A method for configuring energy storage in a self-organizing network distribution radio station area

Abstract

This application belongs to the field of power systems and specifically discloses a method for configuring energy storage in self-organizing grid distribution substations. A closed-loop decision-making process improves the accuracy of energy storage configuration. Specifically, this application uses an improved generative adversarial network to generate basic scenarios, accurately modeling the complex time-series characteristics of load and photovoltaic output, providing a reliable basis for evaluation. Secondly, fuzzy logic is introduced to correct for climate and user behavior factors, generating a target scenario set that covers extreme and special operating conditions. Furthermore, a simulation model is run based on the target scenario set, accurately quantifying the expected unsupplied power under different energy storage capacities, transforming the assessment of power supply capability from qualitative to quantitative, solving the core problem of lacking quantitative support. Finally, this quantitative indicator is selected as the core constraint, and the solution is performed with the goal of minimizing the entire life cycle cost, ensuring that the scheme achieves economic optimization while meeting power supply reliability requirements, thereby solving the problems of unscientific configuration results and insufficient reliability.

Description

Technical Field

[0001] This application belongs to the field of power systems and relates to distribution substation planning and configuration technology; more specifically, it relates to a method for configuring energy storage in self-organizing network distribution substations. Background Technology

[0002] In the construction and renovation of distribution substations, the scientific configuration of energy storage capacity is a key aspect of improving the reliability of power supply and meeting the needs of extreme and special operating conditions. Currently, most methods for configuring energy storage capacity in distribution substations are based on single probability assumptions such as normal distribution to model load and photovoltaic output. This makes it difficult to accurately capture the strong temporal correlation, multi-peak distribution, and abrupt changes in load and photovoltaic output in actual power systems. Furthermore, it fails to effectively quantify the nonlinear impacts of climatic factors such as humid weather and holidays, as well as user behavior, leading to significant biases in the assessment of energy storage capacity under extreme operating conditions.

[0003] Meanwhile, existing configuration methods lack integrated design for scenario generation, operation simulation, and capacity optimization. They often make capacity decisions directly after analyzing a single aspect, failing to form a closed loop from data-driven to engineering decision-making. Furthermore, they do not use quantitative formulas to accurately express operating conditions, operational processes, and optimization objectives. This results in insufficient scientific rigor and reliability of configuration schemes in complex and uncertain power operation environments. Such schemes are prone to problems such as over-configuring energy storage capacity, leading to resource waste, or under-configuring capacity, failing to meet the supply requirements under extreme operating conditions. They are also difficult to adapt to the high requirements for power supply continuity in distribution substations. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the purpose of this application is to provide a method for configuring energy storage in self-organizing network distribution substations, aiming to solve the problem that current configuration methods lack a precise quantitative scheme for ensuring the supply support capacity of different energy storage configurations under different operating conditions, resulting in inaccurate energy storage configuration results.

[0005] The first aspect of this application relates to a method for configuring energy storage in a self-organizing network distribution substation. In one embodiment, the method includes: Step S10, generating a basic scenario set based on a Conditional Time GAN model with gradient penalty, according to historical load data and historical photovoltaic output data of the distribution substation; the basic scenario set is a set of photovoltaic output and load values ​​that reflect historical characteristics; Step S20, performing nonlinear correction on the basic scenario set for climate factors and user behavior factors using fuzzy logic to generate a target scenario set; the target scenario set is a set of photovoltaic output and load values ​​covering various operating conditions caused by climate factors and user behavior factors; Step S30, generating the expected unsupplied electricity amount corresponding to different energy storage configuration capacities as a supply guarantee index based on the energy storage operation simulation model of the distribution substation and the target scenario set; Step S40, aiming to minimize the total life cycle cost of energy storage, combining the energy storage operation simulation model and the supply guarantee index constraints, solving for the optimal energy storage configuration capacity through an optimization algorithm to realize the energy storage configuration of the distribution substation.

[0006] In one implementation, the Conditional Time GAN model with gradient penalty includes a generator and a discriminator, and during its training, a gradient penalty term based on random interpolated samples is introduced for the discriminator to address the mode collapse problem, thereby enabling the model to generate a base scene set with strong temporal correlation, multimodal distribution and abrupt change characteristics.

[0007] In one embodiment, step S20 includes: establishing membership functions for various operating conditions based on climate factors and user behavior factors; defuzzifying the membership functions based on fuzzy inference rules to obtain correction coefficients for load and photovoltaic output; and using the correction coefficients to correct the load scenario and photovoltaic output scenario in the basic scenario set to generate the final target scenario set for simulation.

[0008] In one embodiment, before step S30, the method further includes: constructing a power balance relationship of the distribution substation including the distribution network, photovoltaic, energy storage system and load; establishing charging and discharging power constraints, state of charge update constraints and state of charge boundary constraints of the energy storage system; and determining the energy storage operation simulation model of the distribution substation based on the power balance relationship of the distribution substation, charging and discharging power constraints, state of charge update constraints and state of charge boundary constraints.

