photoelectric characteristic detection technology. Based on physical mechanisms and historical operating data, a simulation model of light and electricity conversion of photovoltaic modules under dust-free conditions was established, and a meteorological data interface was integrated. By utilizing a real-time data synchronization engine, the operating status parameters and environmental data of the actual power plant are continuously input into the model, enabling dynamic calibration and updating of the digital twin model.

[0008] Preferably, the collection of the ash distribution data and environmental condition data includes the following steps: A drone equipped with a multispectral imager was used to collect spectral reflectance data from the surface of a photovoltaic panel by flying along a preset route. Using image recognition and grayscale analysis algorithms, the ash cover density and distribution heat map are extracted from spectral data; Simultaneously collect irradiance, temperature, and wind speed data provided by the local meteorological station at the power station, and align them with the ash distribution data by timestamp to form a comprehensive monitoring dataset.

[0009] Preferably, the calculation of the power generation includes the following steps: Based on the digital twin model, environmental data under an ideal dust-free state is input, and a simulation is run to obtain the baseline ideal power generation. Real-time ash accumulation distribution data is converted into photoelectric conversion efficiency loss coefficients, which are then input into a twin model to calculate the theoretical power generation under the current ash accumulation state. Based on the ash cleaning prediction model, the efficiency recovery after cleaning is simulated, the predicted power generation after cleaning is calculated, and the expected power generation gain is derived accordingly.

[0010] Preferably, the design and generation of the smart service contract includes the following steps: Define key performance indicators in the contract, namely the percentage increase in power generation and the percentage decrease in power generation loss, and set the measurement period and tolerance threshold; Integrate risk assessment models to quantify weather uncertainty and equipment failure risk factors, and design corresponding risk-sharing and cost adjustment clauses; Performance indicators, verification methods, profit sharing ratios, and payment conditions are encoded into automatically executable smart contract scripts and deployed on blockchain and trusted platforms.

[0011] Preferably, the generation and execution of the cleaning operation includes the following steps: Based on the heat map of dust distribution, the optimal cleaning paths for cleaning robots and drones are planned using a greedy algorithm and a genetic algorithm. Based on environmental monitoring data, cleaning instructions are automatically sent to the execution equipment during rainless and low-wind-speed time windows; The system monitors the operational status and energy consumption of cleaning equipment in real time and provides feedback on task completion and abnormal events through an IoT platform.

[0012] Preferably, the cleaning verification includes the following steps: After the agreed verification period ends, the actual theoretical power generation is calculated again by inputting the actual environmental data of that period and the dust accumulation monitoring data after cleaning into the digital twin model. The actual theoretical power generation is compared with the theoretical power generation before the cleaning to calculate the actual power generation gain. The confidence interval of the actual gain is analyzed using statistical significance testing to determine whether the performance target threshold agreed upon in the contract has been reached.

[0013] Preferably, the revenue calculation and fee settlement includes the following steps: The service provider's share of revenue is calculated based on the verified actual power generation gain and the revenue sharing ratio agreed upon in the contract. The risk assessment model is invoked, and the profit share is dynamically adjusted based on the actual risk events that occur during the verification period. The settlement instructions are automatically triggered by smart contracts, and the adjusted fee data is sent to the payment system to complete automatic reconciliation and payment.

[0014] Preferably, the dynamic adjustment of the revenue share includes the following sub-steps: By using historical meteorological data and machine learning algorithms, a short-term prediction model for irradiance and precipitation probability is constructed, and the weather risk weight coefficient is calculated based on the probability distribution of prediction bias. Analyze the real-time operating status and historical fault data of power plant equipment, and use fault tree analysis to quantify the risk value of the impact of sudden equipment failures on power generation and cleaning operations. The weather risk weighting coefficient and equipment failure risk value are input into a preset risk and cost mapping function to dynamically generate risk-adjusted contract terms, including exemption clauses, profit ratio fluctuation ranges, and settlement delay rules.

[0015] A photovoltaic power plant performance-based cleaning service system, guaranteed by power generation, includes: Digital Twin Modeling Module: Used to build and maintain digital twin models of photovoltaic power plants; Dust accumulation and environmental monitoring module: used to collect dust distribution and environmental condition data on the surface of photovoltaic panels; Power generation simulation and gain calculation module: used to calculate ideal, current and predicted power generation; Smart contract management module: used for the generation, storage, and automatic execution of service contracts; Cleaning operation scheduling module: used to plan cleaning paths and control the cleaning equipment to perform the cleaning; Cleaning effect verification module: used to calculate and verify actual power generation gain; Revenue Settlement and Payment Module: Used to automatically settle fees based on contracts and verification results.

