Zone area energy storage multi-objective optimization configuration method for improving source load bearing capacity

Abstract

The invention provides a transformer area energy storage multi-objective optimization configuration method for improving the bearing capacity of a source load, and belongs to the technical field of transformer area energy storage, and the method comprises the steps: firstly, continuously collecting the load, distribution network input, photovoltaic power generation, wind power generation, energy storage and other operation data of a transformer area, and building a transformer area operation response equation set; and obtaining a group of input and output data required by optimization through fitting and data augmentation and expansion. Then, the data are constructed into a multi-dimensional matrix, and main features are extracted by using a tensor decomposition technology; based on the characteristics, a multi-objective optimization model is constructed, a GPU accelerated multi-objective optimization algorithm is adopted for solving, and a non-inferior solution set is obtained. And finally, calculating the weighted score of each non-inferior solution, and selecting the energy storage configuration corresponding to the optimal solution as an output result. According to the process, the actual operation data of the transformer area can be fully utilized, the optimization model highly fitting the reality is constructed, the optimal configuration of the energy storage of the transformer area is found, and the problem that the configuration optimization of the energy storage of the transformer area is not comprehensively considered in the prior art is solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of substation energy storage, and in particular, relates to a multi-objective optimization configuration method for substation energy storage for improving source-load carrying capacity. Background Art

[0002] As the power system transforms towards distributed and low-carbon, the access of distributed power sources such as photovoltaic, wind power and other renewable energy sources is increasing, which brings many challenges to the power grid. On the one hand, the intermittent and volatile nature of renewable energy makes the operation of the power grid face greater uncertainty, affecting the safety, stability and power supply reliability of the power grid. On the other hand, large-scale access to renewable energy will lead to power balance and voltage regulation problems in the power grid, and even over-investment. Therefore, how to improve the flexibility and robustness of distributed power grids has become a major problem in the current planning and operation of the power system.

[0003] As an important means to adjust the power balance of the power grid and improve the power quality, the electric energy storage system plays an increasingly important role in the distributed power grid. By rationally configuring and optimizing the control of the electric energy storage system, the fluctuation of renewable energy output can be effectively absorbed, the power supply reliability of the power grid can be improved, and the investment cost of the power grid can be reduced. However, since the configuration of the electric energy storage system is a complex optimization problem involving multiple objectives and multiple constraints, the existing technology for the configuration optimization of regional energy storage is mostly based on single-objective optimization, such as minimizing costs or maximizing economic benefits, ignoring multiple factors such as power grid efficiency, power supply reliability, and environmental impact. This leads to the optimization results being not comprehensive enough and unable to improve the overall performance of the power grid. Summary of the invention

[0004] In view of this, the present invention provides a multi-objective optimization configuration method for substation energy storage to improve source-load carrying capacity, which can solve the technical problem that the existing configuration optimization methods for substation energy storage are mostly based on single-objective optimization and are not comprehensive enough.

[0005] The present invention is achieved in that: The present invention provides a multi-objective optimization configuration method for energy storage in a substation area for improving source-load carrying capacity, comprising the following steps: S10, continuously collect the operation data of the substation, including load data, distribution network input data, photovoltaic power generation data, wind power generation data and energy storage data; S20, establishing a group of substation operation response equations according to the operation data, including a power balance equation, a voltage balance equation, a power loss equation, an energy storage charging and discharging equation, a photovoltaic power generation equation, a wind power generation equation, a load response equation, and a power grid support equation; S30, fitting the substation operation response equation group using the continuously collected substation operation data to obtain a first equation group; and inputting the operation data into the first equation group to obtain a result set; S40, augmenting the collected operation data by data augmentation to obtain augmented operation data, and bringing the augmented operation data into the first set of equations for calculation to obtain an augmented result set; merging the operation data and the augmented operation data into an operation data collection, and merging the result set and the augmented result set into a result data collection; S50, establishing a multidimensional matrix according to the operation data collection and the result data collection; dividing the multidimensional matrix into an input matrix X and an output matrix Y, wherein X includes all input parameters and Y includes all output parameters related to the optimization target; and performing standardization processing on X and Y to eliminate the influence of different dimensions; S60, using tensor decomposition technology to perform dimension reduction and feature extraction on the multidimensional matrix to obtain a simplified feature matrix F; S70, constructing a multi-objective optimization model based on the feature matrix F; S80, using a multi-objective optimization algorithm based on GPU acceleration to solve the multi-objective optimization model to obtain a non-inferior solution set; S90, calculating the weighted score of each non-inferior solution in the non-inferior solution set, selecting the non-inferior solution with the highest weighted score as the optimal solution, and outputting the energy storage data corresponding to the optimal solution as the optimized configuration data of the substation energy storage.

[0006] Furthermore, the power balance equation is specifically expressed as follows: ; In the formula, is the total load power (kW); is the system power loss (kW); is the grid input power (kW); is the photovoltaic power generation power (kW); is wind power generation power (kW); is the energy storage system power (kW, negative for charging and positive for discharging); , , , is the coefficient to be fitted; is the first error term.

[0007] Parameter acquisition method: , , , and Obtained through real-time measurement. Calculated using the power loss equation. , , , Obtained through least squares fitting methods.

[0008] The voltage balance equation is specifically expressed as follows: ; In the formula, is the voltage of node i (V); is the voltage on the power supply side (V); is the number of line segments from node i to the power source; is the resistance of the kth line (Ω); is the reactance of the kth line (Ω); is the current of the kth line (A); is an imaginary unit; , , is the coefficient to be fitted; is the second error term.

[0009] Parameter acquisition method: and Obtained through voltage sensor measurement; and Calculated based on line parameters; Measured by current sensor. , , Obtained through fitting methods such as complex least squares method.

[0010] The power loss equation is specifically expressed as follows: ; In the formula, is the total power loss (kW); is the number of line segments in the system; is the current of the kth line (A); is the resistance of the kth line (Ω); is the core loss coefficient; is the voltage of the kth line (V); is the switching loss coefficient; is the switching frequency (Hz); is the apparent power of the kth line (kVA); , , is the coefficient to be fitted; is the third error term.

[0011] Parameter acquisition method: and Obtained through real-time measurement; Calculated based on line parameters; and Determined through equipment specifications and experimentation; Obtained according to system settings; pass Calculate, where and They are active power and reactive power respectively, which are obtained through measurement. , , Obtained through least squares fitting methods.

[0012] The energy storage charge and discharge equation is specifically expressed as follows: ; In the formula, is the state of charge at time t; is the state of charge at time t+1; For charging efficiency; is the discharge efficiency; is the charging power at time t (kW); is the discharge power at time t (kW); is the time interval (h); is the rated capacity of the energy storage system (kWh); , , is the coefficient to be fitted; is the fourth error term.

[0013] Parameter acquisition method: Obtained through the energy storage system’s battery management system (BMS); and Determined according to the energy storage system specifications; and Real-time measurement through power measurement equipment; Determined by the frequency of data collection; are the design parameters of the energy storage system. , , Obtained through least squares fitting methods.

[0014] The photovoltaic power generation equation is specifically expressed as follows: ; In the formula, is the photovoltaic power generation power (kW); is the photovoltaic panel efficiency; is the photovoltaic panel area (m²); is the solar irradiance (kW / m²); is the temperature coefficient (typically -0.004 to -0.005 / °C); is the photovoltaic cell temperature (°C); is the reference temperature (usually 25°C); , , is the coefficient to be fitted; is the fifth error term.