[0009] In one embodiment, step S30 includes: for a given energy storage configuration capacity configuration, based on the energy storage operation simulation model of the distribution substation, the target scenario set, and the limit value of the grid-side power supply, obtaining the unsupplied power at each simulation time under different scenarios; and based on the occurrence probability of each scenario, performing a weighted summation of the unsupplied power to obtain the expected unsupplied amount corresponding to the energy storage configuration capacity configuration.

[0010] In one embodiment, minimizing the total lifecycle cost of energy storage specifically involves: constructing a single-objective or multi-objective optimization function with the energy storage configuration capacity as the decision variable, wherein the objective optimization function includes at least minimizing the total lifecycle cost of energy storage, and the total lifecycle cost includes at least investment cost and operation and maintenance cost; combining the energy storage operation simulation model and supply guarantee index constraints specifically involves: constructing constraints for optimization solution, including: supply guarantee index constraints based on the expected unsupplied power corresponding to the energy storage configuration capacity not exceeding the target threshold, and constraints from the energy storage operation simulation model.

[0011] In one embodiment, the optimal energy storage configuration capacity is obtained by using an optimization algorithm, including: using a particle swarm optimization algorithm to solve the objective function under constraints to obtain the optimal energy storage configuration capacity that meets the target supply level.

[0012] In a second aspect, this application provides an energy storage configuration device, comprising: at least one memory for storing a program; and at least one processor for executing the program stored in the memory, wherein when the program stored in the memory is executed, the processor is configured to execute the method described in the first aspect or any possible implementation thereof.

[0013] Thirdly, this application provides a computer-readable storage medium storing a computer program that, when run on a processor, causes the processor to perform the method described in the first aspect or any possible implementation thereof.

[0014] Fourthly, this application provides a computer program product that, when run on a processor, causes the processor to perform the method described in the first aspect or any possible implementation thereof.

[0015] It is understood that the beneficial effects of the second to fifth aspects mentioned above can be found in the relevant descriptions in the first aspect mentioned above, and will not be repeated here.

[0016] Overall, the technical solutions conceived in this application have the following beneficial effects compared with the prior art: The technical solution proposed in this application effectively solves the problem of inaccurate energy storage configuration results in existing technologies by adopting a set of interconnected, closed-loop decision-making techniques, and achieves a comprehensive technical effect of improving the scientificity, reliability, and economy of the configuration scheme. Specifically, as follows: First, by adopting a technique based on an improved Conditional Time GAN model to generate a basic scenario set, the model can capture the strong temporal correlation and abrupt change characteristics of load and photovoltaic output. This solves the problem that traditional methods rely on a single probability assumption and cannot accurately model complex temporal changes in reality, thus providing a scenario basis that is more in line with the actual data patterns for subsequent evaluation.

[0017] Next, by further adopting the technical means of nonlinearly correcting climate and user behavior factors based on fuzzy logic to generate a target scenario set, the aforementioned basic scenarios can cover special working conditions such as the return of spring and holidays, solving the problem that existing methods are difficult to quantify external nonlinear effects, thus generating a set of simulation scenarios that can comprehensively reflect extreme and normal working conditions.

[0018] Furthermore, by employing a technique based on a simulation model running on a target scenario set to generate the expected amount of unsupplied power corresponding to different energy storage capacities, the assessment of energy storage supply support capability has been transformed from a qualitative or rough estimate to a precise quantification based on full-scenario simulation. This solves the core problem of the current method's lack of a precise quantification scheme for supply support capability, thus providing a direct quantitative indicator basis for capacity decision-making.

[0019] Ultimately, by employing a technical approach that aims to minimize the total lifecycle cost and uses the aforementioned quantitative supply guarantee indicators as the core constraint for optimization, the optimal energy storage capacity scheme is ensured to achieve the best balance between investment and operational economy while strictly meeting the predetermined power supply reliability requirements. This comprehensively solves the problems of unscientific configuration results that may be too large or too small, and achieves the technical effect of improving the economy and reliability of energy storage configuration schemes. Attached Figure Description

[0020] Figure 1 This is a flowchart illustrating the self-organizing network distribution station area energy storage configuration method provided in the embodiments of this application; Figure 2 This is a schematic diagram illustrating the generation of the target scene set provided in the embodiments of this application; Figure 3 This is a flowchart illustrating the optimization algorithm provided in an embodiment of this application; Figure 4 This is an overall flowchart of the closed-loop decision-making framework for the self-organizing network distribution station area energy storage configuration method provided in the embodiments of this application; Figure 5 This is a schematic diagram of the energy storage configuration device provided in the embodiments of this application. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0022] In this application, the term "and / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent three cases: A existing alone, A and B existing simultaneously, and B existing alone. In this application, the symbol " / " indicates that the related objects are in an "or" relationship, for example, A / B means A or B.

[0023] In this application, the terms "first" and "second," etc., are used to distinguish different objects, not to describe a specific order of objects. For example, "first response message" and "second response message," etc., are used to distinguish different response messages, not to describe a specific order of response messages.

[0024] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design. Specifically, the use of the terms "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0025] In the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more, for example, multiple processing units means two or more processing units, multiple elements means two or more elements, etc.