[0016] The present invention has the following beneficial effects: 1. In this invention, by constructing a digital twin model of photovoltaic modules and combining it with UAV multispectral imaging technology, the ideal power generation state without dust accumulation is simulated and the current dust accumulation loss is quantified. An objective and verifiable benchmark for power generation performance is established. This solves the problem of the decoupling between charging and power generation improvement caused by the inability to scientifically assess the impact of dust accumulation and cleaning effect in traditional cleaning services. It makes service fees directly related to actual gains, incentivizes high-quality services and activates owners' willingness to invest.

[0017] 2. In this invention, by designing a smart service contract with the ratio of increased power generation to reduced losses as the core performance indicator, and integrating a risk assessment model based on machine learning and fault tree analysis, uncertainties such as weather and equipment failures are quantified, achieving fair and dynamic risk sharing and cost adjustment. This solves the problem that the pay-per-performance model is difficult to implement and promote due to unclear risk definition and uneven risk sharing, and enhances the feasibility of the business model.

[0018] 3. In this invention, by encoding the performance verification logic, profit sharing rules and risk assessment results into an automatically executable smart contract, and automatically triggering settlement based on the calculation results of the digital twin model after the verification period, the entire process from effect verification to fee payment is automated and reliable. This solves the problems of traditional settlement methods that rely on manual labor, have cumbersome processes, low transparency and are prone to disputes, and greatly improves the efficiency and credibility of service settlement. Attached Figure Description

[0019] Figure 1 This is a framework diagram of the photovoltaic power plant performance-based cleaning service method based on power generation guarantee according to the present invention; Figure 2This is a framework diagram of the photovoltaic power plant performance-based pay cleaning service system based on power generation guarantee of the present invention. Figure 3 This is a flowchart of the photovoltaic power plant pay-per-performance cleaning service method based on power generation guarantee according to the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Please see Figures 1-3 This invention provides a technical solution: a method for paying-per-performance cleaning services for photovoltaic power plants based on power generation guarantees, the method comprising the following steps: By using digital twin modeling technology for photovoltaic modules, a virtual power station model that is synchronized with the physical power station in real time is constructed to simulate an ideal power generation state without dust accumulation. Using a multispectral imaging sensor mounted on a drone and a local environmental monitoring device at the power station, real-time data on the surface dust distribution of photovoltaic panels and environmental conditions are collected. Based on the digital twin model and real-time ash accumulation data, the ideal power generation, the theoretical power generation under the current ash accumulation state, and the predicted power generation after cleaning are calculated using a photovoltaic power generation performance simulation algorithm. Design and generate smart service contracts with core terms that guarantee the increase in power generation and the reduction in losses, and clarify performance targets, verification cycles and revenue sharing rules; Based on the contract terms and dust distribution data, a cleaning operation plan is generated through path planning algorithms, and cleaning robots and drones are dispatched to perform targeted cleaning. During the cleanup and verification period, the actual power generation gain was calculated again using the digital twin model and real-time data, and the effect was verified using the confidence interval statistical method. Based on the verified actual power generation gain, and combined with the revenue sharing rules and risk assessment model in the contract, revenue calculation and fee settlement are automatically executed through smart contract technology.

[0022] The construction of a digital twin model of a photovoltaic module includes the following steps: The layout, tilt angle, orientation, and specifications of photovoltaic power plant components are collected using 3D laser scanning and photoelectric characteristic detection technology, including the following steps: Using drones or ground-based mobile scanning equipment equipped with LiDAR, the entire area of ​​the photovoltaic power station is collected in three-dimensional point cloud data. Through point cloud registration and surface reconstruction algorithms, the spatial coordinates, installation tilt angle and orientation of each photovoltaic module are accurately extracted. By combining high-resolution aerial imagery, image recognition technology is used to initially identify and classify component models; Retrieve power plant design drawings, equipment lists and factory test reports, associate and input key electrical specifications of each component, including peak power, open circuit voltage, short circuit current and temperature coefficient under standard test conditions, to form a digital list of components that includes spatial and electrical attributes; Based on physical mechanisms and historical operating data, a simulation model of light and electricity conversion of photovoltaic modules under dust-free conditions was established, and a meteorological data interface was integrated, including the following steps: Based on the aforementioned digital list of components, and using the single-diode equivalent circuit model of a photovoltaic cell as the physical core, its output current I can be described by the formula: ; in, For photocurrent, This is the reverse saturation current of the diode. and These are series and parallel resistors, respectively. As an ideal factor, For thermal voltage, the model parameters are initialized using the component's factory data; The system integrates a meteorological data interface to receive or call real-time or forecast data such as total solar irradiance, ambient temperature, and wind speed provided by the power plant's on-site meteorological station, which are used as input driving variables for the model. By utilizing the actual power generation of the power plant under historical dust-free or clean conditions and corresponding meteorological data, particle swarm optimization or genetic algorithms are used to invert and calibrate the unknown parameters or system loss coefficients of the simulation model, so that the simulation output of the model is statistically highly consistent with the actual power generation data, thereby obtaining a high-fidelity benchmark performance model of the power plant. By utilizing a real-time data synchronization engine, the operating status parameters and environmental data of the actual power plant are continuously input into the model to achieve dynamic calibration and updating of the digital twin model, including the following steps: Through the IoT gateway of the power plant data acquisition and monitoring system, the operating status parameters such as DC side voltage, current, and power of each photovoltaic string or inverter are obtained in real time, as well as the latest minute-level environmental data from the weather station. The design incorporates a data assimilation mechanism that compares real-time operating power data with the predicted power of the established simulation model under the same environmental input. When the deviation continues to exceed a preset threshold, the model calibration process is triggered. By employing online parameter estimation techniques based on Kalman filtering or sliding window least squares, the performance degradation coefficient or specific environmental correction coefficient in the model is dynamically fine-tuned. This enables the output of the digital twin model to continuously track the performance changes of the physical power plant caused by slow-changing factors such as equipment aging and slight obstruction, ensuring its long-term accuracy and reliability as a benchmark for an ideal state without ash accumulation.