[0015] Parameter acquisition method: and Design parameters for photovoltaic systems; Obtained through irradiance sensor measurement; Determined according to the specifications of the photovoltaic panels; Obtained through temperature sensor measurement; The temperature is fixed at 25°C. , , Obtained through fitting methods such as nonlinear least squares method.

[0016] The wind power generation equation is specifically expressed as follows: ; In the formula, is wind power generation power (kW); is the air density (kg / m³); is the swept area of ​​the wind rotor (m²); is the wind energy utilization coefficient; is wind speed (m / s); is the cut-in wind speed (m / s); is the cut-out wind speed (m / s); is the rated wind speed (m / s); is the rated power (kW); , is the coefficient to be fitted; is the sixth error term.

[0017] Parameter acquisition method: Determined according to local meteorological conditions; and Design parameters for wind turbines; Obtained through wind speed sensor measurement; , , and is the rated parameter of the wind turbine. , It is obtained through fitting methods such as piecewise nonlinear least squares method.

[0018] The load response equation is specifically expressed as follows: ; ; In the formula, is the actual load power at time t (kW); is the base load power at time t (kW); is the demand response regulation power at time t (kW); is the demand response sensitivity coefficient; is the electricity price at time t (yuan / kWh); is the reference electricity price (yuan / kWh); , , is the coefficient to be fitted; is the seventh error term.

[0019] Parameter acquisition method: Obtained through historical load data and load forecasting models; Determined through statistical analysis and user behavior research; Obtained based on real-time electricity price information; A fixed reference electricity price. , , Obtained through fitting methods such as nonlinear least squares method.

[0020] The grid support equation is specifically expressed as follows: ; In the formula, is the grid support capacity (kVA); is the maximum active power that the power grid can provide (kW); is the actual active power input to the grid (kW); is the maximum reactive power that the power grid can provide (kvar); is the reactive power input into the actual grid (kvar); , is the coefficient to be fitted; is the eighth error term.

[0021] Parameter acquisition method: and Determined based on transformer capacity and grid operation limitations; and It is obtained through real-time measurement by power measurement equipment at the grid input point. , Obtained through least squares fitting methods.

[0022] In the above equations, all the coefficients to be fitted (such as , , , etc.) and the error term ( , The fitting process can use methods such as least squares, nonlinear least squares, and maximum likelihood estimation. The specific choice depends on the characteristics of the equation and the distribution of the data. The error term is assumed to follow a normal distribution with a mean of 0, and the variance needs to be estimated through data.

[0023] In step S50, the multi-dimensional matrix normalization process is specifically represented as follows: ; ; In the formula, and are the input matrix and the output matrix respectively; and are the standardized input matrix and output matrix respectively; and They are and The mean of and They are and The standard deviation of .

[0024] The formulas for calculating the mean and standard deviation are as follows: ; ; in, is the sample size, For the The value of samples. The mean and standard deviation are calculated in the same way.

[0025] Among them, the step of using tensor decomposition technology to reduce the dimension and extract features of the multidimensional matrix to obtain a simplified feature matrix F is specifically: using the Tucker decomposition method to decompose the multidimensional matrix into a core tensor and a factor matrix; using the element size of the core tensor to identify important features and correlations; based on the decomposition results, constructing a simplified feature matrix F to retain the most important feature information.

[0026] Among them, Tucker decomposition is specifically expressed as follows: ; In the formula, is the original multidimensional matrix; is the core tensor; , and is a factor matrix; Represents n-mode product.

[0027] Core Tensor The size is ,in , , are the sizes of each dimension after dimensionality reduction. , , The sizes are , , ,in , , is the original matrix The three dimensions of size.

[0028] Among them, the step of constructing a multi-objective optimization model based on the characteristic matrix F is specifically: using the energy storage data as the decision variable vector, constructing an objective function matrix, including: the objective function of maximizing the system energy conversion efficiency, the objective function of minimizing energy loss, the objective function of maximizing the support role of energy storage to the power grid, the objective function of maximizing power supply reliability, and the objective function of minimizing total cost; setting a constraint matrix, including energy storage capacity and power restrictions, voltage and frequency restrictions, load balance constraints, energy storage charging and discharging depth and number of times restrictions, renewable energy output restrictions, and investment budget restrictions. Specifically, the objective function matrix construction is specifically expressed as follows: ; In the formula, is the objective function matrix; is the decision variable vector; Maximize the objective function for system energy conversion efficiency; Minimize the objective function for energy loss; The objective function for maximizing the support effect of energy storage on the power grid; Maximizing the objective function for power supply reliability; Minimize the objective function for the total cost.

[0029] The specific expressions of each objective function are as follows: 1. Maximize system energy conversion efficiency: ; 2. Minimize energy loss: ; 3. Maximizing the role of energy storage in supporting the power grid: ; 4. Maximize power supply reliability: ; 5. Minimize total cost: ; in, is the total number of time steps; is the time step; and are the output power and input power at time t respectively; is the power loss at time t; is the grid support capacity at time t; and are the power supply and power demand at time t respectively; For investment costs; is the operating cost at time t.

[0030] The GPU-accelerated non-dominated sorting algorithm consists of the following steps: 1. Dominance relationship matrix calculation: ; In the formula, is the element of the dominance relationship matrix; and Respectively and A solution; For the An objective function.

[0031] 2. Dominance count calculation: ; In the formula, For the The dominated count of solutions; is the total number of solutions.

[0032] 3. Pareto front determination: ; In the formula, is the first-level Pareto front, which contains all solutions with a dominated count of 0.

[0033] 4. Crowding distance calculation: ; In the formula, For the The crowding distance of each solution; is the number of objective functions; and In the The function value of two adjacent solutions on the target; and Respectively The maximum and minimum values ​​of the objective function.

[0034] S90. Calculation of comprehensive score of non-inferior solution: The non-inferior solution comprehensive score calculation is specifically expressed as follows: ; In the formula, For the The comprehensive score of the non-inferior solutions; is the number of objective functions; For the The weight of the objective function satisfies ; For the The non-inferior solution is The value of the objective function; and Respectively The minimum and maximum values ​​of the objective function among all non-inferior solutions.

[0035] Weight The determination of can be calculated using methods such as the analytic hierarchy process (AHP) or the entropy weight method based on the decision maker's preference or the importance of the objective function.

[0036] Selection of the optimal solution: ; In the formula, is the optimal solution, that is, the non-inferior solution with the highest comprehensive score.

[0037] Compared with the prior art, the beneficial effects of the multi-objective optimization configuration method for energy storage in a substation area for improving the source-load carrying capacity provided by the present invention are: 1. A detailed response model covering all aspects of the substation has been established. The present invention continuously collects the operating data of the substation's load, distribution network, renewable energy, and energy storage system, and establishes a comprehensive set of response equations including power balance, voltage characteristics, loss analysis, energy storage behavior, renewable energy output, load response, and grid support. This systematic modeling method can more accurately describe the actual operating characteristics of the substation and provide a reliable basis for subsequent optimization decisions.

[0038] 2. The generalization ability of the model is improved by using data augmentation technology. Based on the original operation data, the present invention generates a richer simulation operation data set by using methods such as random interpolation, rolling window, and time series model. This not only expands the coverage of training data and enhances the model's ability to adapt to future uncertainties, but also provides more extensive data support for subsequent optimization modeling.