[0026] Current methods for configuring energy storage capacity in distribution substations have two main shortcomings: In terms of data modeling, existing methods are mostly based on single probability assumptions such as normal distribution, making it difficult to accurately characterize the strong temporal correlation, multi-peak distribution, and abrupt changes in load and photovoltaic output. They also cannot effectively quantify the impact of nonlinear factors such as climate and user behavior, leading to biases in the assessment of power supply capacity under extreme conditions. In terms of methodology, existing methods lack an integrated design and closed loop for scenario generation, operation simulation, and capacity optimization. They fail to accurately express key aspects through a systematic quantitative formula system, resulting in insufficient scientific rigor and reliability of configuration schemes in complex and uncertain environments. This can easily lead to inappropriate capacity configuration—over-configuration resulting in resource waste, or under-configuration failing to meet power supply requirements, thus making it difficult to adapt to the high requirements for power supply continuity in distribution substations.

[0027] Based on this, this application proposes a method for configuring energy storage in a self-organizing network distribution substation. Please refer to... Figure 1 , Figure 1This is a flowchart illustrating the self-organizing network distribution radio station area energy storage configuration method provided in the embodiments of this application.

[0028] In this embodiment, the method for configuring energy storage in a self-organizing network distribution substation includes: Step S10: Based on the Conditional Time GAN model with gradient penalty, generate a basic scenario set according to the historical load data and historical photovoltaic power output data of the distribution substation.

[0029] It should be noted that step S10 aims to address the problem that traditional probabilistic models struggle to accurately simulate the complex spatiotemporal distribution and dependencies between load and photovoltaic output. This application employs a Conditional Time GAN model based on generative adversarial networks and introduces a gradient penalty mechanism into its training process.

[0030] Specifically, the model comprises a generator and a discriminator. The generator receives a random noise vector sampled from a standard normal distribution and specific conditional information, and outputs synthesized time series data. The discriminator distinguishes whether the input time series data comes from a real historical dataset or is synthesized by the generator, and simultaneously determines whether it matches the given conditional information.

[0031] It should be noted that gradient penalty is a regularization term added to the loss function of the discriminator to constrain the gradient norm of the discriminator with respect to real and synthetic data samples. This can significantly improve the stability of model training, avoid mode collapse, and thus generate higher quality and more diverse time series scenes.

[0032] It should be noted that conditional information is a key variable guiding the data generation process, typically a category label or a continuous variable. In the generation of a distribution area scenario, the conditional information may include date type, season indicator, weather type, etc. For example, date type can be distinguished as weekday, weekend, and holiday; season indicator can be spring, summer, autumn, and winter; weather type can be sunny, cloudy, rainy, or extreme weather.

[0033] It should be noted that historical load data and historical photovoltaic (PV) output data are time-series data collected at fixed time intervals, typically stored in Supervisory Control and Data Acquisition (SCADA) systems or electricity consumption information collection systems, with time resolutions such as 15 minutes or 1 hour. The basic scenario set refers to a large number of statistically representative combinations of load-PV output time-series relationships generated by the model. Each scenario contains a complete cycle, typically point-by-point data for one day or one week. In other words, the basic scenario set is a collection of PV output and load values ​​that reflect historical characteristics.

[0034] Specifically, a Conditional Time GAN model with gradient penalty is first trained to match historical load data from the radio station area. Photovoltaic power output data For training samples, the time step is... The model is generated by the generator. Discriminator The model is structured by introducing gradient penalty (GP) to address the mode collapse problem. The total loss function of the model is: .

[0035] in, For the generator loss, cross-entropy loss is used to characterize the distribution difference between generated samples and real samples: , It is random noise. It is characterized by historical chronology; For discriminator loss: , These are real samples.

[0036] It should be noted that, The gradient penalty term is calculated using the following formula: .

[0037] In the formula, The gradient penalty coefficient is set to 10. , These are the interpolation coefficients; It is a 2-norm.

[0038] Understandably, the generator's input includes historical temporal features. During training, the generator must learn to generate logically coherent subsequent data given a historical sequence, thereby forcing the model to internalize the evolution of data over time, giving the generated scene sequences natural temporal dependencies and coherence.

[0039] Understandably, traditional generative adversarial networks (GANs) are prone to pattern collapse during training, meaning the generator can only reproduce a few data patterns and cannot cover the complex multi-peak distributions in real-world power data, such as various typical load curve shapes. The gradient penalty term, by constraining the norm of the discriminator's gradient, makes the adversarial training process more stable. This stability effectively alleviates the pattern collapse problem, enabling the generator to fully explore and reproduce all major patterns in real-world data, thereby generating curves covering peaks, flat periods, troughs, and other variations.

[0040] Understandably, abrupt changes in load and photovoltaic output data, such as sudden increases / decreases in power, are important extreme features. This model, through its adversarial training framework and temporal structure, can automatically learn correlation patterns in data features before and after abrupt events from historical data. When the conditional information provided to the generator hints at or approaches a possible abrupt change, the generator can generate time-series data containing reasonable, non-smooth abrupt changes based on the learned patterns, thereby preserving these key fluctuation features in the scenario.

[0041] Therefore, by training the model to convergence using equations (1) and (2), a basic load scenario with strong temporal correlation, multi-peak distribution, and abrupt change characteristics can be generated. With basic photovoltaic power output scenarios This refers to the basic scene set.

