Line loss optimization system and method based on distributed photovoltaic topology analysis

Abstract

The invention provides a line loss optimization system and method based on distributed photovoltaic topology analysis, and relates to the technical field of distributed photovoltaic, and the system comprises a topology connection module which is used for carrying out the topology connection, and building a distributed photovoltaic power distribution network topology model; the prediction modeling module is used for performing prediction modeling to obtain a photovoltaic output-power grid load space-time prediction model; the access position optimization module is used for carrying out photovoltaic access position optimization and obtaining a distributed photovoltaic optimization power distribution network topology model through iteration; the strategy analysis module is used for carrying out strategy analysis and determining power distribution network line loss control strategy parameters; and the feedback optimization module is used for carrying out simulation evaluation and carrying out feedback optimization on the power distribution network line loss control strategy parameters. The technical problem that the line loss control effect is poor because the line loss control strategy of the power distribution network in the prior art is often set based on experience and is not fully combined with the dynamic characteristics of photovoltaic power generation and load for real-time adjustment is solved.

Description

Technical Field

[0001] The present invention relates to the field of distributed photovoltaic technology, and in particular to a line loss optimization system and method based on distributed photovoltaic topology analysis. Background Art

[0002] Distributed photovoltaic power generation is one of the most widely used forms of clean energy in modern power systems. However, the intermittent, volatile and geographically dispersed nature of photovoltaic power generation has brought new challenges to the power grid. The problem of power grid line loss has become an important issue that needs to be optimized in the operation of the power grid. Specifically, the selection of photovoltaic system access points has a huge impact on the operation of the power grid. Improper access locations may cause electricity to be transmitted over long distances, resulting in a large amount of transmission losses. In particular, in low-voltage distribution networks, the problem of current loss is particularly prominent. In existing technologies, the line loss control strategy of the distribution network is often based on experience, and does not fully combine the dynamic characteristics of photovoltaic power generation and loads for real-time adjustment, resulting in poor line loss control and energy loss in the transmission line. Summary of the invention

[0003] This application provides a line loss optimization system and method based on distributed photovoltaic topology analysis, aiming to solve the technical problem that the line loss control strategy of the distribution network in the prior art is often set based on experience, and does not fully combine the dynamic characteristics of photovoltaic power generation and load for real-time adjustment, resulting in poor line loss control effect.

[0004] The first aspect disclosed in the present application provides a line loss optimization system based on distributed photovoltaic topology analysis, the system comprising: a topology connection module, the topology connection module is used to collect and obtain basic distribution data of the distribution network and photovoltaic access data, and define graph connection attributes, and topologically connect the basic distribution data of the distribution network and the photovoltaic access data based on the graph connection attributes to establish a distributed photovoltaic distribution network topology model; a prediction modeling module, the prediction modeling module is used to obtain a photovoltaic output data set and a power grid load data set through data mining technology, and use time series analysis to predict and model the photovoltaic output data set and the power grid load data set to obtain a photovoltaic output-power grid load spatiotemporal prediction model; an access location optimization module, the access location optimization module is used to obtain a photovoltaic output-power grid load spatiotemporal prediction model based on the The distributed photovoltaic distribution network topology model initializes the particle swarm parameters, optimizes the photovoltaic access position of the particle swarm parameters according to the line loss optimization target, and iteratively obtains the distributed photovoltaic optimized distribution network topology model; a strategy analysis module, the strategy analysis module is used to obtain the spatiotemporal distribution characteristic information of the photovoltaic load according to the photovoltaic output-grid load spatiotemporal prediction model, and performs strategy analysis based on the spatiotemporal distribution characteristic information of the photovoltaic load and the distributed photovoltaic optimized distribution network topology model to determine the distribution network line loss control strategy parameters; a feedback optimization module, the feedback optimization module is used to simulate and evaluate the distribution network line loss control strategy parameters, obtain the distribution network loss reduction simulation effect, and feedback optimize the distribution network line loss control strategy parameters based on the distribution network loss reduction simulation effect.

[0005] Furthermore, the establishment of a distributed photovoltaic distribution network topology model includes: According to the graph connection attributes, entity node attributes and node connection edge attributes are determined; based on the entity node attributes, entity recognition and attribute identification are performed on the distribution network basic distribution data and photovoltaic access data to obtain a distribution network entity set and an entity node attribute set; connection relationship analysis is performed on the distribution network entity set through the distribution network basic distribution data and photovoltaic access data to construct a node neighbor connection relationship set; topological abstraction is performed on the node neighbor connection relationship set based on the node connection edge attributes to obtain an entity node topological adjacency matrix; topological identification connection is performed on the distribution network entity set based on the entity node topological adjacency matrix and the entity node attribute set to establish the distributed photovoltaic distribution network topology model.