[0039] 3. Advanced tensor decomposition technology is applied to extract key features. The present invention uses the Tucker decomposition method to reduce the dimension and extract features of the multi-dimensional operation data matrix, and obtains a simplified feature matrix. This feature extraction technology can effectively identify the most important information in the data and provide high-quality input for subsequent multi-objective optimization modeling.

[0040] 4. A comprehensive optimization model considering multiple factors was constructed. Based on feature extraction, the present invention constructed a multi-objective optimization model including five objective functions, namely, system energy conversion efficiency, energy loss, grid support, power supply reliability, and total cost. This comprehensive optimization goal can better reflect the various performance indicators of grid operation and provide a more balanced and comprehensive basis for the final optimal energy storage configuration.

[0041] 5. Use GPU-accelerated multi-objective optimization algorithms to solve. The present invention takes advantage of GPU parallel computing and uses efficient multi-objective optimization algorithms such as non-dominated sorting to greatly improve the solution efficiency. This algorithm can find a non-inferior solution set for the electric energy storage system within a limited time, providing decision makers with a variety of alternatives to meet the time requirements in practical applications.

[0042] In summary, the multi-objective optimization configuration method for energy storage in the substation area proposed in the present invention can make full use of real-time monitoring data for comprehensive modeling, and use advanced feature extraction and optimization algorithms to obtain high-quality configuration decisions. While improving the operation performance of the power grid, it also provides effective support for the flexibility and reliability of the distributed power grid. This solves the technical problem that most of the existing configuration optimization methods for substation energy storage are based on single-objective optimization and are not comprehensive enough. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1A flow chart of the method provided by the present invention. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0045] like Figure 1 As shown, it is a flow chart of a multi-objective optimization configuration method for energy storage in a substation area for improving the source-load carrying capacity provided by the present invention. The specific implementation method of each step of the method is described in detail below: The specific implementation of step S10 is: continuously collect the operation data of the station area. Specifically, it includes the following sub-steps: First, collect load data, including active power, reactive power, voltage, current and power factor. These data reflect the power consumption characteristics of users in the substation area and are important inputs for subsequent optimization. It is recommended to set the data collection frequency to 1 minute to 15 minutes to capture the dynamic changes of the substation load.

[0046] Secondly, the distribution network input data is collected, including feeder voltage, current, active power, reactive power and power factor. These data reflect the power supply quality on the input side of the grid and can be used to analyze the grid support capacity. The collection frequency is consistent with the load data.

[0047] Secondly, the photovoltaic power generation data is collected, including output active power, reactive power, voltage, current, irradiance and temperature. These data reflect the power generation status of the photovoltaic system and can be used to analyze the changing characteristics of renewable energy.

[0048] At the same time, wind power generation data is collected, including output active power, reactive power, voltage, current, wind speed and wind direction. These data reflect the power generation status of the wind power generation system and can be used to analyze the changing characteristics of renewable energy.

[0049] Finally, the energy storage system data is collected, including charging and discharging power, state of charge, voltage, current and temperature. These data reflect the operating status of the energy storage system and are direct inputs for optimizing energy storage configuration.

[0050] In summary, the purpose of step S10 is to continuously collect various important operation data of the substation to provide a detailed data basis for subsequent modeling, optimization and decision-making. These data cover all aspects such as load, power grid, renewable energy and energy storage system, laying the foundation for building a comprehensive substation response model.

[0051] The specific implementation of step S20 is: according to the collected operation data, establish the substation operation response equation group. Specifically, it includes the following equations: 1. Power balance equation: ; This equation describes the balance between the total load, power loss and the power supplied from the grid, photovoltaic, wind power and energy storage system. to and the error term It needs to be obtained through historical data fitting.

[0052] 2. Voltage balance equation: ; This equation describes the relationship between the node voltage and the supply voltage, line resistance, reactance, and current. , and It needs to be obtained through historical data fitting.

[0053] 3. Power loss equation: ; This equation describes the total power loss of the system, including line loss, core loss and switching loss. to It needs to be obtained through historical data fitting.

[0054] 4. Energy storage charging and discharging equation: ; This equation describes the charging and discharging process of the energy storage system and its impact on the battery state of charge SOC. to It needs to be obtained through historical data fitting.

[0055] 5. Photovoltaic power generation equation: ; This equation describes the relationship between photovoltaic power generation and irradiance and battery temperature. to It needs to be obtained through historical data fitting.

[0056] 6. Wind power generation equation: ; This equation describes the relationship between wind power generation and wind speed, including the segmented characteristics of cut-in, rated and cut-out wind speeds. and It needs to be obtained through historical data fitting.

[0057] 7. Load response equation: ; ; This equation describes the response characteristics of the load to the electricity price signal, including the benchmark load and the demand response regulation part. to It needs to be obtained through historical data fitting.

[0058] 8. Grid support equation: ; This equation describes the active and reactive support capacity that the grid can provide. and It needs to be obtained through historical data fitting.

[0059] In summary, the purpose of step S20 is to establish a response model covering all aspects of the substation, including power balance, voltage characteristics, loss analysis, energy storage behavior, renewable energy output, load response and grid support, etc. These equations constitute a complete substation simulation model, which provides a basis for subsequent optimization decisions.

[0060] The specific implementation method of step S30 is: using the continuously collected substation operation data to fit the established substation response equation group, obtaining the first set of equation group parameters, and bringing the operation data into the equation group to calculate the result set.

[0061] First, the eight response equations established in step S20 are fitted with the coefficients of each equation using the least squares method, nonlinear least squares method or maximum likelihood estimation method in combination with the actual collected operation data to obtain the first set of parameters. In this way, a preliminary substation response model is established.

[0062] Next, the actual operation data collected, including load, power grid, renewable energy and energy storage system data, are brought into the above fitted equations for calculation. In this way, a set of result data that matches the actual operation situation is obtained, which is recorded as the result set.

[0063] The purpose of this step is: 1) to obtain the initial parameters of the substation response model through fitting; 2) to verify whether the established model can accurately describe the operating characteristics of the actual substation. The result set provides a benchmark and reference for subsequent data augmentation and optimization.

[0064] The specific implementation of step S40 is: augmenting the collected operation data by data augmentation to obtain augmented operation data, and bringing the fitted equation group into calculation to obtain an augmented result set. Then, the original operation data and the augmented operation data are merged into an operation data collection, and the result set and the augmented result set are merged into a result data collection.

[0065] Specifically, firstly, some data augmentation techniques are used, such as random interpolation, rolling window, time series model, etc., to augment and enhance the original operation data collected in step S10. In this way, a set of augmented simulation operation data is obtained, which better covers various possible situations.

[0066] Next, these augmented operating data are brought into the substation response equation group obtained by fitting in step S30 to calculate the corresponding augmented result set.

[0067] Finally, the original operation data and the augmented operation data are merged into a richer operation data collection, and the result set and the augmented result set are merged into a result data collection.

[0068] The purpose of this step is: 1) to expand the coverage of the training data set and improve the generalization ability of the model through data augmentation; 2) to generate richer input and output data pairs to lay a solid foundation for subsequent optimization modeling.

[0069] The specific implementation method of step S50 is: establishing a multidimensional matrix based on the operation data collection and the result data collection, and standardizing the input matrix X and the output matrix Y.