[0042] Next, step S20 is executed, and the basic scene set is nonlinearly modified for climate factors and user behavior factors through fuzzy logic to generate the target scene set.

[0043] It should be noted that step S20 aims to address the difficulty of existing methods in quantifying the impact of complex nonlinear factors such as climate and user behavior on load and photovoltaic output. This is achieved by employing a fuzzy logic module to correct the basic scenario set. Fuzzy logic is a mathematical framework for handling imprecise, nonlinear information. Its core lies in transforming precise input variables into fuzzy language descriptions such as "low temperature," "medium temperature," and "high temperature" through membership functions. Reasoning is then performed based on an expert knowledge base constructed using "IF-THEN" rules, and finally, the fuzzy reasoning results are inversely solved to output precise correction quantities. This method effectively captures and simulates the complex, nonlinear impact of climate and user behavior factors on energy consumption characteristics without establishing precise physical or statistical equations, making the generated scenarios closer to physical reality.

[0044] It should be noted that the inputs for this step include time-series data points from the basic scenario set in step S10, as well as the corresponding precise input variables that drive the correction. These input variables are mainly divided into two categories: first, climate factor variables, which may include real-time temperature, humidity, wind speed, light intensity, precipitation intensity, etc., with data sourced from meteorological monitoring stations or numerical weather prediction grids associated with the distribution area; second, user behavior factor variables, which may include date type, holiday markers, real-time electricity price signals, special social event markers, etc.

[0045] It should be noted that the output of the fuzzy logic module is a precise power correction value, used to adjust the load value or photovoltaic output value at each moment in the basic scenario. For each input / output variable, its membership function needs to be designed. For example, for the "temperature" variable, fuzzy sets such as "cold" and "hot" can be defined, and a triangular, trapezoidal, or Gaussian membership function can be specified for each set to characterize the degree to which a specific temperature value (such as 25℃) belongs to the "comfortable" set.

[0046] In one feasible implementation, step S20 includes: establishing membership functions for various operating conditions based on climate factors and user behavior factors; defuzzifying the membership functions based on fuzzy inference rules to obtain correction coefficients for load and photovoltaic output; and using the correction coefficients to correct the load scenario and photovoltaic output scenario in the basic scenario set to generate the final target scenario set for simulation.

[0047] It should be noted that this implementation aims to precisely embed the specific impacts of two complex nonlinear factors—climate and user behavior—on electricity into the preliminary scenarios obtained from the generative model in a structured and quantifiable manner. Multiple operating conditions refer to different combinations of states of variables from both climate and user behavior. Operating conditions caused by climate factors include, but are not limited to: clear skies with strong radiation, cloudy skies with fluctuating radiation, overcast skies with low radiation, high temperature and high humidity, and low temperature and cold waves. Operating conditions caused by user behavior factors include, but are not limited to: typical weekdays, weekends, statutory holidays, days of major social events, and typical behavioral patterns during the implementation of time-of-use pricing or demand response events. The specific operating condition corresponding to each scenario in the set is defined by the precise factor variable values ​​input to the fuzzy logic system, and the data can be derived from weather forecasts, calendar information, power grid dispatch plans, etc.

[0048] It should be noted that the target scenario set is a core dataset, defined as: after completing nonlinear correction based on fuzzy logic, the time-series data set that can characterize the various power supply and demand situations that may occur in a distribution substation within a specific time range, such as the next day, under the combined effects of different climate conditions and user behavior patterns. This set consists of numerous scenarios, each of which is a complete time series pair, containing synchronous, time-series-specific photovoltaic power output forecasts and power load forecasts.

[0049] It should be noted that fuzzy inference rules are a set of premise-conclusion pairs stored in a database or knowledge base. For example, a rule might be: IF temperature is high temperature AND date type is weekday, THEN load correction coefficient is positive. During system operation, the specific operating conditions input variables at the current moment are substituted into all membership functions to obtain the membership degree of each premise; fuzzy operators, such as taking the smaller value, are used to obtain the overall activation strength of the rule; subsequently, defuzzification methods such as weighted average or centroid method are used to aggregate the output fuzzy sets corresponding to all activated rules into a precise correction coefficient value. This process is completed cyclically by the fuzzy inference engine software built into a microprocessor or industrial computer.

[0050] Specifically, fuzzy logic is introduced to address climate factors (such as the return of spring). ,high temperature ,rainstorm ) and user behavior factors (holidays) Weekday peak Nonlinear corrections are performed, and correction coefficients are obtained through membership functions, fuzzy inference rules, and defuzzification formulas. (Load Correction) (Photovoltaic output correction), the final corrected scenario value is: .

[0051] in, , For the scene Down The load and photovoltaic output at any given time.

[0052] Specifically, the following steps are taken: A triangular membership function is used to fuzzify the fuzzy input (climate / user behavior level, ranging from 0 to 1) and output (correction coefficient, ranging from 0.5 to 2.0) to reflect the humid weather. For example, its membership function is: .

[0053] It should be noted that, These are the characteristic parameters of the triangular membership function, calibrated based on actual operating data of the distribution substation area. The level of the return of spring The degree of membership.

[0054] Next, the centroid method is used for defuzzification to obtain the determined correction coefficients, as shown in the formula: .