[0006] Furthermore, the step of obtaining a photovoltaic output-grid load spatiotemporal prediction model includes: The photovoltaic output data set and the grid load data set are arranged in time series and analyzed for characteristics by using time series analysis to obtain photovoltaic output characteristic data sets and grid load characteristic data sets; the photovoltaic output characteristic data sets and the grid load characteristic data sets are respectively trained for prediction by using an LSTM network to obtain a photovoltaic output prediction model and a grid load prediction model; the photovoltaic output prediction model and the grid load prediction model are connected in parallel to fuse the photovoltaic output prediction model and the grid load prediction model to generate an initial photovoltaic load spatiotemporal prediction model; the initial photovoltaic load spatiotemporal prediction model is verified and optimized by using a model optimizer to obtain the photovoltaic output-grid load spatiotemporal prediction model.

[0007] Furthermore, the iterative acquisition of the distributed photovoltaic optimized distribution network topology model includes: A line loss evaluation index is extracted for the line loss optimization target to obtain a line loss evaluation index set, and a line loss optimization effect fitness function is constructed according to the line loss evaluation index set; a parameter particle position and a parameter particle speed are determined according to the particle swarm parameters, and the fitness of the particle swarm parameters is evaluated using the line loss optimization effect fitness function to obtain a particle swarm fitness set; the parameter particle position and the parameter particle speed are iteratively updated based on the particle swarm fitness set until a preset termination condition is met, and the optimal parameter particle with the largest fitness is determined by optimization, and the optimal parameter particle includes a photovoltaic access optimization position; the distributed photovoltaic distribution network topology model is optimized and updated based on the photovoltaic access optimization position to obtain the distributed photovoltaic optimized distribution network topology model.

[0008] Furthermore, the determining of the distribution network line loss control strategy parameters includes: Based on the distributed photovoltaic optimized distribution network topology model, a power grid flow distribution impact analysis is performed to obtain power grid flow distribution impact parameters, wherein the power grid flow distribution impact parameters include line load impact parameters and power grid line loss impact parameters; a line loss control strategy library is constructed, and based on the photovoltaic load spatiotemporal distribution characteristic information and the power grid flow distribution impact parameters, a matching analysis is performed with the line loss control strategy library to obtain a distribution network line loss matching control strategy; based on the distribution network line loss matching control strategy, the photovoltaic load spatiotemporal distribution characteristic information and the power grid flow distribution impact parameters are selected by threshold division to obtain a line loss control strategy parameter selection threshold; the line loss optimization effect fitness function is used to perform global optimization within the line loss control strategy parameter selection threshold to determine the distribution network line loss control strategy parameters.

[0009] Furthermore, the determining of the distribution network line loss control strategy parameters includes: A plurality of strategy parameters are randomly selected within the line loss control strategy parameter selection threshold, and the plurality of strategy parameters are evaluated using the line loss optimization effect fitness function to obtain a plurality of parameter fitnesses; the line loss control strategy parameter selection threshold is approximated in an optimization region based on the plurality of parameter fitnesses to determine a local optimization region for strategy parameters; a parameter search step is set according to the local optimization region for strategy parameters; a strategy parameter search and evaluation is performed in the local optimization region for strategy parameters according to the parameter search step, and an optimization region is iteratively approximated according to the parameter search and evaluation results until a preset number of iterations are reached, and the distribution network line loss control strategy parameters are determined by comparing the fitnesses.

[0010] Furthermore, the feedback optimization of the distribution network line loss control strategy parameters based on the distribution network loss reduction simulation effect includes: Based on the distribution network loss reduction simulation effect, the distribution network line loss control strategy parameters are optimized in the direction of analysis to obtain parameter variation optimization rules; according to the parameter variation optimization rules, the distribution network line loss control strategy parameters are mutated and expanded to obtain a line loss control strategy parameter cluster; parameter comparison and optimization are performed within the line loss control strategy parameter cluster to obtain line loss control optimization strategy parameters, and distribution network line loss optimization control is performed through the line loss control optimization strategy parameters.

