Power distribution network energy storage collaborative planning method and system based on space-time clustering and convex optimization

Abstract

The invention belongs to the technical field of power distribution network optimization, and particularly discloses a power distribution network energy storage collaborative planning method and system based on space-time clustering and convex optimization, and the method comprises the steps: collecting the multi-dimensional data of a power distribution network; clustering the photovoltaic output data of the multi-dimensional data, and determining the number of typical scenes to construct typical output scenes; constructing a convex optimization model based on the multi-dimensional data and a typical output scene, and solving the convex optimization model to obtain an evaluation result of the bearing capacity of the power distribution area; and generating an energy storage investment scheme by using a double-layer optimization model, and continuously optimizing and updating the energy storage investment scheme in combination with the evaluation result to obtain a globally optimal energy storage investment scheme. According to the invention, the calculation efficiency and accuracy of the energy storage system can be improved.

Description

Technical Field

[0001] This application belongs to the technical field of distribution network optimization. More specifically, it relates to a collaborative planning method and system for distribution network energy storage based on spatio-temporal clustering and convex optimization. Background Technique

[0002] With the continuous increase in the penetration rate of distributed photovoltaic power generation systems in the power grid, energy storage devices, as the core elements to support the stable operation of the new power system, have become increasingly prominent in their strategic position. In the intelligent power grid architecture, the energy storage system, through precise peak shaving and valley filling control strategies, effectively suppresses the output fluctuations of distributed power sources, and plays an irreplaceable role in ensuring power grid frequency stability, voltage quality, and power supply reliability. Especially in the scenario of high-proportion renewable energy access, energy storage devices have become the key infrastructure to enhance the operation resilience of the distribution network and achieve coordinated optimization of the power source, grid, load, and energy storage.

[0003] However, the existing energy storage systems face dual technical bottlenecks in engineering applications: Firstly, in a complex and changeable operating environment, the time-varying characteristics of the energy storage charge and discharge strategies are strongly coupled with the multi-objective optimization requirements, resulting in essential difficulties in the mathematical modeling of the global optimal control strategy. Secondly, the quantitative correlation mechanism between energy storage capacity configuration and the improvement of the bearing capacity of the distribution transformer area has not been fully decoupled. Its evaluation process involves multi-period dynamic power flow calculation of the distribution network with distributed power sources, and the inherent non-convex and non-linear characteristics of this mathematical model make traditional analytical methods difficult to apply, resulting in low solution efficiency of the model.

[0004] Therefore, how to improve the calculation efficiency of the distribution network energy storage system is a technical problem that urgently needs to be solved at present. Summary of the Invention

[0005] Aiming at the defects of the existing technology, the purpose of this application is to provide a collaborative planning method and system for distribution network energy storage based on spatio-temporal clustering and convex optimization, aiming to solve the problem of low solution efficiency of the distribution network energy storage system in the existing technology.

[0006] To achieve the above purpose, in the first aspect, this application provides a collaborative planning method for distribution network energy storage based on spatio-temporal clustering and convex optimization, including: Collect multi-dimensional data of the distribution network; Cluster the photovoltaic output data of the multi-dimensional data to determine the number of typical scenarios and construct typical output scenarios; Construct a convex optimization model based on the multi-dimensional data and typical output scenarios, solve the convex optimization model, and obtain the evaluation result of the bearing capacity of the distribution transformer area; Use a two-layer optimization model to generate an energy storage investment plan and continuously optimize and update the energy storage investment plan in combination with the evaluation result to obtain a globally optimal energy storage investment plan.

[0007] Optionally, the method for constructing the typical output scenario includes: Segment the photovoltaic output data on a daily basis to form multiple daily vectors, and select a target number of target vectors as the initial cluster centers, where the dimensions of the daily vectors and the target vectors are the same; Use the k-means clustering method to calculate the Euclidean distance between each daily vector and the initial cluster centers; According to the Euclidean distance, assign each daily vector to the nearest cluster center to update the cluster centers, so that the cluster centers are the means of all points within the clusters; Repeat the assignment and update steps until the cluster centers no longer change or reach the set number of iterations to obtain the final cluster centers, cluster assignment scheme, and error value; Gradually increase the number of clusters, obtain the change information of the error value, and select the point where the first error value begins to level off as the number of typical scenarios; Construct the photovoltaic typical output scenario for each node based on the number of typical scenarios and the photovoltaic capacity connected to each node.

[0008] Optionally, the method for constructing the convex optimization model includes: Determine the basic equations for power flow calculation of multi-dimensional data, where the basic equations for power flow calculation include: active power equation, reactive power equation, node voltage equation, and non-linear quadratic equation; Perform linear processing and second-order cone constraint processing on the non-linear quadratic equation to form cone constraint conditions that can be solved by convex optimization, and obtain the convex optimization model.

[0009] Optionally, it further includes: Determine the operation constraints of the distribution network energy storage device, and constrain the voltage fluctuation and power flow feedback ratio through a linearization method; where the operation constraints include: dynamic update constraint of the energy storage power, upper and lower limits constraint of the power, charge and discharge power limit constraint, and node voltage fluctuation range constraint.