[0070] First, organize the run data collection and result data collection into a multidimensional matrix Specifically, the rows of the matrix represent different time points, the columns represent different data types and parameters, and the depth dimension represents different scenarios or augmented data sets. In this way, a rich multidimensional data structure is constructed, laying the foundation for subsequent feature extraction and optimization modeling.

[0071] Next, the multidimensional matrix is ​​divided into input matrix X and output matrix Y. X contains all input parameters, such as measured data of load, power grid, renewable energy and energy storage system; Y contains all output parameters related to the optimization target, such as energy conversion efficiency, energy loss, power grid support, etc.

[0072] Finally, the input matrix X and the output matrix Y are standardized to eliminate the influence of different dimensions. The standardization formula is: ; ; in, , , and are the mean and standard deviation of X and Y respectively.

[0073] The purpose of this step is: 1) to build a multidimensional data structure to better organize and represent complex operating data; 2) to standardize the input and output data to provide a standardized data foundation for subsequent feature extraction and optimization modeling.

[0074] The specific implementation method of step S60 is: using the Tucker decomposition method to reduce the dimension and extract features of the multidimensional matrix to obtain a simplified feature matrix F.

[0075] Tucker decomposition is a multidimensional data decomposition method that can convert the original multidimensional matrix into Approximate decomposition into a core tensor and three factor matrices , , : ; in, Represents an n-mode product. Core tensor The size is , reflects the most important features and correlations in the data. Factor matrix , , The sizes are , , , which contains the principal components of the data in each dimension.

[0076] Through Tucker decomposition, key features in multidimensional data can be effectively extracted and data dimensions can be greatly compressed. In this problem, the results of Tucker decomposition can be used to construct a simplified feature matrix F, which only contains the most important feature information and provides high-quality input for subsequent optimization modeling.

[0077] The purpose of this step is: 1) to use tensor decomposition techniques such as Tucker decomposition to extract key features from multidimensional data; 2) to construct a simplified feature matrix F, which significantly reduces the data dimension while retaining the main information, providing efficient input for optimization modeling.

[0078] The specific implementation method of step S70 is: based on the feature matrix F, construct a multi-objective optimization model.

[0079] Specifically, the decision variable vector of the energy storage system is recorded as , construct the objective function matrix , including the following five objective functions: ; 1. Objective function for maximizing system energy conversion efficiency : ; 2. Energy loss minimization objective function : ; 3. Objective function for maximizing the support role of energy storage for the power grid : ; 4. Power supply reliability maximization objective function : ; 5. Total cost minimization objective function : ; At the same time, the corresponding constraint matrix is ​​set , including energy storage capacity and power limitations, voltage and frequency limitations, load balance constraints, energy storage charging and discharging depth and frequency limitations, renewable energy output limitations, and investment budget limitations.

[0080] In summary, the purpose of step S70 is to construct a multi-objective optimization model based on the feature extraction results, taking into account factors such as system efficiency, loss, reliability, and cost, so as to provide a basis for subsequent optimal energy storage configuration.

[0081] The specific implementation method of step S80 is: using a multi-objective optimization algorithm based on GPU acceleration to solve the above multi-objective optimization model to obtain a non-inferior solution set.

[0082] Specifically, the objective function matrix is ​​first calculated in parallel on the GPU and the constraint matrix Then, the non-dominated sorting algorithm implemented on GPU is used to solve the multi-objective optimization problem through the following steps: 1. Calculate the dominance relationship matrix , determine whether each solution is dominated by other solutions.

[0083] 2. Compute the dominance count of each solution in parallel and the dominated set.

[0084] 3. Determine the Pareto frontier , that is, the set of solutions whose dominated count is 0.

[0085] 4. Calculate the crowding distance of each solution , reflects the distribution of solutions in the target space.

[0086] 5. Perform selection, crossover and mutation operations based on non-dominated sorting and crowding distance, and iteratively optimize.

[0087] 6. Repeat steps 2-5 until the termination condition is reached and the final set of non-inferior solutions is obtained.

[0088] The algorithm makes full use of the parallel computing power of GPU and greatly improves the efficiency of multi-objective optimization. Non-dominated sorting and crowding distance calculation ensure the convergence and dispersion of the final solution set, meeting the requirements of multi-objective optimization.

[0089] The purpose of this step is to use GPU acceleration technology to solve complex multi-objective optimization problems, obtain a non-inferior solution set for the substation energy storage system, and provide alternative solutions for subsequent optimal selection.

[0090] The specific implementation method of step S90 is: selecting the solution with the highest comprehensive score from the non-inferior solution set as the optimal energy storage configuration solution.

[0091] Specifically, firstly, each objective function value of each non-inferior solution is normalized so that its numerical range is unified to the interval [0,1]: ; in, For the The comprehensive score of non-inferior solutions, For the The weight of the objective function, For the The non-inferior solution is The value of the objective function, and Respectively The minimum and maximum values ​​of the objective function among all non-inferior solutions.

[0092] Weight The analytic hierarchy process (AHP) or entropy weight method can be used to determine the decision-maker's preference or the importance of the objective function to meet the requirements. .

[0093] Finally, select the comprehensive score The highest non-inferior solution is taken as the optimal solution , and outputs the corresponding energy storage system capacity, power and optimized values ​​of each objective function.

[0094] The purpose of this step is: 1) to calculate the comprehensive score of each non-inferior solution according to the decision maker's preference; 2) to select the solution with the highest score from the non-inferior solution set as the optimal energy storage configuration plan to provide a basis for subsequent implementation.

[0095] In this multi-objective optimization configuration method for energy storage in a substation, a multidimensional matrix is ​​a key data structure used to organize and represent complex operating data. The following is a step-by-step analysis of this multidimensional matrix: 1. Dimensions of the matrix: Row: represents different time points; Column: represents different data types and parameters; Depth: Representing different scenarios or augmenting datasets; 2. The specific meaning of depth: The depth dimension here represents different scenes or augmented data sets. This means that the matrix is ​​not just two-dimensional, but contains multiple "layers" or "slices". Each layer represents a specific scene or a set of augmented data.

[0096] 3. The role of depth: Scenario representation: Each layer may represent a different operating scenario, such as normal working days, weekends, holidays, or different seasons.

[0097] Data augmentation: Additional data sets generated by data augmentation techniques (such as those mentioned in step S40), each augmented data set may occupy one depth layer of the matrix.

[0098] Rich information: The depth dimension allows the model to consider more changes and possibilities, improving the robustness and adaptability of the optimization results.

[0099] 4. Example description: Suppose there is a 3x3x3 multidimensional matrix: 3 rows represent 3 time points (e.g. morning, midday, evening); The 3 columns represent 3 data types (e.g., load, photovoltaic, and energy storage); The three depth layers represent three scenes (e.g., sunny, cloudy, and rainy); This matrix can be represented as: Depth 1 (sunny day): [ [load 1, photovoltaic 1, energy storage 1], [load 2, photovoltaic 2, energy storage 2], [load 3, photovoltaic 3, energy storage 3]]; Depth 2 (cloudy): [ [load 1', photovoltaic 1', energy storage 1'], [load 2', photovoltaic 2', energy storage 2'], [load 3', photovoltaic 3', energy storage 3']]; Depth 3 (rainy day): [ [load 1", PV 1", energy storage 1"], [load 2", PV 2", energy storage 2"], [load 3", PV 3", energy storage 3"]]; 5. Advantages of multidimensional matrix: Data Integration: Integrate large amounts of complex operational data into a unified structure.