[0011] The second aspect disclosed in the present application provides a line loss optimization method based on distributed photovoltaic topology analysis, which is implemented by the above-mentioned line loss optimization system based on distributed photovoltaic topology analysis, and the method includes: collecting and acquiring basic distribution data of the distribution network and photovoltaic access data, and defining graph connection attributes, and topologically connecting the basic distribution data of the distribution network and the photovoltaic access data based on the graph connection attributes to establish a distributed photovoltaic distribution network topology model; obtaining a photovoltaic output data set and a power grid load data set by associating them through data mining technology, and using time series analysis to predict and model the photovoltaic output data set and the power grid load data set to obtain a photovoltaic output-power grid load spatiotemporal prediction model; Initialize particle swarm parameters based on the distributed photovoltaic distribution network topology model, optimize photovoltaic access positions for the particle swarm parameters according to the line loss optimization target, and iteratively obtain the distributed photovoltaic optimized distribution network topology model; obtain photovoltaic load spatiotemporal distribution characteristic information based on the photovoltaic output-grid load spatiotemporal prediction model, perform strategy analysis based on the photovoltaic load spatiotemporal distribution characteristic information and the distributed photovoltaic optimized distribution network topology model, and determine distribution network line loss control strategy parameters; perform simulation evaluation on the distribution network line loss control strategy parameters to obtain distribution network loss reduction simulation effect, and perform feedback optimization on the distribution network line loss control strategy parameters based on the distribution network loss reduction simulation effect.

[0012] One or more technical solutions provided in this application have at least the following technical effects or advantages: A distributed photovoltaic distribution network topology model is established. This model effectively integrates the basic distribution data of the distribution network and the photovoltaic access data to form a clear network structure. The graph theory topology connection method is used to accurately reflect the connection relationship between each node in the power grid, so that the impact of photovoltaic access on the power grid flow can be visualized and analyzed; data mining and time series analysis technology are used to construct a photovoltaic output-grid load spatiotemporal prediction model. This model integrates the spatiotemporal characteristics of photovoltaic power generation and the dynamic changes of grid load demand, and realizes the matching prediction of photovoltaic power generation and grid load demand; the particle swarm algorithm is used to optimize the photovoltaic access location, and the optimal location of the photovoltaic access point is iteratively adjusted based on the line loss optimization target, and the optimized distributed photovoltaic optimized distribution network topology model is formed. This process combines the grid topology structure and the prediction model, so that Photovoltaic access points can minimize line losses during grid operation, reduce losses during power transmission, and achieve reasonable allocation of grid resources and maximize power efficiency. Through comprehensive analysis of the spatiotemporal distribution characteristics of photovoltaic loads and combined with the distributed photovoltaic optimization distribution network topology model, effective distribution network line loss control strategy parameters can be formulated to respond to dynamic changes in photovoltaic output and grid load in real time, and further reduce line losses by reasonably scheduling photovoltaic power generation and adjusting power flow paths. Simulation evaluation and feedback optimization are carried out to verify the effectiveness of the formulated line loss control strategy through simulation, and optimize the parameters based on the simulation results. This process ensures the effectiveness and pertinence of the line loss control strategy, so that the strategy can be continuously adjusted and improved according to actual conditions, realizing closed-loop optimization and ensuring that line losses are minimized during grid operation.

[0013] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 A schematic diagram of the structure of a line loss optimization system based on distributed photovoltaic topology analysis provided in an embodiment of the present application; Figure 2 A schematic flow chart of a line loss optimization method based on distributed photovoltaic topology analysis provided in an embodiment of the present application.

[0015] Explanation of the reference numerals: topology connection module 10 , prediction modeling module 20 , access location optimization module 30 , strategy analysis module 40 , feedback optimization module 50 . DETAILED DESCRIPTION

[0016] The embodiments of the present application provide a line loss optimization system and method based on distributed photovoltaic topology analysis, thereby solving the technical problem that the line loss control strategy of the distribution network in the prior art is often set based on experience and does not fully combine the dynamic characteristics of photovoltaic power generation and load for real-time adjustment, resulting in poor line loss control effect.

[0017] After introducing the basic principles of the present application, various non-limiting implementation methods of the present application will be specifically introduced below in conjunction with the accompanying drawings of the specification. 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.