[0023] The collection of ash distribution data and environmental condition data includes the following steps: The process involves using a drone equipped with a multispectral imager to collect spectral reflectance data from the surface of a photovoltaic panel by flying along a preset route, including the following steps: Based on the component spatial distribution map derived from the digital twin model of the photovoltaic power station, the flight path planning software automatically generates a full-coverage parallel scanning flight path for the drone, ensuring that each photovoltaic panel is photographed and that the images of adjacent flight paths have sufficient overlap. The drone is equipped with a multispectral imager in the visible and near-infrared bands. It performs automatic flight missions during clear, cloudless days when the solar altitude angle is moderate. It flies at a constant altitude and speed to collect spectral reflectance images of the photovoltaic panel surface in multiple specific bands. The POS system is used to record the precise geographical location, pose angle, and shooting timestamp of each image, providing a spatial reference for subsequent image registration and data analysis; Using image recognition and grayscale analysis algorithms, the ash cover density and distribution heatmap are extracted from spectral data, including the following steps: The acquired multispectral images are preprocessed, including radiometric calibration to convert the original gray values ​​into surface reflectance, and orthorectification based on POS data and digital surface model to eliminate perspective distortion and topographic relief effects, generating reflectance orthophotos with accurate geographic coordinates. Image segmentation algorithms (such as threshold-based segmentation or edge detection) are applied to accurately segment each independent photovoltaic module region from the orthophoto, eliminating background interference. For each module region, its average reflectance in different bands is calculated. Based on the physical characteristic that clean photovoltaic panels and dust accumulation exhibit significant differences in reflectivity at specific wavelengths (such as the visible blue light band and the near-infrared band), a dust index (DI) model is established, which is calculated using, for example, the following formula: ; in and The reflectance was measured in the blue light and near-infrared bands, respectively. Then, the pre-established DI value was compared with the dust cover density (e.g., g / m³). 2The calibration relationship between the components (which can be obtained through laboratory calibration or historical data fitting) is used to convert the DI value of each component into physical cover density and generate a heat map of ash density level distribution across the entire power plant. Simultaneously collect irradiance, temperature, and wind speed data from the local meteorological station at the power plant, and align them with ash distribution data by timestamp to form a comprehensive monitoring dataset, including the following steps: Through the data interface of the power plant monitoring system or the Internet of Things gateway, minute-level or second-level data recorded by the irradiance meter, temperature sensor and anemometer installed at the power plant site are collected in real time, including total horizontal irradiance, component backplane temperature and ambient wind speed. Establish a unified time synchronization protocol to mark each meteorological data record and each frame of UAV image data with a UTC timestamp accurate to the second; In the data processing server, using timestamps as key indexes, the meteorological data sequence within the synchronization period (usually a few hours before and after the drone flight) is associated, aligned, and packaged with the processed ash distribution heat map data with the collection time point, forming a comprehensive monitoring data package containing spatial dimensions (ash distribution), temporal dimensions (collection time), and environmental dimensions (meteorological conditions), providing accurate and synchronous input for subsequent power generation simulation calculations.