[0100] Scenario Analysis: Ability to consider multiple operating scenarios and conditions simultaneously.

[0101] Optimization basis: Provides a rich data foundation for subsequent feature extraction and multi-objective optimization.

[0102] In summary, this multi-objective optimization configuration method for energy storage in the substation area to improve the source-load carrying capacity makes full use of the detailed operation data of the substation area, establishes a response model covering all aspects, uses data augmentation technology to expand the coverage of the training data set, uses Tucker decomposition to extract key features, constructs a multi-objective optimization model, and uses GPU-accelerated multi-objective optimization algorithm to solve the non-inferior solution set, and finally selects the optimal energy storage configuration plan with the highest comprehensive score. This method fully considers multiple objectives such as system efficiency, loss, reliability, and cost, and provides a systematic and comprehensive solution for the optimal configuration of substation energy storage systems.

[0103] Specifically, the principle of the present invention is: First, this method continuously collects various operating data of the substation, including real-time monitoring data of load, distribution network, renewable energy and energy storage system, and establishes a detailed response model covering all aspects of the substation. These response equations include power balance, voltage characteristics, power loss, energy storage charging and discharging, renewable energy output and grid support, which can accurately describe the actual operating characteristics of the substation. Compared with the existing methods that only focus on a single objective, this comprehensive modeling method lays a solid foundation for subsequent multi-objective optimization.

[0104] Secondly, in order to enhance the adaptability of the model to future uncertainties, the present invention adopts data augmentation technology. Through random interpolation, rolling window, time series prediction and other methods, a richer simulation operation data set is generated. This not only expands the coverage of training data, but also provides more extensive data support for subsequent optimization modeling. Compared with existing methods that rely only on limited historical data, this data augmentation technology can better predict and deal with uncertainties in power grid operation.

[0105] Thirdly, the present invention applies the advanced Tucker decomposition technology to reduce the dimension and extract features from the multi-dimensional operation data matrix. This tensor decomposition method can effectively identify the most important features and correlations in the data, greatly compress the data dimension, and provide high-quality input for subsequent multi-objective optimization modeling. Compared with the existing methods that rely solely on empirical feature selection, this data-driven feature extraction technology is more objective, systematic and efficient.

[0106] Finally, based on the extracted key features, the present invention constructs a comprehensive optimization model that considers multiple factors. The model includes five objective functions, namely, system energy conversion efficiency, energy loss, grid support, power supply reliability, and total cost, which can more comprehensively reflect the various performance indicators of grid operation. Compared with the existing methods that only focus on single-objective optimization, this multi-objective optimization method can seek a more balanced solution between various objectives and provide a more scientific decision-making basis for the final optimal energy storage configuration.

[0107] In order to efficiently solve this complex multi-objective optimization problem, the present invention adopts a non-dominated sorting algorithm based on GPU acceleration. This algorithm makes full use of the parallel computing power of the GPU and greatly improves the solution efficiency. By calculating the dominance relationship and congestion distance of each solution, the algorithm can find a non-inferior solution set for the electric energy storage system within a limited time, providing decision makers with a variety of alternatives. This optimization algorithm based on GPU acceleration meets the demand for fast decision-making in practical applications.

[0108] To sum up, the multi-objective optimization configuration method for substation energy storage proposed in the present invention is based on a number of key technologies such as comprehensive modeling, data augmentation, feature extraction, multi-objective optimization, and GPU accelerated optimization algorithm, and systematically solves the problems existing in the existing electric energy storage configuration methods.

[0109] In order to better understand and implement the present invention, a specific embodiment 1 of the present invention is provided below. The specific implementation of each step in this embodiment 1 is described in detail as follows: The specific implementation of step S10 is as follows: The purpose of this step is to continuously collect various operating data of the substation to provide a detailed data basis for subsequent modeling and optimization. The collected data includes load data, distribution network input data, photovoltaic power generation data, wind power generation data and energy storage data.

[0110] First, for load data, the following parameters were collected: Active Power : reflects the electricity demand of users in the area; reactive power : reflects the power consumption of users in the area; voltage : reflects the voltage level on the user side of the substation; current : reflects the current change on the user side of the substation; power factor : Reflects the power factor on the user side of the substation.

[0111] These load parameters can fully describe the power consumption characteristics of users in the substation area. The data collection frequency is recommended to be set to minutes to capture the dynamic changes of load.

[0112] Secondly, for the distribution network input data, the following parameters were collected: feeder voltage : reflects the voltage level on the grid input side; feeder current : reflects the current change on the input side of the power grid; active power : reflects the active power on the input side of the power grid; reactive power : reflects the reactive power on the input side of the power grid; power factor : Reflects the power factor on the grid input side.

[0113] These distribution network input parameters can reflect the power supply quality and support capacity of the power grid. The data collection frequency is also set to minute.

[0114] Again, for photovoltaic power generation data, the following parameters were collected: Active Power : reflects the output active power of the photovoltaic system; Reactive power : reflects the output reactive power of the photovoltaic system; Voltage : reflects the output voltage of the photovoltaic system; Current : reflects the output current of the photovoltaic system; Irradiance : reflects the lighting conditions; temperature : reflects the operating temperature of photovoltaic cells.

[0115] These photovoltaic power generation parameters can fully reflect the power generation status of the photovoltaic system and provide a basis for analyzing the changing characteristics of renewable energy.

[0116] At the same time, wind power generation data was also collected, including: Active Power : reflects the output active power of the wind power generation system; Reactive power : reflects the output reactive power of the wind power generation system; Voltage : reflects the output voltage of the wind power generation system; Current : reflects the output current of the wind power generation system; Wind speed : reflects wind conditions; wind direction :Reflects changes in wind direction.

[0117] These wind power generation parameters can fully reflect the power generation status of the wind system and provide a basis for analyzing the changing characteristics of renewable energy.

[0118] Finally, for the energy storage system data, the following parameters were collected: Charge and discharge power : reflects the charge and discharge status of the energy storage system, a positive value indicates discharge, and a negative value indicates charge; State of Charge : reflects the battery charge level of the energy storage system; Voltage : reflects the output voltage of the energy storage system; Current : reflects the output current of the energy storage system; temperature : reflects the operating temperature of the energy storage system.

[0119] These energy storage data can comprehensively reflect the operating status of the energy storage system and provide key input for subsequent optimization configuration.

[0120] In summary, this step provides a detailed data basis for subsequent modeling, optimization and decision-making by continuously collecting the above-mentioned types of operating data, laying the foundation for the efficient configuration of the substation energy storage system.

[0121] The specific implementation of step S20 is as follows: The purpose of this step is to establish a set of response equations covering all aspects of the substation based on the collected operating data, providing a basic model for subsequent optimization modeling. The established set of equations includes: 1. Power balance equation: ; in, is the total load power ; is the total power loss of the system ; Power input to the grid ; Photovoltaic power generation ; Wind power generation ; The power of the energy storage system , charging is negative, discharging is positive; , , , is the coefficient to be fitted; is the error term. This equation describes the balance between the total load and power loss of the substation and the power supplied by the power grid, renewable energy and energy storage system.

[0122] 2. Voltage balance equation: ; in, For Node Voltage ; The power supply voltage ; For Node Number of line segments to the power source; For the Segment line resistance ; For the Reactance of the line segment ; For the Current of the segment line ; is an imaginary unit; , , is the coefficient to be fitted; is the error term. This equation describes the relationship between the node voltage and the power supply voltage, line parameters and current.