[0010] Optionally, solving the convex optimization model to obtain the evaluation result of the bearing capacity of the distribution substation area includes: Taking the minimization of the comprehensive index of voltage fluctuation and power flow feedback as the objective, combined with the weight coefficient determined by the analytic hierarchy process and the weight of the typical scenario, establish an objective function for multi-scenario optimization; Solve the objective function to obtain the bearing capacity evaluation results of the maximum voltage volatility and the maximum feedback ratio.

[0011] Optionally, the double-layer optimization model includes an upper-layer sub-model and a lower-layer sub-model; The upper-layer sub-model is used to minimize the investment cost and scheduling weight of the energy storage capacity, and set the total investment limit of the energy storage capacity; The lower-level sub-model is used to minimize the operating cost under the operating constraints.

[0012] Optionally, the method for generating a globally optimal energy storage investment plan includes: Generating an energy storage investment plan using the upper-level sub-model; Using the lower-level sub-model to generate cut planes by minimizing the comprehensive index of voltage fluctuation and power flow feedback, and feeding the cut planes back to the upper-level sub-model; Updating the energy storage investment plan by the upper-level sub-model according to the cut planes, and gradually narrowing the upper bound of the investment cost; Verifying the updated energy storage investment plan by the lower-level sub-model and generating new cut planes, and dynamically adjusting the lower bound to approach the optimal solution; Terminating the optimization when the difference between the upper bound and the lower bound reaches a preset convergence threshold or the maximum number of iterations is reached, and outputting a globally optimal energy storage investment plan that satisfies the investment and operating constraints.

[0013] Optionally, the collection of multi-dimensional data of the distribution network includes: Collecting grid topology parameters, node types, historical load data, and distributed photovoltaic output data to establish a distribution network global information database.

[0014] In a second aspect, the present application further provides a distribution network energy storage collaborative planning system based on spatio-temporal clustering and convex optimization, including: A collection module for collecting multi-dimensional data of the distribution network; A clustering module for clustering the photovoltaic output data of the multi-dimensional data to determine the number of typical scenarios and construct typical output scenarios; An evaluation module for constructing a convex optimization model based on the multi-dimensional data and typical output scenarios, solving the convex optimization model, and obtaining an evaluation result of the bearing capacity of the distribution transformer area; An optimization module for generating an energy storage investment plan using a two-layer optimization model and continuously optimizing and updating the energy storage investment plan in combination with the evaluation result to obtain a globally optimal energy storage investment plan.

[0015] In a third aspect, the present application provides an electronic device, including: at least one memory for storing a program; at least one processor for executing the program stored in the memory, and when the program stored in the memory is executed, the processor is used to execute the method described in the first aspect or any possible implementation manner of the first aspect.

[0016] In a fourth aspect, the present application provides a computer-readable storage medium storing a computer program, and when the computer program runs on a processor, it causes the processor to execute the method described in the first aspect or any possible implementation manner of the first aspect.

[0017] In a fifth aspect, the present application provides a computer program product, which, when running on a processor, causes the processor to execute the method described in the first aspect or any possible implementation of the first aspect.

[0018] It can be understood that for the beneficial effects of the above second aspect to fifth aspect, reference can be made to the relevant descriptions in the first aspect above, which will not be elaborated here.

[0019] Generally speaking, compared with the prior art, the above technical solutions conceived by the present application have the following beneficial effects: (1) By clustering multi-dimensional data, the present application realizes the construction of typical scenarios for photovoltaic output data, which can effectively reduce the amount of data to be processed. Using the clustering method to compress complex daily vector data into a limited number of typical scenarios enables subsequent optimization calculations to be carried out on a smaller scale, thus significantly improving the calculation speed. By adopting a two-layer model design, the solution of the problem is made more efficient, and independent optimization can be carried out at each level, thereby reducing the overall computational complexity and greatly enhancing the computational efficiency. This not only effectively saves time costs but also makes the entire workflow more efficient and smooth, contributing to the rapid progress of related projects and meeting the high requirements for timeliness in practical applications.

[0020] (2) By combining the weights of typical scenarios and the weight coefficients determined by the analytic hierarchy process, the present application can more accurately reflect the actual situation. The optimization method based on multiple scenarios of the present application can consider the system behavior under different circumstances, thereby improving the accuracy of the evaluation results.

[0021] (3) The present application is excellent in ensuring the stability of calculation results. Through innovative technical means and optimized algorithm mechanisms, it can more significantly enhance the stability of calculation results. This greatly reduces the situations of repeated calculation and frequent adjustment caused by poor stability, ensures the reliability of calculation results, provides a stable and accurate data basis for subsequent decision-making, planning, etc. based on the results, and effectively avoids many inconveniences and resource waste caused by unstable results. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 is one of the schematic flowcharts of the distribution network energy storage collaborative planning method based on spatio-temporal clustering and convex optimization provided by an embodiment of the present application; Figure 2 is another schematic flowchart of the distribution network energy storage collaborative planning method based on spatio-temporal clustering and convex optimization provided by an embodiment of the present application; Figure 3 is a schematic diagram of the curve of the total squared error varying with the k value in an embodiment of the present application; Figure 4 is a schematic diagram of a typical scenario under the corresponding value in an embodiment of the present application; Figure 5 is a schematic diagram of the time and energy storage discharge power curve in an embodiment of the present application; Figure 6 is a schematic diagram of the time and energy storage energy state curve in an embodiment of the present application; Figure 7 is a schematic diagram of the structure of a distribution network energy storage collaborative planning system based on spatio-temporal clustering and convex optimization provided in an embodiment of the present application; Figure 8 is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. Detailed implementation manners

[0023] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0024] The term "and / or" in this document is a relational term describing associated objects, indicating that three relationships may exist. For example, A and / or B may represent: A exists alone, A and B exist simultaneously, and B exists alone. The symbol " / " in this document represents an "or" relationship between associated objects. For example, A / B represents A or B.