[0024] The calculation of power generation includes the following steps: Based on a digital twin model, environmental data under an ideal dust-free state is input, and a simulation is run to obtain a baseline ideal power generation, including the following steps: Set the target time period for calculation (such as the comparison period before cleaning or the historical cleaning period), and obtain the actual, quality-verified solar irradiance, ambient temperature and wind speed sequence data within that time period from the meteorological database integrated in the digital twin model, as the ideal dust-free input representing the real environmental conditions of that time period. The time-series environmental data stream is input into the calibrated digital twin model, which is based on its inherent photoelectric physics equations (such as the single diode equivalent circuit model mentioned above) and sets the dust accumulation efficiency loss coefficient to 0. The model performs hourly simulation calculations in minutes or hours and outputs the theoretical power of each photovoltaic string without dust accumulation at each moment. Integrating (summing) the simulated power at all times within this time period yields the total baseline ideal power generation of the power station under specific meteorological conditions in a completely dust-free ideal state. ; The real-time ash accumulation distribution data is converted into photoelectric conversion efficiency loss coefficients, input into a twin model, and the theoretical power generation under the current ash accumulation state is calculated, including the following steps: The system retrieves the dust distribution heatmap generated from the most recent drone inspection, and for each individual photovoltaic module unit, reads its dust coverage density value. Then, it uses a pre-defined mapping function between dust density and efficiency loss. Physical ash density Converted to the relative loss factor of the current photoelectric conversion efficiency of the component (Typically a value between 0 and 1, where 1 represents complete cleanliness); The calculated efficiency loss coefficients of all components As a new performance correction parameter, it is input into the corresponding component instance of the digital twin model and drives the model simulation together with the same real-time or historical environmental data time series. During computation, the model takes into account the efficiency degradation of each component due to dust accumulation (e.g., when calculating its photocurrent). Multiply by a factor This outputs the theoretical power timing under the current ash accumulation state, and after integration, the theoretical power generation of the power plant under the current ash accumulation state is obtained. ; The spatial variability allows the model to accurately reflect the comprehensive impact of uneven ash distribution on the overall power generation of the power plant. Based on the ash removal prediction model, the efficiency recovery after cleaning is simulated, the predicted power generation after cleaning is calculated, and the expected power generation gain is derived accordingly, including the following steps: The core of constructing a dust accumulation cleaning prediction model is to define an efficiency recovery function after cleaning. The function uses the dust density before component cleaning. Using the cleaning technology employed (e.g., dry brushing, water washing) as input, predict the efficiency recovery coefficient of the component after cleaning (e.g., recovery to...). (i.e., 98% initial efficiency). The predicted efficiency coefficients of each component after cleaning Input the digital twin model and replace the original one. Then, input the predicted meteorological data or typical meteorological year data for the target future time period (as in the next verification period), run the model simulation, and obtain the theoretical power generation after the predicted cleanup. ;.Predict the amount of electricity generated after cleaning Theoretical power generation under current ash accumulation conditions Compare and calculate the absolute gain: ; Calculate the relative gain ratio: ; this The key technical basis for service contract negotiation and expected revenue assessment is the expected power generation gain.

[0025] The design and generation of smart service contracts includes the following steps: Define key performance indicators (KPIs) in the contract, namely the percentage increase in power generation and the percentage decrease in power generation losses, and set measurement periods and tolerance thresholds, including the following steps: The theoretical power generation under the current ash accumulation state, derived from the aforementioned power generation calculation steps. Compared with the predicted power generation after cleaning Negotiate and determine the core KPIs of the contract, such as defining the percentage increase in power generation. or the proportion of power generation loss reduction ,in To verify the actual cleaning and power generation, This represents the ideal power generation for the same period. Define the measurement period for KPIs, such as setting it as a 30 consecutive calendar days after the completion of the cleaning operation as the effect verification period, and stipulate that the power generation data used to calculate KPIs during this period must be based on a digital twin model and simulated using actual meteorological data input. Set performance tolerance thresholds, for example, agreeing on the actual percentage increase. Below the contract target value If the performance reaches 90% (the tolerance threshold), the fee adjustment clause will be triggered. This threshold is used to define whether the performance target has been met and to initiate the corresponding business rules. Integrate risk assessment models, quantify weather uncertainty and equipment failure risk factors, and design corresponding risk-sharing and cost adjustment clauses, including the following steps: To address weather uncertainties, a short-term irradiance prediction bias model is constructed. The inputs are weather forecast data and historical error distributions, and the output is the risk weight coefficient. This is used to adjust the determination of responsibility for fluctuations in power generation caused by abnormal weather (such as continuous rain), for example when When a certain threshold is exceeded, the corresponding time period will be removed from the KPI calculation or discounted proportionally. To address equipment failure risks, an equipment health status assessment model based on real-time monitoring and historical failure rates is established to quantify the potential impact of sudden failures on power generation. And design corresponding risk-sharing clauses, such as stipulating that power generation losses caused by equipment failures not caused by cleaning operations will not be included in the cleaning effect assessment; Based on quantization and The final service fee is dynamically calculated using a predefined fee adjustment function. ; This function encapsulates the risk-sharing rules agreed upon by both parties, such as setting deductibles and a floating range for profit sharing ratios, to ensure that the cost calculation fairly reflects the controllable cleaning effect; Encode performance indicators, verification methods, profit-sharing ratios, and payment terms into automatically executable smart contract scripts and deploy them on a blockchain and trusted platform, including the following steps: The KPI calculation formula, measurement period, tolerance threshold, risk quantification model and cost adjustment function, as well as the profit sharing ratio and payment trigger conditions (such as automatic payment after KPI is met) confirmed by both parties to the contract are all transformed into formal, unambiguous logical statements and mathematical expressions. Using the Solidity smart contract programming language, the above logic is encoded into a contract script that can run automatically in the blockchain virtual machine. Key logic includes: automatically calling the authorized digital twin model interface to obtain [data / equipment] after the verification period ends. Wait for data, calculate Query the risk assessment model output and finally execute. The function calculates the amount due; The completed smart contract script is deployed on a selected blockchain platform (such as an enterprise-level consortium blockchain), and the contract address is published, all parties' digital signatures are authenticated, and the interface with the payment gateway is connected to ensure that the contract status is publicly verifiable and the execution process is transparent and tamper-proof, thereby achieving full automation and trustworthiness from effect verification to fee settlement.