[0055] in, The overall membership degree of the output; These are the upper and lower limits for the output value.

[0056] It should be noted that all scenarios are corrected through equations (3) to (5), generating the final target scenario set covering extreme and special working conditions.

[0057] In summary, please refer to the above process. Figure 2 , Figure 2 This is a schematic diagram illustrating the generation of the target scene set provided in this application embodiment. This flowchart clearly demonstrates a two-stage technical framework for generating the final power scene of a distribution substation, consisting of two core modules. The module on the left is an improved conditional temporal generative adversarial network module. It receives random noise and historical time-series features as conditional inputs. Its core is an adversarial training process: the generator attempts to generate realistic fake samples using the input, while the discriminator is responsible for distinguishing real samples from real historical data from the generated samples produced by the generator. The introduction of a gradient penalty mechanism aims to stabilize this training process. The output of this module is the initial base scene.

[0058] It should be noted that the module on the right is the fuzzy logic correction module. It receives the basic scenario generated by the module on the left and introduces specific external conditions such as "humid weather," "high temperature," and "holidays." Internally, this module follows a standard fuzzy logic processing flow: first, it fuzzifies the precise input conditions into semantic descriptions; then, it performs fuzzy inference based on preset rules; finally, it defuzzifies the inference results and outputs precise correction values. After correcting the basic scenario, this module outputs the final scenario.

[0059] Overall, this diagram illustrates a hybrid modeling process that combines data generation with knowledge correction. The first stage utilizes deep generative models to learn the inherent patterns in historical data and generate diverse basic scenarios. The second stage uses fuzzy logic to inject prior knowledge or explicit rules, such as specific climate and user behavior, into the system to perform targeted calibration of the basic scenarios, thereby obtaining a final scenario set that better reflects actual complex working conditions.

[0060] In step S30, it is necessary to generate the expected amount of unsupplied power corresponding to different energy storage configuration capacities as a supply guarantee indicator based on the energy storage operation simulation model of the distribution substation and the target scenario set.

[0061] It should be noted that the core purpose of step S30 is to establish a quantitative evaluation system for scientifically comparing the supply performance of different energy storage configuration capacity schemes when dealing with complex and uncertain operating conditions.

[0062] The energy storage operation simulation model for distribution substations is a digital virtual model used to simulate and evaluate the dynamic operation and effects of energy storage systems in distribution substations. Its core essence lies in establishing a complete and computable rule system to depict the real-time flow of electricity from production, storage to consumption, as well as the physical constraints and control logic followed by energy storage devices during this process.

[0063] It should be noted that the expected power shortfall is a core power supply guarantee indicator. It is defined as the probability-weighted average of the cumulative power shortage (usually expressed in kilowatt-hours, kWh) caused by a momentary imbalance between photovoltaic output and load in a distribution substation under all possible operating conditions covered by the target scenario set, where the energy storage system cannot fully compensate. It is a risk expectation value that comprehensively reflects the average power supply reliability level of the configuration scheme in long-term operation.

[0064] In one feasible implementation, before step S30, the construction of the distribution substation energy storage operation simulation model includes: constructing the power balance relationship of the distribution substation including the distribution network, photovoltaic, energy storage system and load; establishing the charging and discharging power constraints, state of charge update constraints and state of charge boundary constraints of the energy storage system; and determining the distribution substation energy storage operation simulation model based on the distribution substation power balance relationship, charging and discharging power constraints, state of charge update constraints and state of charge boundary constraints.

[0065] Specifically, the scenario Down The power balance relationship of the distribution radio stations at any given time is as follows: .

[0066] In the formula, Power supplied to the grid side Supplying power to the grid For power transmission from the distribution area to the power grid; For the charging and discharging power of the energy storage system, For energy storage and discharge, Charge the energy storage; The line loss in the transformer substation is calculated based on the impedance parameters of the substation.

[0067] The charging and discharging power of an energy storage system is constrained by power limits, and the state of charge (SOC) is constrained by capacity limits. The core formulas include charging and discharging power constraints and SOC calculations. The charging and discharging power constraints are shown in equation (7): .

[0068] In the formula, This represents the maximum charging power of the energy storage (negative value). This represents the maximum discharge power of the stored energy (positive value).

[0069] For the calculation of charging and discharging of state of charge (SOC), please refer to equations (8) and (9) respectively: ; .

[0070] It should be noted that, The simulation time step is set to 15 min / 60 min. For energy storage charging efficiency, The energy storage discharge efficiency is set to a value of 0.9 to 0.95. The energy storage capacity is configured as the optimization variable in this application; the state-of-charge boundary constraints are also considered. , , The upper and lower limits of the energy storage state of charge are set to 0.2 and 0.9, respectively.

[0071] In one feasible implementation, step S30 includes: for a given energy storage configuration capacity configuration, based on the energy storage operation simulation model of the distribution substation, the target scenario set, and the limit value of the grid-side power supply, obtaining the unsupplied power at each simulation time under different scenarios; based on the occurrence probability of each scenario, weighted summing of the unsupplied power to obtain the expected unsupplied amount corresponding to the energy storage configuration capacity configuration.

[0072] Specifically, using the expected amount of unsupplied electricity as the core indicator of supply guarantee capability, the continuity of power supply in distribution areas under different energy storage capacities is quantified, and the formula is as follows: .