[0025] The terms "first", "second", etc. in the description and claims of this document are used to distinguish different objects, rather than to describe a specific order of the objects. For example, the first response message and the second response message are used to distinguish different response messages, rather than to describe the specific order of the response messages.

[0026] In the embodiments of the present application, words such as "exemplary" or "for example" are used to represent examples, illustrations or explanations. Any embodiment or design solution described as "exemplary" or "for example" in the embodiments of the present application should not be construed as being more preferred or having more advantages than other embodiments or design solutions. Rather, the use of words such as "exemplary" or "for example" is intended to present relevant concepts in a specific manner.

[0027] In the description of the embodiments of the present application, unless otherwise specified, the meaning of "a plurality of" refers to two or more. For example, a plurality of processing units refers to two or more processing units, etc.; a plurality of elements refers to two or more elements, etc.

[0028] The embodiments of the present application will be described below with reference to the accompanying drawings in the embodiments of the present application.

[0029] Refer toFigure 1 , this application provides a collaborative planning method for distribution network energy storage based on spatio-temporal clustering and convex optimization, including: S101. Collect multi-dimensional data of the distribution network; S102. Cluster the photovoltaic output data of the multi-dimensional data, determine the number of typical scenarios to construct typical output scenarios; S103. Construct a convex optimization model based on the multi-dimensional data and typical output scenarios, solve the convex optimization model, and obtain the evaluation result of the bearing capacity of the distribution transformer area; S104. Use a two-layer optimization model to generate an energy storage investment plan and continuously optimize and update the energy storage investment plan in combination with the evaluation result to obtain a globally optimal energy storage investment plan.

[0030] Specifically, the multi-dimensional data in the embodiments of this application includes, but is not limited to, grid topology parameters, node types, historical load data, and distributed photovoltaic output data to establish a distribution network global information database.

[0031] The acquisition methods of each dimension data are as follows: (1) Collect grid connection topology information and the resistance of each line and reactance , (2) Obtain the node types assigned to the nodes in the transformer area, including slack nodes, PV nodes, and PQ nodes; (3) Collect node historical load data, including the active power of the generator at each time point and reactive power as well as the active power and reactive power of the load; (4) Collect the access capacity of each distributed photovoltaic , and the historical photovoltaic output data of the distributed photovoltaic equipment with a capacity of in this transformer area.

[0032] Secondly, the photovoltaic output data of the multi-dimensional data is clustered through S102. In the data analysis stage, the photovoltaic output data is mainly clustered and analyzed. Through clustering, the performance of photovoltaic power generation can be divided into several typical scenarios. In some alternative implementation schemes, this step may include: data preprocessing: removing outliers and performing normalization processing; selecting a clustering algorithm: such as K-means, DBSCAN, etc., and the specific selection depends on the data characteristics; determining the number of typical scenarios: evaluating the clustering quality through methods such as the elbow method and the silhouette coefficient to determine a reasonable number of clusters. Finally, according to the clustering results, each typical scenario is selected to form a set of typical output scenarios, providing basic elements for subsequent modeling.

[0033] Further, based on the multi-dimensional data and typical output scenarios through S103, a convex optimization model is constructed. The goal of constructing the convex optimization model is to evaluate the carrying capacity of the distribution substation area, mainly including the following steps: Combining the collected multi-dimensional data and typical output scenarios, a mathematical model is established to describe the operating state of the power grid under different conditions; Define the objective function and constraints: The objective function can be to maximize the carrying capacity or minimize the cost, while considering constraints such as power balance, equipment stability, and safety; Use a suitable optimization algorithm to solve the model and obtain the bearing capacity evaluation results for each scenario.

[0034] Finally, through S104, a double-layer optimization model is used to generate and continuously optimize and update the energy storage investment plan. The process of continuously generating and optimizing the plan through the model in this application embodiment is as follows: Construct a double-layer optimization model, where the upper layer of the model optimizes for investment decisions, and the lower layer optimizes for power operation scheduling; Objectives and constraints: The upper-layer objective can be to maximize the investment return or minimize the cost, and the lower layer ensures that the energy storage system can effectively play its role; Dynamic optimization: Based on the feedback of actual operation data, continuously update and optimize the model, and seek the global optimal solution through an iterative algorithm, including the energy storage capacity, investment timing, etc.

[0035] In this application, by clustering multi-dimensional data, the construction of typical scenarios for photovoltaic output data is realized, which can effectively reduce the amount of data to be processed. The clustering method is used to compress complex daily vector data into a limited number of typical scenarios, enabling subsequent optimization calculations to be carried out on a smaller scale, thereby significantly improving the calculation speed. By adopting a double-layer model design, the solution of the problem is more efficient, and independent optimization can be carried out at each level, thereby reducing the overall computational complexity and greatly improving the computational efficiency. It not only effectively saves time costs but also makes the entire workflow more efficient and smooth, helping to quickly promote the progress of related projects and meet the high requirements for timeliness in practical applications.