[0026] The generation and execution of a cleaning operation includes the following steps: Based on the heat map of dust distribution, the optimal cleaning paths for the cleaning robot and drone are planned using a greedy algorithm and a genetic algorithm, including the following steps: The heat map of dust accumulation distribution is converted into a weighted task point map. Each vertex in the map represents a photovoltaic module or a group of photovoltaic modules, and its weight is the dust accumulation density value of that area. The edges between vertices represent the path that the device can move, and their weights are the moving distance or time. A two-layer planning strategy is adopted. The first layer uses the local optimum characteristic of the greedy algorithm to quickly generate the initial path and prioritizes cleaning the area with the most severe dust accumulation (highest weight). Using the initial path as an individual in the population, the genetic algorithm is used for global optimization. The fitness function is defined as maximizing the total amount of dust cleaned (the sum of the weights of each task point) within a limited total operation time or total energy consumption, while minimizing the invalid movement distance. Through selection, crossover, mutation and other operations, the algorithm iteratively evolves and finally outputs one or a set of optimal operation path sequences that take into account both cleaning efficiency and coverage effect. Based on environmental monitoring data, during rainless and low-wind-speed time windows, cleaning instructions are automatically issued to the execution equipment, including the following steps: Establish an environmental adaptability assessment model and access real-time weather forecast data for the next 6-24 hours provided by meteorological stations, including precipitation probability, wind speed and direction, and irradiance predictions; Define the safe and efficient execution window conditions for cleaning operations, such as: the probability of precipitation is less than 10% in the next 3 hours, the average wind speed is less than 5 m / s, and the daytime irradiance is high to ensure that the component surface is dry and conducive to cleaning. When the system detects that the above window conditions are met in the future, it automatically packages the generated optimal path and operation parameters (such as the brush head pressure of the cleaning robot and the water flow rate of the drone) into a standard operation instruction and sends it to the designated ground cleaning robot base station and drone control station through the wireless communication network (such as 4G / 5G or local area network), triggering the equipment to start automatically and execute the cleaning task according to the plan. Real-time monitoring of the cleaning equipment's operational status and energy consumption, and feedback on task completion and abnormal events via an IoT platform, including the following steps: Cleaning robots and drones are equipped with IoT sensors to collect and upload their own status data in real time, including GPS / BeiDou positioning coordinates, travel speed, battery level, current working mode, brush head speed, water pump pressure, and real-time energy consumption. The central monitoring platform receives these IoT data streams and compares them in real time with the predetermined work paths and plans. It uses electronic fences and status thresholds to determine whether the equipment is operating according to the plan, whether it has deviated from the path, has insufficient power, has abnormal components (such as a stuck brush head), or has encountered sudden obstacles. Once an abnormal event is detected or a task is completed, the platform immediately generates a structured status report and alarm information, which is fed back to the operation and maintenance personnel through a visual interface, SMS or application push. The platform also automatically records the actual path, total time and total energy consumption of the task, which serve as input data for task cost accounting and subsequent path optimization algorithm training, thus forming a closed-loop management of task execution.