[0123] 3. Power loss equation: ; in, The total power loss ; is the number of line segments in the system; For the Current of the segment line ; For the Segment line resistance ; is the core loss coefficient; For the Voltage of the segment line ; is the switching loss coefficient; is the switching frequency ; For the Apparent power of the line segment ; , , is the coefficient to be fitted; is the error term. This equation describes the total power loss of the system, including line loss, core loss and switching loss.

[0124] 4. Energy storage charging and discharging equation: ; in, for State of charge at the moment; for State of charge at the moment; For charging efficiency; is the discharge efficiency; for Charging power at the moment ; for Discharge power at the moment ; For time interval ; is the rated capacity of the energy storage system ; , , is the coefficient to be fitted; is the error term. This equation describes the charging and discharging process of the energy storage system and its effect on the battery state of charge impact.

[0125] 5. Photovoltaic power generation equation: ; in, Photovoltaic power generation ; is the photovoltaic panel efficiency; is the photovoltaic panel area ; Solar irradiance ; is the temperature coefficient Usually ; is the photovoltaic cell temperature ; is the reference temperature Usually ; , , is the coefficient to be fitted; is the error term. This equation describes the variation characteristics of photovoltaic power generation with irradiance and battery temperature.

[0126] 6. Wind power generation equation: ; in, Wind power generation ; is the air density ; The swept area of ​​the wind rotor ; is the wind energy utilization coefficient; Wind speed ; Cut-in wind speed ; To cut out wind speed ; Rated wind speed ; Rated power ; , is the coefficient to be fitted; is the error term. This equation describes the segmented characteristics of wind power generation with wind speed.

[0127] 7. Load response equation: ; ; in, for Actual load power at the moment ; for Base load power at the time ; for Moment-by-moment demand response to adjust power ; is the demand response sensitivity coefficient; for Electricity price at the time ; Reference electricity price ; , , is the coefficient to be fitted; is the error term. This equation describes the dynamic response characteristics of the load to the electricity price signal.

[0128] 8. Grid support equation: ; in, Supporting capacity for the power grid ; The maximum active power that the power grid can provide ; The active power input from the actual grid ; The maximum reactive power that the power grid can provide ; Reactive power input from the actual power grid ; , is the coefficient to be fitted; is the error term. This equation describes the active and reactive support capabilities that the power grid can provide.

[0129] In summary, this step establishes a detailed response model covering all aspects of the substation, which provides a basis for subsequent optimization modeling and decision-making. All coefficients and error terms to be fitted need to be estimated through a large amount of historical data.

[0130] The specific implementation of step S30 is as follows: The purpose of this step is to use the continuously collected substation operation data to fit the substation response equation group established in step S20, obtain preliminary model parameters, and verify the accuracy of the model.

[0131] Specifically, the least squares method, nonlinear least squares method or maximum likelihood estimation method is first used to perform parameter fitting on the eight response equations in step S20 in combination with the actual operation data collected in step S10 to obtain the first set of fitting coefficients: ; In this way, a preliminary substation response model was established.

[0132] Next, the operating data actually collected in step S10, including

[0133] ; etc., and substitute them into the above fitted equations for calculation, and obtain a set of result data that is consistent with the actual situation, which is recorded as the result set.

[0134] Through this step, a preliminary substation response model was established, and it was verified whether the model can accurately describe the operating characteristics of the actual substation, which laid the foundation for subsequent data augmentation and optimization modeling.

[0135] The specific implementation of step S40 is as follows: The purpose of this step is to use data augmentation technology to expand the coverage of the training data set, improve the generalization ability of the model, and provide richer data support for subsequent optimization modeling.

[0136] Specifically, the original operation data collected in step S10 is first expanded and enhanced using the following data augmentation techniques: 1. Random interpolation method: Randomly interpolate some missing points in the collected data to generate new data samples.

[0137] 2. Rolling window method: Roll the time window of collected data to generate more time series data.

[0138] 3. Time series model: Train the time series prediction model based on the collected data and generate simulated data.

[0139] Through the above technology, a set of augmented simulation operation data is obtained, which is recorded as augmented operation data.

[0140] Next, these augmented operation data are substituted into the substation response equation group obtained by fitting in step S30, and calculations are performed and the results are recorded to obtain an augmented result set.

[0141] Finally, the original operation data collected in step S10 and the augmented operation data generated above are merged into a richer operation data collection, and the result set of step S30 and the augmented result set are merged into a result data collection.

[0142] Through this step, the coverage of the training data set is expanded, the generalization ability of the model is improved, and a more solid data foundation is laid for subsequent optimization modeling.

[0143] The specific implementation of step S50 is as follows: The purpose of this step is to construct a multidimensional data matrix based on the operation data collection and the result data collection, and to standardize the input and output data to provide a basis for subsequent feature extraction and optimization modeling.

[0144] First, organize the run data collection and result data collection into a multidimensional matrix ,in: OK Represents different time points ; List Represents different data types and parameters, such as

[0145] wait; depth Represent different scenarios or augmented datasets.

[0146] In this way, a rich multidimensional data structure is constructed , which laid the foundation for subsequent feature extraction and optimization modeling.

[0147] Next, the multidimensional matrix Divide into input matrix And the output matrix .in, Contains all input parameters, such as measured data of loads, power grids, renewable energy sources and energy storage systems; Contains output parameters related to the optimization objectives, such as energy conversion efficiency, energy loss, grid support, etc.

[0148] Finally, the input matrix And the output matrix Standardize them separately to eliminate the influence of different dimensions: ; ; in, and They are The mean and standard deviation of and They are The mean and standard deviation of .

[0149] Through this step, a well-organized multidimensional data structure is constructed. , and the input and output data were standardized, providing a standardized data basis for subsequent feature extraction and optimization modeling.

[0150] The specific implementation of step S60 is as follows: The purpose of this step is to use the Tucker decomposition method to decompose the multidimensional matrix constructed in step S50. Perform dimensionality reduction and feature extraction to obtain a simplified feature matrix , providing high-quality input for subsequent optimization modeling.

[0151] Tucker decomposition is a multidimensional data decomposition method that can convert the original multidimensional matrix into Approximate decomposition into a core tensor and three factor matrices , , : ; in, Represents an n-mode product. Core tensor The size is , reflects the most important features and correlations in the data. Factor matrix , , The sizes are , , , which contains the principal components of the data in each dimension.

[0152] Specifically, firstly, the multidimensional matrix Perform Tucker decomposition to obtain the core tensor and the factor matrix , , Among them, the core tensor The size of the elements in reflects the importance of each feature in the data.

[0153] Next, according to The size of the elements in the graph, select the most important features, construct a simplified feature matrix This not only retains the main information of the data, but also greatly compresses the data dimension, providing high-quality input features for subsequent optimization modeling.

[0154] Through this step, tensor decomposition techniques such as Tucker decomposition are used to effectively extract key features in multidimensional data, laying the foundation for subsequent optimization modeling.

[0155] The specific implementation of step S70 is as follows: The purpose of this step is to extract the feature matrix based on step S60. ,A multi-objective optimization model is constructed to provide a basis for the subsequent optimal energy storage configuration.