[0036] Optionally, the method for constructing the typical output scenario includes: The photovoltaic output data is segmented by day to form multiple daily vectors, and a target number of target vectors are selected as the initial cluster centers. The daily vectors and the target vectors have the same dimension; Use the k-means clustering method to calculate the Euclidean distance between each daily vector and the initial cluster center; According to the Euclidean distance, each daily vector is assigned to the nearest cluster center to update the cluster center so that the cluster center is the mean of all points within the cluster; Repeat the allocation and update steps until the cluster centers no longer change or the set number of iterations is reached to obtain the final cluster centers, cluster allocation scheme, and error value; Gradually increase the number of clusters, obtain the change information of the error value, and select the point where the first error value begins to level off as the number of typical scenarios; Construct the photovoltaic typical output scenarios for each node based on the number of typical scenarios and the installed photovoltaic capacity of each node.

[0037] Specifically, the embodiment of the present application uses the k-means algorithm and the elbow method to select the optimal number of typical scenarios value, and construct the typical photovoltaic output scenarios based on this. Abstract and generalize the complex photovoltaic historical output data while retaining its original characteristics.

[0038] (1) For the photovoltaic output data, segment it by day to form a number of daily vectors with a dimension of ; .

[0039] (2) Randomly initialize vectors with a dimension of n as the initial cluster centers; (3) Calculate the Euclidean distance from each vector to the two centroids. For vector j, the distance to cluster center k is:

[0040] (4) Allocate the daily vectors to the nearest cluster center, and update the center of each cluster to the mean of all points within the cluster; (5) Repeat the above allocation and update steps until the cluster centers no longer change or the iteration upper limit is reached to obtain the final cluster centers, cluster allocation, and relative error SEE; (6) Start from and gradually increase , observe the change of the final relative error, and select the point where the first relative error begins to level off as the appropriate number of typical scenarios ; (7) Construct the photovoltaic typical output scenarios for each node based on the installed photovoltaic capacity of each node.

[0041] The embodiment of the present application systematically analyzes the photovoltaic output data through the k-means clustering method to extract representative typical scenarios. This process not only helps to understand the change characteristics of photovoltaic power generation, but also provides important data support for power grid scheduling, energy storage investment, etc. Through reasonable selection of the number of clusters and the clustering process, it can be ensured that the extracted photovoltaic typical output scenarios can truly reflect the actual operation situation, thereby optimizing the operation and management of the power system.

[0042] Optionally, the method for constructing the convex optimization model includes: Determine the basic power flow calculation equations for multi-dimensional data, where the basic power flow calculation equations include: active power equation, reactive power equation, node voltage equation, and non-linear quadratic equation; Perform linear processing and second-order cone constraint processing on the non-linear quadratic equation to form cone constraint conditions that can be solved by convex optimization, thereby obtaining a convex optimization model.

[0043] In some embodiments, the process of constructing the convex model further includes: Determine the operation constraints of the distribution network energy storage device, and constrain the voltage fluctuation and power flow feedback ratio through a linearization method; wherein, the operation constraints include: dynamic update constraint of the energy storage power, upper and lower limit constraints of the power, charge and discharge power limit constraints, and node voltage fluctuation range constraints.

[0044] Optionally, solving the convex optimization model to obtain the evaluation result of the bearing capacity of the distribution transformer area includes: Taking the minimization of the comprehensive index of voltage fluctuation and power flow feedback as the objective, combining the weight coefficients determined by the analytic hierarchy process and the weights of typical scenarios, and establishing an objective function for multi-scenario optimization; Solve the objective function to obtain the bearing capacity evaluation results of the maximum voltage volatility and the maximum feedback ratio.

[0045] Specifically, in the embodiments of the present application, by establishing an optimization model based on mixed integer second-order cone programming (MISOCP), the traditional non-linear power flow calculation equation is converted into a convex optimization equation, so that the power flow equation can be quickly and accurately solved by a commercial convex optimization solver.

[0046] (1) The basic power flow calculation equations are as follows:

[0047]

[0048]

[0049]

[0050]

[0051]

[0052] Among them, is the active power flow from node to node at time is The active power flow from node to node is For the reactive power flow from node to node is For the reactive power flow from node to node is The resistance of line is The reactance of line is The current of line at time is The voltage magnitude of node at time is The voltage of node at time is The voltage of node at time is The current magnitude of line at time is The current of line at time is

[0053] And are the active power and reactive power injected into the node respectively, and the calculation formulas are as follows:

[0054]

[0055] Where 、 represent the active power and reactive power of the load of node at time respectively, 、 The active power and reactive power of the generator of node at time respectively, represents the active power of the distributed PV of node at time, represents the discharge power of the energy storage device of node at time.

[0056] Perform linear processing and second-order cone constraint processing on the basic power flow equations to form cone constraint conditions that can be solved by convex optimization; (2) In the basic power flow equations, the equation is non-linear. To convert the power flow solution problem into a convex optimization problem, formula 4 needs to be relaxed. Here, second-order cone relaxation is used to obtain:

[0057] (3) Considering the limitations of the energy storage capacity and its relationship with power, add dynamic update constraints and upper and lower limits constraints for the energy storage power:

[0058]

[0059] represents the energy storage power of node at time, represents the energy storage capacity of the energy storage device connected to node and is a number less than 1 representing the loss during the charge and discharge process.