[0027] Cleaning verification includes the following steps: After the agreed verification period ends, the actual theoretical power generation is calculated again using the digital twin model, by inputting the actual environmental data of that period and the monitoring data of dust accumulation after cleaning. This includes the following steps: Acquire and lock the verification period stipulated in the contract (such as 30 days after cleaning), and extract actual, high-precision time-series environmental data within this period from the meteorological database, including total solar irradiance, ambient temperature and wind speed; Immediately after the cleaning operation is completed and before the verification cycle begins, a multispectral inspection using a drone is arranged to obtain the baseline dust distribution after cleaning, and this data is input into the model as the initial efficiency coefficient after cleaning. ; The time series of actual environmental data during the verification period and based on The component efficiency status is input into a dynamically calibrated digital twin model for simulation. The model calculates the theoretical power generation under the actual ash accumulation state after cleaning and the actual weather conditions within the cycle hourly. Finally, the actual theoretical power generation for the entire verification cycle is obtained through time integration. ; The actual theoretical power generation is compared with the theoretical power generation during the same period before the cleaning, and the actual power generation gain is calculated, including the following steps: Select a comparable pre-cleaning period, typically the same number of consecutive days before cleaning, when dust accumulation is stable and weather conditions are similar. Then, retrieve the actual environmental data and corresponding dust distribution data for that period (i.e., pre-cleaning data). The theoretical power generation before cleaning was calculated by inputting the digital twin model. ; Calculate absolute power generation gain: ; Calculate the actual power generation increase rate of the core performance indicator: ; The calculated Compared with the power generation increase target defined in the contract This correlation serves as a direct quantitative basis for measuring the actual effectiveness of cleaning services; The statistical significance test method is used to analyze the confidence interval of the actual gain and determine whether the performance target threshold agreed in the contract has been reached. This includes the following steps: Considering the uncertainties in the input of the simulation model (such as meteorological data) and its inherent nature, statistical analysis was performed on the daily power generation data during the verification and comparison periods to calculate the sample mean and standard deviation of the daily power generation gain. Calculate the actual power generation gain ratio based on the t-distribution theory. Confidence interval at a certain confidence level (e.g., 95%) This range reflects the reliable range of the gain estimate after taking into account random fluctuations; Lower bound of the confidence interval Performance target thresholds agreed in the contract Comparison: If If the cleaning effect meets the standard, then a high statistical confidence level is used to determine that the cleaning effect has met the standard; if If the result is not satisfactory, the performance may not meet the target. This statistical judgment result serves as an objective and scientific basis for triggering different fee settlement branches in smart contracts, effectively avoiding performance misjudgments caused by short-term accidental weather fluctuations.

[0028] The revenue calculation and expense settlement process includes the following steps: Based on the verified actual power generation gain and the revenue sharing ratio agreed upon in the contract, the service provider's share of the revenue is calculated, including the following steps: Obtain the actual power generation gain results confirmed by statistical testing from the cleanup verification module, including the actual theoretical power generation during the verification period. Theoretical power generation during the comparison period The absolute gain was calculated. and increase ratio ; Read the pre-agreed revenue-sharing rules between the two parties from the deployed smart contract, such as the basic revenue-sharing ratio. (e.g., the service provider shares 60% of the gain) and possible tiered incentive coefficients; The service provider's basic revenue share is calculated based on the contract terms. The calculation formula is: ,in The price per unit of electricity stipulated in the contract (such as the local grid connection price or negotiated price), This refers to the initial expected return before risk adjustment. The risk assessment model is invoked, and the profit share is dynamically adjusted based on the actual risk events that occur during the verification period. This includes the following steps: By using historical meteorological data and machine learning algorithms, a short-term prediction model for irradiance and precipitation probability is constructed, and weather risk weighting coefficients are calculated based on the probability distribution of prediction bias. The process includes the following steps: Collect historical meteorological data of the power station location over many years, including time series of measured irradiance, precipitation, cloud cover, and temperature, as a training set; use machine learning algorithms such as long short-term memory networks or random forests to establish a time series prediction model of irradiance and precipitation probability for the next 24-72 hours. The input of this model is the current and historical meteorological characteristics, and the output is a sequence of predicted values. During the validation period, the predicted data output by the model is compared with the actual observation data from the weather station, and the daily prediction deviation is calculated, such as the relative error of irradiance prediction. And statistically analyze the probability distribution of these deviations over a certain period of time (e.g., fit it to a normal distribution). Based on the deviation probability distribution, combined with the abnormal weather thresholds defined in the contract (such as daily irradiance below 50% of the annual average and prediction error), ), calculate the weather risk weighting coefficient The calculation formula is as follows: ; This coefficient quantifies the proportion of power generation losses that cannot be attributed to any single party due to significant deviations in weather forecasts, and serves as the basis for cost adjustments. Analyze the real-time operating status and historical fault data of power plant equipment, and use fault tree analysis to quantify the risk value of the impact of sudden equipment failures on power generation and cleaning operations, including the following steps: Establish a fault tree with unexpected power generation loss as the top event, and decompose it layer by layer to basic events such as inverter shutdown, combiner box failure, and module hot spot, and statistically analyze the occurrence probability of each basic event based on historical operation and maintenance records. During the verification period, alarm data and equipment status telemetry data (such as inverter efficiency and string current dispersion) from the power plant monitoring system are accessed in real time to identify whether basic equipment fault events defined in the fault tree that are unrelated to the cleaning operation have occurred, and to record their duration. ; Quantify the risk value of the impact of this failure event on power generation. The calculation method is as follows:

[0029] in The probability of the i-th type of fault occurring (from historical statistics). For the affected installed capacity, the formula estimates the proportion of expected power generation loss caused by non-purge factors. This risk value will be used to deduct the corresponding liability in the revenue calculation. By inputting the weather risk weighting coefficient and equipment failure risk value into a preset risk-cost mapping function, risk-adjusted contract terms, including deductible clauses, profit ratio fluctuation ranges, and settlement delay rules, are dynamically generated, including the following steps: Define the risk-cost mapping function The function calculates the risk coefficient and the underlying return. As input, its internal logic encapsulates all risk-sharing rules agreed upon by both parties; The function dynamically determines the final terms based on the risk value: for example, if If a certain threshold is exceeded, the weather deductible clause will be triggered, and the gains for that period will not be included in the revenue calculation; if If significant, the base revenue share will be reduced proportionally. Meanwhile, the function can output a judgment on whether the settlement is delayed (e.g., the verification period needs to be extended in case of major equipment failure). The function outputs the adjusted profit. The specific terms and conditions that take effect will serve as the final logic executed by the smart contract during automatic settlement. Its mathematical expression or decision tree rules are pre-deployed as part of the contract code, thereby achieving automatic and fair binding of returns and dynamic risk assessment results. The settlement instruction is automatically triggered by a smart contract, and the adjusted fee data is sent to the payment system to complete automatic reconciliation and payment, including the following steps: The smart contract script deployed on the blockchain automatically executes the settlement logic: first, it verifies whether the above calculation process is complete and whether the data source is trustworthy (e.g., through pre-set digital signature verification), and then reads the finally determined share of revenue. ; Smart contracts according to Generate a payment instruction containing the payment amount, the payee's (service provider's) account address, and a unique settlement ID. After digitally signing the instruction using the contract's own private key, it is automatically submitted to the connected bank payment gateway or digital currency payment system through the application programming interface. After receiving the instruction, the payment system performs compliance verification, then executes the fund transfer, and sends the payment success receipt and transaction hash back to the smart contract and the management platforms of both parties. The smart contract status is updated to "settled". The entire reconciliation and payment process does not require manual intervention, achieving a fully automated and auditable closed loop.

[0030] A photovoltaic power plant performance-based cleaning service system, guaranteed by power generation, includes: Digital Twin Modeling Module: Used to build and maintain digital twin models of photovoltaic power plants; Dust accumulation and environmental monitoring module: used to collect dust distribution and environmental condition data on the surface of photovoltaic panels; Power generation simulation and gain calculation module: used to calculate ideal, current and predicted power generation; Smart contract management module: used for the generation, storage, and automatic execution of service contracts; Cleaning operation scheduling module: used to plan cleaning paths and control the cleaning equipment to perform the cleaning; Cleaning effect verification module: used to calculate and verify the actual power generation gain; Revenue Settlement and Payment Module: Used to automatically settle fees based on contracts and verification results.

[0031] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus.

[0032] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for paying-per-performance cleaning services for photovoltaic power plants based on power generation guarantees, characterized in that: The method includes the following steps: By using digital twin modeling technology for photovoltaic modules, a virtual power station model that is synchronized with the physical power station in real time is constructed to simulate an ideal power generation state without dust accumulation. Using a multispectral imaging sensor mounted on a drone and a local environmental monitoring device at the power station, real-time data on the surface dust distribution of photovoltaic panels and environmental conditions are collected. Based on the digital twin model and real-time ash accumulation data, the ideal power generation, the theoretical power generation under the current ash accumulation state, and the predicted power generation after cleaning are calculated using a photovoltaic power generation performance simulation algorithm. Design and generate smart service contracts with core terms that guarantee the increase in power generation and the reduction in losses, and clarify performance targets, verification cycles and revenue sharing rules; Based on the contract terms and dust distribution data, a cleaning operation plan is generated through path planning algorithms, and cleaning robots and drones are dispatched to perform targeted cleaning. During the cleanup and verification period, the actual power generation gain is calculated again using the digital twin model and real-time data, and the effect is verified using the confidence interval statistical method. Based on the verified actual power generation gain, and combined with the revenue sharing rules and risk assessment model in the contract, revenue calculation and fee settlement are automatically executed through smart contract technology.

2. The photovoltaic power plant performance-based cleaning service method according to claim 1, characterized in that: The construction of the digital twin model of the photovoltaic module includes the following steps: The layout, tilt angle, orientation, and specifications of photovoltaic power plants are collected using 3D laser scanning and photoelectric characteristic detection technology. Based on physical mechanisms and historical operating data, a simulation model of light and electricity conversion of photovoltaic modules under dust-free conditions was established, and a meteorological data interface was integrated. By utilizing a real-time data synchronization engine, the operating status parameters and environmental data of the actual power plant are continuously input into the model, enabling dynamic calibration and updating of the digital twin model.