[0073] In the formula, For the scene The probability of occurrence satisfies ; For the scene Down Power not supplied at any given time ; This represents the maximum power supply from the grid side (the value for grid power rationing / outage under extreme conditions is taken).

[0074] The formulas are understandable, and equations (6) to (10) are used to represent different energy storage capacities. Perform full-scenario simulation to obtain the expected unpowered quantity corresponding to each capacity. This forms a quantitative curve of capacity-supply guarantee capability.

[0075] Finally, step S40 is executed: with the goal of minimizing the total life cycle cost of energy storage, and combining the energy storage operation simulation model and supply guarantee index constraints, the optimal energy storage configuration capacity is solved through optimization algorithm to realize the energy storage configuration of the distribution area.

[0076] It should be noted that step S40 is the final decision-making step of this method. Its core objective is to determine the most reasonable scale of energy storage system construction from an economic perspective, while meeting the requirements of power supply reliability. This step integrates the outputs of the preceding steps into a complete mathematical optimization problem. Specifically, it aims to minimize the total life-cycle cost of energy storage, using the supply guarantee index defined in step S30—that is, the expected amount of unsupplied power not exceeding a certain allowable threshold—as the core constraint. The energy storage operation simulation model established in step S30 serves as the basic tool for calculating this supply guarantee index and cost items. The decision variable of this optimization problem is the rated energy capacity of the energy storage system, which can also be extended to the rated power capacity. By solving this problem, a quantitative configuration scheme that achieves a balance between technical reliability and economic optimality is obtained.

[0077] Specifically, minimizing the total lifecycle cost of energy storage involves constructing a single-objective or multi-objective optimization function with energy storage configuration capacity as the decision variable. The objective optimization function includes at least minimizing the total lifecycle cost of energy storage, which includes at least investment cost and operation and maintenance cost.

[0078] It should be noted that the total lifecycle cost of energy storage is an economic evaluation indicator, referring to the sum of the present value of all relevant costs incurred by an energy storage system over its entire service life. It mainly includes the following components: Initial investment cost: This includes one-time investments in the energy storage battery itself, battery management system, power conversion system, civil engineering and installation, and is usually expressed as a function of the energy storage capacity (e.g., yuan / kWh). Operation and maintenance cost: This includes periodic costs such as daily maintenance, power consumption of the management system, and equipment depreciation and replacement, which can be simplified to annual costs proportional to the initial investment.

[0079] Specifically, we establish minimizing the total lifecycle cost of energy storage as a single-objective optimization function: .

[0080] In the formula, Energy storage investment costs: , Investment cost per unit capacity of energy storage (RMB / kWh); Annual operation and maintenance costs for energy storage: , The annual maintenance coefficient is set at 0.02 to 0.05. The annual revenue from energy storage (peak-valley arbitrage, supply guarantee compensation) is determined based on the regional electricity price and supply guarantee policy.

[0081] It should be noted that for multi-objective optimization, a minimum energy storage capacity objective can be added, which can then be transformed into a single objective using a weighted approach. , These are the weighting coefficients.

[0082] Specifically, combining the energy storage operation simulation model and the supply guarantee index constraints involves: constructing constraints for optimization solutions, including: supply guarantee index constraints based on the expected unsupplied power amount corresponding to the energy storage configuration capacity not exceeding the target threshold, and constraints from the energy storage operation simulation model.

[0083] Specifically, the supply guarantee targets are as follows: .

[0084] In the formula, The threshold for the expected amount of unsupplied electricity corresponding to the target supply level of the distribution substation is determined by the power grid planning standards.

[0085] Specifically, the energy storage operation constraints are the same as those in equations (7)-(9), ensuring that the energy storage charging and discharging and state of charge are within a safe range.

[0086] Specifically, the capacity boundary constraints are as follows: .

[0087] In the formula, , The upper and lower limits of energy storage configuration capacity are determined based on the load scale of the distribution area and the photovoltaic installed capacity.

[0088] Finally, please refer to Figure 3 , Figure 3 This is a flowchart illustrating the optimization algorithm provided in this application. An optimization algorithm refers to a class of mathematical computation methods used to automatically search for and determine the optimal decision-making scheme. These include particle swarm optimization (PSO) algorithms, genetic algorithms, etc. Their core mechanism is to systematically iteratively search for the combination of variables that minimizes or maximizes the objective function, while satisfying all given constraints.

[0089] Specifically, the particle swarm optimization algorithm is used to solve the objective function under constraints in order to obtain the optimal energy storage configuration capacity that meets the target supply guarantee level.

[0090] Specifically, the solution to equation (11) is based on the energy storage capacity. For particle position, with objective function value The fitness value is obtained by iteratively solving the particle velocity and position update formulas to find the optimal solution: ; .

[0091] In the formula, Inertial weights; The learning factor is set to 2.0. A random number between 0 and 1; For particles The optimal position of an individual; The globally optimal position; This represents the number of iterations.

[0092] Understandably, the process begins with the data input phase, where basic data of the distribution area, the reliability target for energy supply, and key parameters of the energy storage system are used as known conditions for the algorithm. Subsequently, the algorithm enters the initialization phase, randomly generating a group of particles representing different energy storage capacity schemes and assigning them initial positions and velocities.