[0060] (4) Considering that the charge and discharge power and voltage fluctuations are limited, add charge and discharge power limit constraints and node voltage fluctuation range constraints:

[0061]

[0062] (5) To evaluate the bearing capacity of the distribution substation area, consider two parameters, the maximum voltage fluctuation and the maximum power flow return ratio of the substation area. The calculation formula is:

[0063]

[0064] Among them, represents the standard voltage of the node, which is generally 1, represents the power injected into the power grid by the balance node connected to the upper-level power grid, represents the transformer capacity of this balance node.

[0065] (6) Since and are both non-linear in the calculation formula and cannot be solved, but can be converted by adding constraints, that is

[0066]

[0067] In the optimal case, it is obvious that the rewritten formula can satisfy the original formula, thus converting all calculation formulas into linear constraints. The optimization function is:

[0068] and The value of is determined by the analytic hierarchy process, which represents the degree of emphasis on the maximum voltage fluctuation and the maximum reverse power flow in the substation area. represents the typical scenario weight.

[0069] (7) Solve this convex optimization problem using a commercial convex optimization solver to obtain the bearing capacity evaluation result.

[0070] Optionally, the two-layer optimization model includes an upper-layer sub-model and a lower-layer sub-model; The upper-layer sub-model is used to minimize the investment cost and scheduling weight of the energy storage capacity and set the total investment limit of the energy storage capacity; The lower-layer sub-model is used to minimize the operating cost under the operating constraints.

[0071] Optionally, the method for generating a globally optimal energy storage investment plan includes: Generate an energy storage investment plan using the upper-layer sub-model; Use the lower-layer sub-model to generate cutting planes by minimizing the comprehensive index of voltage fluctuation and power flow feedback and feed them back to the upper-layer sub-model; Update the energy storage investment plan by the upper-layer sub-model according to the cutting planes, and gradually narrow the upper bound of the investment cost; Verify the updated energy storage investment plan by the lower-layer sub-model and generate new cutting planes to dynamically adjust the lower bound to approach the optimal solution; Terminate the optimization when the difference between the upper bound and the lower bound reaches the preset convergence threshold or the maximum number of iterations, and output the globally optimal energy storage investment plan that satisfies the investment and operation constraints.

[0072] Specifically, the embodiment of the present application uses Benders decomposition to construct a master-slave game optimization framework, divides an optimization problem into a master problem and a sub-problem, thereby reducing the constraint dimension and accelerating the solution speed.

[0073] It should be noted that "benders" refers to the Benders decomposition algorithm, which is an important algorithm for solving large-scale mixed-integer programming (MIP) problems and related optimization problems. Its core idea is to decompose a complex large-scale optimization problem into a master problem and one or more subproblems. By iteratively solving between the master problem and the subproblems, the optimal solution of the original problem is gradually approximated. For mixed-integer programming problems, the integer variable part is usually placed in the master problem, while the continuous variable part is placed in the subproblem. The master problem updates the values of the integer variables based on the information provided by the subproblems, and the subproblems solve the optimal solution of the continuous variables under the given values of the integer variables.

[0074] In problems such as unit commitment and power grid planning in the power system, the Benders decomposition algorithm can decompose large-scale mixed-integer programming problems into multiple smaller subproblems for solution, improve the computational efficiency, and help optimize the operation and planning of the power system.

[0075] The process of model establishment and the process of generating and updating solutions based on the model are as follows: (1) To accelerate the solution, a two-layer optimization master-slave game optimization framework for energy storage collaborative planning with upper-layer minimizing investment and lower-layer operation verification is constructed based on the Benders distribution solution strategy.

[0076] The upper-layer investment submodel is:

[0077]

[0078] Among them, is the cost of investing one unit of energy storage capacity at node , and is a parameter used to adjust the weight of the optimal dispatch .

[0079] (2) The lower-layer operation verification submodel is:

[0080] The lower-layer constraints are all the constraint formulas mentioned in the above embodiments.

[0081] (3) During the solution process, the upper-layer investment model generates investment plans for new energy storage, and the lower-layer model is responsible for verifying the feasibility of the master problem solution and generating cut planes to further optimize the solution.

[0082] Taking the optimal value of the upper-layer model obtained from the initial investment plan as , after iteration, if there is a smaller value in the investment function, taking it as , Update in rounds simultaneously . When after the maximum number of iterations there is still no update or the following conditions are met:

[0083] Stop the program, the bender solution ends, and the optimal new energy storage investment plan is obtained.

[0084] Refer to Figure 2 , Figure 2 is the complete process schematic diagram of this application, including the following steps: 1. Collect multi-dimensional data of the distribution network; 2. Perform kmeans clustering on the photovoltaic output data; 3. Draw the K-SEE graph to determine the number of typical scenarios; 4. Perform second-order cone relaxation on the non-linear power flow equation; 5. Add power flow constraints and equipment physical constraints; 6. Solve the convex optimization problem to obtain the bearing capacity assessment of the distribution substation area; 7. Construct a two-layer optimization model for energy storage collaborative planning; 8. Use benders distribution to solve.

[0085] 9. If the maximum number of iterations is not reached, return to step 8 and continue the iteration; 10. Check whether the conditions are met ; 11. If met, end the process.