3. The photovoltaic power plant performance-based cleaning service method according to claim 1, characterized in that: The collection of ash distribution data and environmental condition data includes the following steps: A drone equipped with a multispectral imager was used to collect spectral reflectance data from the surface of a photovoltaic panel by flying along a preset route. Using image recognition and grayscale analysis algorithms, the ash cover density and distribution heat map are extracted from spectral data; Simultaneously collect irradiance, temperature, and wind speed data provided by the local meteorological station at the power station, and align them with the ash distribution data by timestamp to form a comprehensive monitoring dataset.

4. The photovoltaic power plant performance-based cleaning service method according to claim 1, characterized in that: The calculation of the power generation includes the following steps: Based on the digital twin model, environmental data under an ideal dust-free state is input, and a simulation is run to obtain the baseline ideal power generation. Real-time ash accumulation distribution data is converted into photoelectric conversion efficiency loss coefficients, which are then input into a twin model to calculate the theoretical power generation under the current ash accumulation state. Based on the ash cleaning prediction model, the efficiency recovery after cleaning is simulated, the predicted power generation after cleaning is calculated, and the expected power generation gain is derived accordingly.

5. The photovoltaic power plant performance-based cleaning service method according to claim 1, characterized in that: The design and generation of the smart service contract includes the following steps: Define key performance indicators in the contract, namely the percentage increase in power generation and the percentage decrease in power generation loss, and set the measurement period and tolerance threshold; Integrate risk assessment models to quantify weather uncertainty and equipment failure risk factors, and design corresponding risk-sharing and cost adjustment clauses; Performance indicators, verification methods, profit sharing ratios, and payment conditions are encoded into automatically executable smart contract scripts and deployed on blockchain and trusted platforms.

6. The photovoltaic power plant performance-based cleaning service method according to claim 1, characterized in that: The generation and execution of the cleaning operation includes the following steps: Based on the heat map of dust distribution, the optimal cleaning paths for cleaning robots and drones are planned using a greedy algorithm and a genetic algorithm. Based on environmental monitoring data, cleaning instructions are automatically sent to the execution equipment during rainless and low-wind-speed time windows; The system monitors the operational status and energy consumption of cleaning equipment in real time and provides feedback on task completion and abnormal events through an IoT platform.

7. The photovoltaic power plant performance-based cleaning service method according to claim 1, characterized in that: The cleaning verification includes the following steps: After the agreed verification period ends, the actual theoretical power generation is calculated again by inputting the actual environmental data of that period and the dust accumulation monitoring data after cleaning into the digital twin model. The actual theoretical power generation is compared with the theoretical power generation before the cleaning to calculate the actual power generation gain. The confidence interval of the actual gain is analyzed using statistical significance testing to determine whether the performance target threshold agreed upon in the contract has been reached.

8. The photovoltaic power plant performance-based cleaning service method according to claim 1, characterized in that: The calculation of revenue and settlement of fees includes the following steps: The service provider's share of revenue is calculated based on the verified actual power generation gain and the revenue sharing ratio agreed upon in the contract. The risk assessment model is invoked, and the profit share is dynamically adjusted based on the actual risk events that occur during the verification period. The settlement instructions are automatically triggered by smart contracts, and the adjusted fee data is sent to the payment system to complete automatic reconciliation and payment.

9. The photovoltaic power plant performance-based cleaning service method according to claim 8, characterized in that: The dynamic adjustment of the revenue share includes the following sub-steps: By using historical meteorological data and machine learning algorithms, a short-term prediction model for irradiance and precipitation probability is constructed, and the weather risk weight coefficient is calculated based on the probability distribution of prediction bias. Analyze the real-time operating status and historical fault data of power plant equipment, and use fault tree analysis to quantify the risk value of the impact of sudden equipment failures on power generation and cleaning operations. The weather risk weighting coefficient and equipment failure risk value are input into a preset risk and cost mapping function to dynamically generate risk-adjusted contract terms, including exemption clauses, profit ratio fluctuation ranges, and settlement delay rules.

10. A photovoltaic power plant performance-based cleaning service system with guaranteed power generation, comprising the photovoltaic power plant performance-based cleaning service method with guaranteed power generation as described in any one of claims 1-9, characterized in that, include: Digital Twin Modeling Module: Used to build and maintain digital twin models of photovoltaic power plants; Dust accumulation and environmental monitoring module: used to collect dust distribution and environmental condition data on the surface of photovoltaic panels; Power generation simulation and gain calculation module: used to calculate ideal, current and predicted power generation; Smart contract management module: used for the generation, storage, and automatic execution of service contracts; Cleaning operation scheduling module: used to plan cleaning paths and control the cleaning equipment to perform the cleaning; Cleaning effect verification module: used to calculate and verify actual power generation gain; Revenue Settlement and Payment Module: Used to automatically settle fees based on contracts and verification results.

Images