[0156] Specifically, the decision variable vector of the energy storage system is recorded as , construct the following five objective functions: 1. Objective function for maximizing system energy conversion efficiency: ; in, for Output power at the moment, for The input power at the moment, is the time step, is the total number of time steps. The objective function aims to maximize the energy conversion efficiency of the system.

[0157] 2. Energy loss minimization objective function: ; in, for The objective function aims to minimize the total energy loss of the system.

[0158] 3. Objective function for maximizing the support effect of energy storage on the power grid: ; in, for The objective function aims to maximize the support of the energy storage system to the power grid.

[0159] 4. Power supply reliability maximization objective function: ; in, for The power supply at the time, for The objective function aims to maximize the power supply reliability.

[0160] 5.Total cost minimization objective function: ; in, is the investment cost, for The objective function aims to minimize the total cost.

[0161] At the same time, corresponding constraints are set, including energy storage capacity and power limitations, voltage and frequency limitations, load balance constraints, energy storage charging and discharging depth and number limitations, renewable energy output limitations, and investment budget limitations.

[0162] In summary, this step constructs a multi-objective optimization model containing 5 objective functions and multiple constraints, which provides a basis for the subsequent optimal energy storage configuration.

[0163] The specific implementation of step S80 is as follows: The purpose of this step is to use a multi-objective optimization algorithm based on GPU acceleration to solve the multi-objective optimization model constructed in step S70 and obtain a non-inferior solution set.

[0164] First, the objective function matrix is ​​calculated in parallel on the GPU and the constraint matrix .

[0165] Next, the non-dominated sorting algorithm implemented by GPU is used to solve the multi-objective optimization problem. The specific steps are as follows: 1. Dominance relationship matrix calculation: ; in, is the element of the dominating relationship matrix, and Respectively and A solution, For the This step calculates the dominance relationship between each solution.

[0166] 2. Dominance count calculation: ; in, For the The dominated count of a solution, is the total number of solutions. This step counts the number of times each solution is dominated by other solutions.

[0167] 3. Pareto front determination: ; This step determines the first-level Pareto front, which is the set of solutions with a dominated count of 0.

[0168] 4. Crowding distance calculation: ; in, For the The crowding distance of a solution, is the number of objective functions, and In the The function value of two adjacent solutions on the target, and Respectively This step calculates the crowding distance of each solution in the target space.

[0169] 5. Selection, Crossover and Mutation: Based on non-dominated sorting and crowding distance, selection, crossover and mutation operations are performed for iterative optimization.

[0170] 6. Termination condition check: Repeat steps 2-5 until the termination condition is reached and the final non-inferior solution set is obtained.

[0171] The algorithm makes full use of the parallel computing power of GPU and greatly improves the efficiency of multi-objective optimization. At the same time, the calculation of non-dominated sorting and crowding distance ensures the convergence and dispersion of the final solution set, meeting the requirements of multi-objective optimization.

[0172] In order to further better understand and implement the present invention, the following provides Example 2 of a specific application scenario of the present invention: A certain area under the jurisdiction of a certain power company in a certain city, whose distribution network consists of a 110kV substation and several 10kV distribution transformers, supplies about 20,000 households and commercial users. In recent years, the area has gradually connected to photovoltaic power generation and wind power generation systems, and the installed capacity of renewable energy is about 10MW. In order to smooth the volatility of renewable energy and improve the flexibility and reliability of power supply of the power grid, the power company intends to deploy an energy storage system in the area.

[0173] Based on the multi-objective optimization configuration method for energy storage in the substation area proposed in this invention, the power company conducted detailed data collection and modeling analysis on the substation area, and finally determined the optimal energy storage system configuration plan. The specific implementation process is as follows: 1. Data Collection and Preliminary Modeling 1. Run Data Collection The power company deployed a large number of monitoring equipment in the substation area and collected various operating data for one year, including: Table 1 Operation data collection of the substation area

[0174] Through one year of continuous data collection, the power company obtained comprehensive operating data of the substation.

[0175] 2. Initial response model establishment Based on the collected operation data, the power company established a response model covering all aspects of the substation, mainly including: (1) Power balance equation: ; in, is the total load power; is the system power loss; Input power to the grid; is the photovoltaic power generation power; is the wind power generation power; is the energy storage system power; is the error term.

[0176] (2) Voltage balance equation: ; in, is the node voltage; is the voltage on the power supply side; and Respectively resistance and reactance of the line segment; For the Current in the line segment; is the error term.

[0177] (3) Power loss equation: ; in, is the total power loss; is the number of system line segments; For the Segment line current; For the Segment line resistance; is the core loss coefficient; For the Segment line voltage; is the switching loss coefficient; is the switching frequency; For the Apparent power of the line segment; is the error term.

[0178] (4) Energy storage charge and discharge equation: ; in, for The state of charge at the moment, is the time interval; For charging efficiency; is the discharge efficiency; and are charging and discharging power respectively, and their value range is ; is the rated capacity of the energy storage system; is the error term.

[0179] (5) Photovoltaic power generation equation: ; in, is the photovoltaic power generation power; is the photovoltaic panel efficiency; is the photovoltaic panel area; is the solar irradiance; is the temperature coefficient; is the photovoltaic cell temperature; is the error term.

[0180] (6) Wind power generation equation: ; in, is the wind power generation power; is the air density; is the swept area of ​​the wind wheel; is the wind energy utilization coefficient; is the wind speed; , , are cut-in, rated and cut-out wind speeds respectively; is the rated power; is the error term.

[0181] (7) Load response equation: ; in, is the actual load power; is the reference load power; is the demand response sensitivity coefficient; For electricity prices; is the error term.

[0182] (8) Grid support equation: ; in, Provide support capacity for the grid; and are the maximum active and reactive output of the power grid respectively; and are the actual grid active and reactive input respectively; is the error term.

[0183] By fitting the above eight response equations, the power company established a preliminary operation model for the substation.

[0184] 3. Model validation and data augmentation In order to verify the accuracy of the established response model, the power company substituted the actual collected operation data into the model for calculation and compared it with the measured results. After repeated adjustments, the fitting accuracy of each response equation reached more than 95%.

[0185] In order to further enhance the model's adaptability to future uncertainties, power companies have adopted data augmentation technology to generate a richer simulation data set based on the original operating data. The specific methods include: (1) Random interpolation method: Randomly interpolate some missing points in the original data to generate new data samples.

[0186] (2) Rolling window method: Roll the time window of the original data to generate more time series data.

[0187] (3) Time series model: Train time series prediction models such as ARIMA based on the original data to generate simulated data.

[0188] Through the above methods, power companies have expanded the coverage of training data sets, including various possible situations such as different load levels, meteorological conditions, and electricity price changes. These expanded data, together with the original collected data, constitute the substation operation data collection and result data collection, providing a broad basis for subsequent optimization modeling.

[0189] 2. Optimization Modeling and Solution 1. Data organization and feature extraction The power company organizes the above operation data collection and result data collection into a multidimensional matrix ,in: OK Represents different time points ; List Represents different data types and parameters, such as , , , , , , , , , , , , , , , , , , , , , , , , , , wait; depth Represent different scenarios or augmented datasets.

[0190] Then, the power company uses this multidimensional matrix Tucker decomposition is performed to obtain the core tensor and the factor matrix , , .according to The size of the elements, the power companies extracted the most important , , features, construct a simplified feature matrix .