[0086] The following describes the solution of this application in combination with specific embodiments: Step 1: Multi-dimensional data collection and feature modeling Grid topology parameter collection: By organizing a qualified operation and maintenance team and using professional detection equipment to obtain the line connection topology; Establish a line parameter matrix and record the measured values of the resistance (r) and reactance (x) of each branch; Construct a network topology diagram and mark the connection relationship between nodes and line parameters; Node type identification: Divide node types according to equipment configuration: Slack node: The voltage reference point connected to the superior grid; PV node: The photovoltaic access point equipped with a voltage regulating device; PQ node: The conventional load access point; Generate a node attribute table and enter it into the database; Time series data collection: Deploy intelligent measurement devices and set a fixed sampling period; Continuously collect the time-series data of generator output and load power; Record the nameplate capacity of photovoltaic equipment and its historical output curve.

[0087] Step 2: Spatiotemporal feature clustering and scenario generation Data preprocessing: Eliminate outliers using the 3σ criterion and complement missing data with the mean value; Segment the photovoltaic output data by natural day and construct daily feature vectors; Normalization processing: Perform per-unit conversion according to the equipment capacity; Calculate the Euclidean distance between the daily vectors and the cluster centers according to the k-means clustering method; Execute the cluster assignment and centroid update operations; Refer to Figure 3 , Figure 3 which is a schematic diagram of the curve of the sum of squared errors (SEE) varying with the k value; As can be seen from the figure, determine the optimal number of scenarios when the decline rate of SEE significantly decreases, and select ; Refer to Figure 4 , Figure 4 which is a schematic diagram of the typical scenario at the corresponding value; Statistically analyze the occurrence frequencies of each scenario and calculate the scenario probability weights, as shown in Table 1 below: Table 1

[0088] Generate the typical output curves at the node level and establish the scenario-node mapping relationship.

[0089] Step 3: Construction of a convex optimization bearing capacity evaluation model Add power flow constraints:

[0090]

[0091]

[0092]

[0093]

[0094]

[0095] Implement second-order cone relaxation on the non-linear power flow equation:

[0096] Construction of Energy Storage Dynamic Equation Constraints:

[0097]

[0098] In the calculation example, take ; Operation Safety Constraints: Set the voltage fluctuation range and the threshold of the power fed back;

[0099]

[0100] Multi-objective Optimization:

[0101] Construct the weight matrix of the Analytic Hierarchy Process (AHP) to determine and the value of. The analytic hierarchy matrix is:

[0102] It can be obtained that . After verification, the consistency check passes.

[0103] Step 4: Two-layer optimization of energy storage collaborative planning; Upper-layer planning model: Define the upper-layer energy storage investment cost function and set the node capacity constraint:

[0104]

[0105] Define the lower-layer operation verification function: Inherit the constraint system of the bearing capacity evaluation model;

[0106] Perform benders decomposition calculation and set the iteration limit to ; When the output meets the convergence condition, output the optimal planning scheme. According to the calculation of the IEEE33-node sample example, the best scheme is shown in Table 2 below: Table 2

[0107] Refer to Figure 5 and Figure 6 , Figure 5 is the schematic diagram of the time and the energy storage discharge power curve, Figure 6 is the schematic diagram of the time and the energy storage energy state curve. From Figure 5It can be seen that the power of different energy storage nodes shows different change trends at different times. The power change trends of different energy storage nodes vary. Some nodes have large power fluctuations during specific time periods, which are related to the charge-discharge strategies of the energy storage system and the load demand of the power grid. From Figure 6 It can be seen that the energy states of different energy storage nodes show different change trends at different times. The energy state change trends of different energy storage nodes vary. Some nodes have large energy state fluctuations during specific time periods, Figure 6 which reflects the charge-discharge conditions of the energy storage system at different times and the changes in the energy storage level.

[0108] Referring to Figure 7 , this application also provides a distribution network energy storage collaborative planning system based on spatio-temporal clustering and convex optimization, including: A collection module 710 for collecting multi-dimensional data of the distribution network; A clustering module 720 for clustering the photovoltaic output data of the multi-dimensional data to determine the number of typical scenarios and construct typical output scenarios; An evaluation module 730 for constructing a convex optimization model based on the multi-dimensional data and the typical output scenarios, solving the convex optimization model, and obtaining the evaluation result of the bearing capacity of the distribution transformer area; An optimization module 740 for generating an energy storage investment plan using a two-layer optimization model and continuously optimizing and updating the energy storage investment plan in combination with the evaluation result to obtain a globally optimal energy storage investment plan.

[0109] Optionally, the method for constructing the typical output scenario includes: Dividing the photovoltaic output data by day to form multiple daily vectors, and selecting a target number of target vectors as the initial cluster centers. The daily vectors and the target vectors have the same dimension; Using the k-means clustering method to calculate the Euclidean distance between each daily vector and the initial cluster centers; According to the Euclidean distance, allocating each daily vector to the nearest cluster center to update the cluster centers so that the cluster centers are the means of all points within the clusters; Repeating the allocation and update steps until the cluster centers no longer change or reach the set number of iterations to obtain the final cluster centers, cluster allocation scheme, and error value; Gradually increasing the number of clusters, obtaining the change information of the error value, and selecting the point where the first error value begins to flatten out as the number of typical scenarios; Based on the number of typical scenarios and the photovoltaic capacity connected to each node, constructing the photovoltaic typical output scenario for each node.