[0093] Understandably, the core component is an iterative optimization loop. In each iteration, the algorithm first calculates the fitness value of each particle, which typically corresponds to the total lifecycle cost calculated after substituting the capacity scheme represented by that particle into the simulation model. Simultaneously, the algorithm rigorously checks whether the scheme satisfies supply constraints, operational constraints, and capacity boundaries. Based on the evaluation results, the algorithm updates the optimal solution found by each particle itself and the current optimal solution found by the entire swarm. Then, based on individual and swarm experience, it updates the velocity and position of each particle, thus adjusting its next capacity scheme exploration.

[0094] Understandably, this loop will continue until the preset convergence criteria are met (e.g., reaching the maximum number of iterations or the optimal solution stabilizing). Ultimately, the process outputs the energy storage capacity represented by the globally optimal particle, which serves as the optimal configuration scheme that satisfies all technical constraints and minimizes costs.

[0095] That is, by iterating through equations (11)-(15) until convergence, the optimal energy storage configuration capacity that satisfies the target supply guarantee level is obtained. .

[0096] In summary, the complete framework of the self-organizing grid distribution substation energy storage configuration method proposed in this application can be found in [reference needed]. Figure 4 , Figure 4 This is an overall flowchart of the closed-loop decision-making framework for the self-organizing network distribution station energy storage configuration method provided in the embodiments of this application.

[0097] Specifically, the process begins with historical data collection, gathering multi-source data such as load, photovoltaic (PV) output, climate, and user behavior. This is followed by a scenario generation phase, where Conditional Time GAN is first used to generate scenarios reflecting basic time-series characteristics. Then, fuzzy logic is used to correct and incorporate specific factors such as climate and behavior, ultimately outputting a load-PV output scenario set S for subsequent calculations.

[0098] Specifically, during the simulation phase, the scenario set, the energy storage capacity to be evaluated, and the transformer area parameters are used as inputs. The simulation is carried out under strict constraints such as power balance, charging and discharging, and state of charge, and the key quantitative indicator for ensuring power supply is output—the expected amount of unsupplied power (EENS).

[0099] Specifically, based on the simulation output, the process enters the core stage of optimization and solution. This stage uses economic indicators such as investment cost as the objective function, with the EENS meeting the target supply guarantee level as the core constraint, and employs a particle swarm optimization algorithm to automatically search for and ultimately output the optimal energy storage capacity.

[0100] Specifically, the results will be directly used to guide engineering applications, thus forming a complete and scientific energy storage planning solution from data-driven modeling and simulation evaluation to intelligent optimization decision-making.

[0101] Understandably, in response to the problem of inaccurate assessment of power supply capacity under extreme operating conditions in distribution substations, this application breaks through the limitations of traditional single probability assumptions. It accurately captures the temporal, distribution, and abrupt change characteristics of load and photovoltaic output through a Conditional Time GAN loss function with gradient penalty. Combined with fuzzy logic quantification formulas, it realizes the nonlinear expression of climate factors and user behavior, which greatly enhances the coverage of extreme and special operating conditions. This provides a realistic quantitative scenario basis for power supply capacity assessment and effectively improves the accuracy of assessment results.

[0102] Understandably, this application constructs a closed-loop decision-making framework integrating scenario generation, operational simulation, and capacity optimization, and establishes a comprehensive quantitative formula system. From model training and fuzzy correction in scenario generation, to power balance, energy storage operation, and supply guarantee indicators in operational simulation, and then to the objective function, constraints, and algorithm solution for capacity optimization, all are precisely expressed through mathematical formulas. This solves the problem of the lack of quantitative support in existing methods, making the formulation of energy storage capacity configuration schemes systematic and evidence-based, and improving the scientific nature of the configuration process.

[0103] Understandably, this application uses quantitative formulas from an energy storage operation simulation model to accurately quantify the supply support capabilities of different capacity configurations. It solves for the optimal solution with the expected unsupplied power as the core constraint, ensuring that the configuration scheme meets the target supply level of the distribution substation area. It also takes into account economic efficiency by minimizing the total life cycle cost target, avoiding the problem of excessive or insufficient capacity. This effectively improves the reliability and practicality of the energy storage configuration scheme in complex and uncertain environments, and can directly guide the energy storage planning and construction of distribution substation areas.

[0104] It is understandable that the formula system and model of this application have good universality. By adjusting the feature parameters, weight coefficients, thresholds, etc., it can be adapted to distribution areas of different regions and sizes, solving the problem of poor adaptability of traditional configuration methods and having broad engineering application value.

[0105] Based on the methods in the above embodiments, please refer to Figure 5This application provides an energy storage configuration device, which may include a processor, a communication interface, a memory, and a communication bus. The processor, communication interface, and memory communicate with each other via the communication bus. The processor can invoke logical instructions stored in the memory to execute the methods described in the above embodiments.

[0106] Furthermore, the logical instructions in the aforementioned memory can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application.

[0107] Based on the methods in the above embodiments, this application provides a computer-readable storage medium storing a computer program that, when run on a processor, causes the processor to execute the methods in the above embodiments.