[0191] 2. Construction of multi-objective optimization model Based on the above feature matrix ,The power enterprise built a multi-objective optimization model with the following 5 objective functions: (1) Maximizing system energy conversion efficiency: ; in, for Output power at the moment, for The input power at the moment, is the total number of time steps. The objective function aims to maximize the energy conversion efficiency of the system.

[0192] (2) Minimize energy loss: ; in, for The objective function aims to minimize the total energy loss of the system.

[0193] (3) Maximizing the role of energy storage in supporting the power grid: ; in, for The objective function aims to maximize the support of the energy storage system to the power grid.

[0194] (4) Maximizing power supply reliability: ; in, for The power supply at the time, for The objective function aims to maximize the power supply reliability.

[0195] (5) Minimize total cost: ; in, is the investment cost, for The objective function aims to minimize the total cost.

[0196] At the same time, the power company also set the following constraints: Energy storage system capacity and power limitations: , ; Voltage and frequency limitations: , ; Load balancing constraints: ; Energy storage charge and discharge depth and frequency limits: , , ; Renewable energy output limits: , ; Investment budget constraints: .

[0197] 3. GPU-accelerated multi-objective optimization solution Based on the above multi-objective optimization model, the power company adopted a non-dominated sorting algorithm based on GPU acceleration to solve it. The main steps are as follows: (1) Parallel calculation of the objective function matrix on the GPU and the constraint matrix .

[0198] (2) Calculate the dominance relationship matrix , determine whether each solution is dominated by other solutions.

[0199] (3) Parallel calculation of the dominance count of each solution and the dominated set.

[0200] (4) Determine the Pareto frontier , that is, the set of solutions whose dominated count is 0.

[0201] (5) Calculate the crowding distance of each solution , reflects the distribution of solutions in the target space.

[0202] (6) Perform selection, crossover and mutation operations based on non-dominated sorting and crowding distance, and iteratively optimize.

[0203] (7) Repeat steps (2)-(6) until the termination condition is reached and the final set of non-inferior solutions is obtained.

[0204] The entire solution process was parallelized on the GPU, which greatly improved the computing efficiency. The power company finally obtained a set of alternative solutions containing 20 non-inferior solutions.

[0205] 3. Optimal Energy Storage Configuration Solution 1. Comprehensive evaluation of non-inferior solutions In order to select the best solution from the non-inferior solution set, the power company conducted a comprehensive evaluation of each non-inferior solution. The specific steps are as follows: (1) Normalize the five objective function values ​​so that their numerical ranges are unified to the interval [0, 1]: ; in, For the The comprehensive score of non-inferior solutions, is the weight of each objective function, and Respectively The minimum and maximum values ​​of the objective function among all non-inferior solutions.

[0206] (2) Select the comprehensive score The highest non-inferior solution is taken as the optimal solution .

[0207] 2. Optimal energy storage configuration solution After the above evaluation, the power company selected the following optimal energy storage configuration solution: Table 2 Optimal energy storage configuration scheme

[0208] The values ​​of this scheme on the five objective functions are: , , , , , comprehensive score , is the highest.

Claims

1. A multi-objective optimization configuration method for energy storage in a substation to improve the source-load carrying capacity, characterized in that: The following steps are involved: S10, continuously collect the operation data of the substation, including load data, distribution network input data, photovoltaic power generation data, wind power generation data and energy storage data; S20, establishing a group of substation operation response equations according to the operation data, including a power balance equation, a voltage balance equation, a power loss equation, an energy storage charging and discharging equation, a photovoltaic power generation equation, a wind power generation equation, a load response equation, and a power grid support equation; S30, fitting the substation operation response equation group using the continuously collected substation operation data to obtain a first equation group; and inputting the operating data into the first set of equations to obtain a result set; S40, augmenting the collected operation data by using a data augmentation method to obtain augmented operation data, and bringing the augmented operation data into the first set of equations for calculation to obtain an augmented result set; Merging the operation data and the augmented operation data into an operation data collection, and merging the result set and the augmented result set into a result data collection; S50, establishing a multidimensional matrix according to the operation data collection and the result data collection; Dividing the multidimensional matrix into an input matrix X and an output matrix Y, wherein X includes all input parameters and Y includes all output parameters related to the optimization target; performing standardization on X and Y to eliminate the influence of different dimensions; S60, using tensor decomposition technology to perform dimension reduction and feature extraction on the multidimensional matrix to obtain a simplified feature matrix F; S70, constructing a multi-objective optimization model based on the feature matrix F; S80, using a multi-objective optimization algorithm based on GPU acceleration to solve the multi-objective optimization model to obtain a non-inferior solution set; S90, calculating the weighted score of each non-inferior solution in the non-inferior solution set, selecting the non-inferior solution with the highest weighted score as the optimal solution, and outputting the energy storage data corresponding to the optimal solution as the optimized configuration data of the substation energy storage.

2. A multi-objective optimization configuration method for energy storage in a substation for improving source-load carrying capacity according to claim 1, characterized in that: in, The frequency of data collection is 1 minute to 15 minutes.

3. According to claim 1, a multi-objective optimization configuration method for energy storage in a substation for improving source-load carrying capacity is characterized in that: The load data specifically includes active power, reactive power, voltage, current, and power factor.

4. A multi-objective optimization configuration method for energy storage in a substation for improving source-load carrying capacity according to claim 1, characterized in that: The distribution network input data specifically includes feeder voltage, current, active power, reactive power, and power factor.

5. The multi-objective optimization configuration method for energy storage in a substation for improving source-load carrying capacity according to claim 1 is characterized in that: The photovoltaic power generation data specifically includes output active power, reactive power, voltage, current, irradiance, and temperature.

6. A multi-objective optimization configuration method for energy storage in a substation for improving source-load carrying capacity according to claim 1, characterized in that: The wind power generation data specifically includes output active power, reactive power, voltage, current, wind speed, and wind direction.

7. A multi-objective optimization configuration method for energy storage in a substation for improving source-load carrying capacity according to claim 1, characterized in that: The energy storage data specifically includes charging and discharging power, state of charge, voltage, current, and temperature.

8. A multi-objective optimization configuration method for energy storage in a substation for improving source-load carrying capacity according to claim 7, characterized in that: The rows of the multidimensional matrix represent different time points, the columns represent different data types and parameters, and the depth represents different scenes or augmented data sets.

9. A multi-objective optimization configuration method for energy storage in a substation for improving source-load carrying capacity according to claim 1, characterized in that: The step of using tensor decomposition technology to reduce the dimension and extract features of the multidimensional matrix to obtain a simplified feature matrix F is specifically: using Tucker decomposition method to decompose the multidimensional matrix into a core tensor and a factor matrix; using the element size of the core tensor to identify important features and correlations; Based on the decomposition results, a simplified feature matrix F is constructed to retain important feature information.

10. A multi-objective optimization configuration method for energy storage in a substation for improving source-load carrying capacity according to claim 1, characterized in that: The step of constructing a multi-objective optimization model based on the characteristic matrix F is specifically as follows: using the energy storage data as a decision variable vector, constructing an objective function matrix, including: an objective function for maximizing system energy conversion efficiency, an objective function for minimizing energy loss, an objective function for maximizing the support role of energy storage for the power grid, an objective function for maximizing power supply reliability, and an objective function for minimizing total cost; setting a constraint matrix, including energy storage capacity and power limitations, voltage and frequency limitations, load balance constraints, energy storage charging and discharging depth and number limitations, renewable energy output limitations, and investment budget limitations.

Images