[0110] Optionally, the method for constructing the convex optimization model includes: Determine the basic equations for power flow calculation of multi-dimensional data, where the basic equations for power flow calculation include: active power equation, reactive power equation, node voltage equation, and non-linear quadratic equation; Perform linear processing and second-order cone constraint processing on the non-linear quadratic equation to form a cone constraint condition that can be solved by convex optimization, and obtain a convex optimization model.

[0111] Optionally, it further includes: Determine the operation constraints of the distribution network energy storage device, and constrain the voltage fluctuation and power flow feedback ratio through a linearization method; where the operation constraints include: dynamic update constraint of energy storage power, upper and lower limits constraint of power, charge and discharge power limit constraint, and node voltage fluctuation range constraint.

[0112] Optionally, solving the convex optimization model to obtain the evaluation result of the bearing capacity of the distribution transformer area includes: Taking the goal of minimizing the comprehensive index of voltage fluctuation and power flow feedback, combining the weight coefficients determined by the analytic hierarchy process and the weights of typical scenarios, and establishing an objective function for multi-scenario optimization; Solve the objective function to obtain the bearing capacity evaluation results of the maximum voltage volatility and the maximum feedback ratio.

[0113] Optionally, the two-layer optimization model includes an upper-layer sub-model and a lower-layer sub-model; The upper-layer sub-model is used to minimize the investment cost and scheduling weight of the energy storage capacity, and set the total investment limit of the energy storage capacity; The lower-layer sub-model is used to minimize the operation cost under the operation constraints.

[0114] Optionally, the method for generating a globally optimal energy storage investment plan includes: Generate an energy storage investment plan using the upper-layer sub-model; Use the lower-layer sub-model to generate a cut plane by minimizing the comprehensive index of voltage fluctuation and power flow feedback and feed it back to the upper-layer sub-model; Update the energy storage investment plan by the upper-layer sub-model according to the cut plane, and gradually narrow the upper bound of the investment cost; Verify the updated energy storage investment plan by the lower-layer sub-model and generate a new cut plane, and dynamically adjust the lower bound to approach the optimal solution; Terminate the optimization when the difference between the upper bound and the lower bound reaches the preset convergence threshold or reaches the maximum number of iterations, and output a globally optimal energy storage investment plan that satisfies the investment and operation constraints.

[0115] Optionally, collecting the multi-dimensional data of the distribution network includes: Collect grid topology parameters, node types, historical load data, and distributed PV output data to establish a global information database for the distribution network.

[0116] It can be understood that the detailed function implementation of each of the above units / modules can be referred to the introduction in the foregoing method embodiments, and will not be elaborated here.

[0117] It should be understood that the above system is used to execute the method in the above embodiment. For the corresponding program modules in the system, their implementation principles and technical effects are similar to those described in the above method. The working process of this system can refer to the corresponding process in the above method, and will not be elaborated here.

[0118] Refer to Figure 8 , based on the method in the above embodiment, an embodiment of the present application provides an electronic device, which may include: a processor (Processor) 810, a communication interface (Communications Interface) 820, a memory (Memory) 830, and a communication bus 840. Among them, the processor 810, the communication interface 820, and the memory 830 communicate with each other through the communication bus 840. The processor 810 can call the logical instructions in the memory 830 to execute the method in the above embodiment.

[0119] In addition, when the logical instructions in the above memory 830 are implemented in the form of a software functional unit and sold or used as an independent product, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or a part of this 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 for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application.

[0120] Based on the method in the above embodiment, an embodiment of the present application provides a computer-readable storage medium, which stores a computer program. When the computer program runs on a processor, it causes the processor to execute the method in the above embodiment.

[0121] Based on the method in the above embodiment, an embodiment of the present application provides a computer program product. When the computer program product runs on a processor, it causes the processor to execute the method in the above embodiment.

[0122] It can be understood that the processor in the embodiments of the present application may be a central processing unit (CPU), or may also be 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. The general-purpose processor may be a microprocessor or any conventional processor.

[0123] The method steps in the embodiments of the present application may be implemented in a hardware manner or by a processor executing software instructions. The software instructions may be composed of corresponding software modules, and the software modules may be stored in a random access memory (RAM), flash memory, read-only memory (ROM), programmable ROM (PROM), erasable PROM (EPROM), electrically erasable PROM (EEPROM), registers, hard disks, removable hard disks, CD-ROMs, or any other form of storage medium well-known in the art. An exemplary storage medium is coupled to the processor so that the processor can read information from the storage medium and write information to the storage medium. Of course, the storage medium may also be a component of the processor. The processor and the storage medium may be located in the ASIC.

[0124] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of 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, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. 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 in a wired manner (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or a wireless manner (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or a data center that includes one or more integrated available media. The available medium can be a magnetic medium (such as a floppy disk, a hard disk, a magnetic tape), an optical medium (such as a DVD), or a semiconductor medium (such as a solid state disk (SSD)), etc.

[0125] It can be understood that the various numerical numbers involved in the embodiments of the present application are only for the convenience of description and are not used to limit the scope of the embodiments of the present application.