[0108] Based on the methods in the above embodiments, this application provides a computer program product that, when run on a processor, causes the processor to execute the methods in the above embodiments.

[0109] It is understood that the processor in the embodiments of this application can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. A general-purpose processor can be a microprocessor or any conventional processor.

[0110] The method steps in this application embodiment can be implemented in hardware or by a processor executing software instructions. The software instructions can consist of corresponding software modules, which can be stored in random access memory (RAM), flash memory, read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), registers, hard disks, portable hard disks, CD-ROMs, or any other form of storage medium known in the art. An exemplary storage medium is coupled to the processor, enabling the processor to read information from and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can reside in an ASIC.

[0111] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially as a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted through the computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state disk (SSD)).

[0112] It is understood that the various numerical designations used in the embodiments of this application are merely for the convenience of description and are not intended to limit the scope of the embodiments of this application.

[0113] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for configuring energy storage in a self-organizing network distribution radio station area, characterized in that, include: Step S10: Based on the Conditional Time GAN model with gradient penalty, generate a basic scenario set according to the historical load data and historical photovoltaic output data of the distribution substation; the basic scenario set is a set of photovoltaic output and load values ​​that reflect historical characteristics. Step S20: The basic scenario set is nonlinearly corrected for climate factors and user behavior factors using fuzzy logic to generate a target scenario set; the target scenario set is a collection of photovoltaic output and load values ​​under various operating conditions caused by climate factors and user behavior factors. Step S30: Based on the energy storage operation simulation model of the distribution substation, and based on the target scenario set, generate the expected amount of unsupplied power corresponding to different energy storage configuration capacities as a supply guarantee indicator. Step S40: With the goal of minimizing the total life cycle cost of energy storage, and combining the energy storage operation simulation model and supply guarantee index constraints, the optimal energy storage configuration capacity is solved through optimization algorithm to realize the energy storage configuration of the distribution area.

2. The self-organizing network distribution substation energy storage configuration method as described in claim 1, characterized in that, The Conditional Time GAN model with gradient penalty includes a generator and a discriminator. During its training, a gradient penalty term based on random interpolated samples is introduced for the discriminator to solve the mode collapse problem, thereby enabling the model to generate a basic scene set with strong temporal correlation, multi-peak distribution and mutation characteristics.

3. The self-organizing network distribution substation energy storage configuration method as described in claim 1, characterized in that, Step S20 includes: Membership functions for various operating conditions are established, taking into account climate and user behavior factors. The membership function is defuzzified based on the fuzzy inference rules to obtain correction coefficients for load and photovoltaic output. Using the aforementioned correction coefficients, the load scenarios and photovoltaic output scenarios in the basic scenario set are corrected respectively to generate the final target scenario set for simulation.

4. The self-organizing network distribution substation energy storage configuration method as described in claim 1, characterized in that, The steps preceding step S30 also include: Construct a power balance relationship for distribution substations that includes distribution networks, photovoltaics, energy storage systems, and loads; Establish the charging and discharging power constraints, state of charge update constraints, and state of charge boundary constraints for the energy storage system. The simulation model for energy storage operation in the distribution substation is determined based on the power balance relationship, charging and discharging power constraints, state of charge update constraints, and state of charge boundary constraints of the distribution substation.

5. The method for configuring energy storage in a self-organizing network distribution substation as described in claim 1, characterized in that, Step S30 includes: For a given energy storage configuration capacity, based on the energy storage operation simulation model of the distribution substation, the target scenario set, and the limit value of the grid-side power supply, the unsupplied power at each simulation time under different scenarios is obtained; Based on the probability of occurrence of each scenario, the unpowered power is weighted and summed to obtain the expected unpowered amount corresponding to the energy storage configuration capacity.

6. The method for configuring energy storage in a self-organizing network distribution substation as described in claim 1, characterized in that, The goal of minimizing the total lifecycle cost of energy storage is specifically: Construct a single-objective or multi-objective optimization function with the energy storage configuration capacity as the decision variable, wherein the objective optimization function includes at least minimizing the total life cycle cost of energy storage, and the total life cycle cost includes at least investment cost and operation and maintenance cost; The specific constraints, combining the energy storage operation simulation model and supply guarantee indicators, are as follows: The constraints for optimization are constructed, including: a supply guarantee index constraint based on the expected unsupplied power amount corresponding to the energy storage configuration capacity not exceeding the target threshold, and constraints of the energy storage operation simulation model.

7. The method as described in claim 6, characterized in that, The optimal energy storage configuration capacity is determined through optimization algorithms, including: The objective function is solved using a particle swarm optimization algorithm under the constraints to obtain the optimal energy storage configuration capacity that satisfies the target supply guarantee level.

8. An energy storage configuration device, characterized in that, The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the self-organizing network distribution substation energy storage configuration method as described in any one of claims 1 to 7.

9. A readable storage medium comprising instructions, characterized in that, When the instruction is executed on the energy storage configuration device, the energy storage configuration device performs the method as described in any one of claims 1 to 7.

10. A computer program product, comprising a computer program or instructions, characterized in that, When the computer program or instructions are run on the energy storage configuration device, the energy storage configuration device performs the method as described in any one of claims 1 to 7.

Images