[0126] Those skilled in the art can easily understand that the above is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent replacements, and improvements made within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A distribution network energy storage collaborative planning method based on spatiotemporal clustering and convex optimization, characterized in that: include: Collect multi-dimensional data of distribution network; Cluster the photovoltaic output data of multi-dimensional data, determine the number of typical scenarios to construct typical output scenarios; Constructing a convex optimization model based on the multi-dimensional data and typical output scenarios, solving the convex optimization model, and obtaining an evaluation result of the carrying capacity of the distribution station area; The energy storage investment plan is generated by using the double-layer optimization model and is continuously optimized and updated in combination with the evaluation results to obtain the globally optimal energy storage investment plan.

2. The distribution network energy storage collaborative planning method based on spatiotemporal clustering and convex optimization according to claim 1 is characterized in that: The method for constructing the typical output scenario includes: The photovoltaic output data is divided into a day unit to form a plurality of daily vectors, and a target number of target vectors are selected as initial cluster centers, wherein the daily vector and the target vector have the same dimension; The k-means clustering method was used to calculate the Euclidean distance between each daily vector and the initial cluster center; According to the Euclidean distance, each daily vector is assigned to the nearest cluster center to update the cluster center so that the cluster center is the mean of all points in the cluster; Repeat the allocation and update steps until the cluster center no longer changes or the set number of iterations is reached to obtain the final cluster center, cluster allocation scheme, and error value; Gradually increase the number of clusters, obtain the change information of the error value, and select the first point where the error value begins to flatten out as the number of typical scenes; Based on the number of typical scenarios and the connected photovoltaic capacity of each node, a typical photovoltaic output scenario for each node is constructed.

3. The distribution network energy storage collaborative planning method based on spatiotemporal clustering and convex optimization according to claim 1 is characterized in that: The method for constructing the convex optimization model includes: Determine a basic equation for power flow calculation of multi-dimensional data, wherein the basic equation for power flow calculation includes: an active power equation, a reactive power equation, a node voltage equation, and a nonlinear quadratic equation; The nonlinear quadratic equation is subjected to linear processing and second-order cone constraint processing to form cone constraint conditions solvable by convex optimization, thereby obtaining a convex optimization model.

4. The distribution network energy storage collaborative planning method based on spatiotemporal clustering and convex optimization according to claim 3 is characterized in that: Also includes: Determine the operating constraints of the distribution network energy storage equipment, and constrain the voltage fluctuation and power flow return ratio through a linearization method; wherein the operating constraints include: dynamic update constraints of energy storage power, upper and lower limit constraints of power, charge and discharge power limit constraints, and node voltage fluctuation range constraints.

5. The distribution network energy storage collaborative planning method based on spatiotemporal clustering and convex optimization according to claim 1 is characterized in that: The convex optimization model is solved to obtain the evaluation result of the distribution station area bearing capacity, including: Taking the comprehensive index of minimizing voltage fluctuation and power flow return as the goal, the objective function of multi-scenario optimization is established by combining the weight coefficients determined by the hierarchical analysis method and the weights of typical scenarios. The objective function is solved to obtain the carrying capacity evaluation results of the maximum voltage fluctuation rate and the maximum return ratio.

6. The distribution network energy storage collaborative planning method based on spatiotemporal clustering and convex optimization according to claim 4 is characterized in that: The two-layer optimization model includes an upper layer sub-model and a lower layer sub-model; The upper sub-model is used to minimize the investment cost and dispatch weight of energy storage capacity and set a total investment limit for energy storage capacity; The lower layer sub-model is used to minimize the operation cost under the operation constraints.

7. The distribution network energy storage collaborative planning method based on spatiotemporal clustering and convex optimization according to claim 6 is characterized in that: The method for generating the global optimal energy storage investment plan includes: Generate an energy storage investment plan using the upper sub-model; Using the lower layer sub-model to minimize the comprehensive index of voltage fluctuation and power flow return, a cutting plane is generated and fed back to the upper layer sub-model; Updating the energy storage investment plan according to the cutting plane through the upper sub-model to gradually reduce the upper bound of the investment cost; Verify the updated energy storage investment plan through the lower sub-model and generate a new cutting plane, dynamically adjust the lower bound to approach the optimal solution; When the difference between the upper bound and the lower bound reaches the preset convergence threshold or the maximum number of iterations is reached, the optimization is terminated and the globally optimal energy storage investment plan that meets the investment and operation constraints is output.

8. The distribution network energy storage collaborative planning method based on spatiotemporal clustering and convex optimization according to claim 7 is characterized in that: The multi-dimensional data of the distribution network is collected, including: Collect grid topology parameters, node types, historical load data and distributed photovoltaic output data to establish a distribution network global information database.

9. A distribution network energy storage collaborative planning system based on spatiotemporal clustering and convex optimization, characterized in that: include: The acquisition module is used to collect multi-dimensional data of the distribution network; A clustering module is used to cluster the photovoltaic output data of multi-dimensional data and determine the number of typical scenarios to construct a typical output scenario; An evaluation module, used to construct a convex optimization model based on the multi-dimensional data and typical output scenarios, solve the convex optimization model, and obtain an evaluation result of the carrying capacity of the distribution station area; The optimization module is used to generate an energy storage investment plan using a double-layer optimization model and continuously optimize and update the energy storage investment plan in combination with the evaluation results to obtain a globally optimal energy storage investment plan.

10. An electronic device, characterized in that: include: at least one memory for storing a computer program; At least one processor is used to execute the program stored in the memory. When the program stored in the memory is executed, the processor is used to execute the method according to any one of claims 1 to 8.

Images