[0018] Embodiment 1, as Figure 1 As shown, the embodiment of the present application provides a line loss optimization system based on distributed photovoltaic topology analysis, and the system includes: A topological connection module 10, which is used to collect and obtain basic distribution data of the distribution network and photovoltaic access data, and define graph connection attributes, and topologically connect the basic distribution data of the distribution network and the photovoltaic access data based on the graph connection attributes to establish a distributed photovoltaic distribution network topology model; a prediction modeling module 20, which is used to obtain a photovoltaic output data set and a power grid load data set through data mining technology, and use time series analysis to predict and model the photovoltaic output data set and the power grid load data set to obtain a photovoltaic output-power grid load spatiotemporal prediction model; an access location optimization module 30, which is used to initialize a particle swarm based on the distributed photovoltaic distribution network topology model Parameters, optimize the photovoltaic access position of the particle swarm parameters according to the line loss optimization target, and iteratively obtain the distributed photovoltaic optimized distribution network topology model; a strategy analysis module 40, the strategy analysis module 40 is used to obtain the spatiotemporal distribution characteristic information of the photovoltaic load according to the photovoltaic output-grid load spatiotemporal prediction model, and perform strategy analysis based on the spatiotemporal distribution characteristic information of the photovoltaic load and the distributed photovoltaic optimized distribution network topology model to determine the distribution network line loss control strategy parameters; a feedback optimization module 50, the feedback optimization module 50 is used to simulate and evaluate the distribution network line loss control strategy parameters, obtain the distribution network loss reduction simulation effect, and feedback optimize the distribution network line loss control strategy parameters based on the distribution network loss reduction simulation effect.

[0019] Furthermore, the establishment of a distributed photovoltaic distribution network topology model includes: According to the graph connection attributes, entity node attributes and node connection edge attributes are determined; based on the entity node attributes, entity recognition and attribute identification are performed on the distribution network basic distribution data and photovoltaic access data to obtain a distribution network entity set and an entity node attribute set; connection relationship analysis is performed on the distribution network entity set through the distribution network basic distribution data and photovoltaic access data to construct a node neighbor connection relationship set; topological abstraction is performed on the node neighbor connection relationship set based on the node connection edge attributes to obtain an entity node topological adjacency matrix; topological identification connection is performed on the distribution network entity set based on the entity node topological adjacency matrix and the entity node attribute set to establish the distributed photovoltaic distribution network topology model.

[0020] Furthermore, the step of obtaining a photovoltaic output-grid load spatiotemporal prediction model includes: The photovoltaic output data set and the grid load data set are arranged in time series and analyzed for characteristics by using time series analysis to obtain photovoltaic output characteristic data sets and grid load characteristic data sets; the photovoltaic output characteristic data sets and the grid load characteristic data sets are respectively trained for prediction by using an LSTM network to obtain a photovoltaic output prediction model and a grid load prediction model; the photovoltaic output prediction model and the grid load prediction model are connected in parallel to fuse the photovoltaic output prediction model and the grid load prediction model to generate an initial photovoltaic load spatiotemporal prediction model; the initial photovoltaic load spatiotemporal prediction model is verified and optimized by using a model optimizer to obtain the photovoltaic output-grid load spatiotemporal prediction model.

[0021] Furthermore, the iterative acquisition of the distributed photovoltaic optimized distribution network topology model includes: A line loss evaluation index is extracted for the line loss optimization target to obtain a line loss evaluation index set, and a line loss optimization effect fitness function is constructed according to the line loss evaluation index set; a parameter particle position and a parameter particle speed are determined according to the particle swarm parameters, and the fitness of the particle swarm parameters is evaluated using the line loss optimization effect fitness function to obtain a particle swarm fitness set; the parameter particle position and the parameter particle speed are iteratively updated based on the particle swarm fitness set until a preset termination condition is met, and the optimal parameter particle with the largest fitness is determined by optimization, and the optimal parameter particle includes a photovoltaic access optimization position; the distributed photovoltaic distribution network topology model is optimized and updated based on the photovoltaic access optimization position to obtain the distributed photovoltaic optimized distribution network topology model.

[0022] Furthermore, the determining of the distribution network line loss control strategy parameters includes: Based on the distributed photovoltaic optimized distribution network topology model, a power grid flow distribution impact analysis is performed to obtain power grid flow distribution impact parameters, wherein the power grid flow distribution impact parameters include line load impact parameters and power grid line loss impact parameters; a line loss control strategy library is constructed, and based on the photovoltaic load spatiotemporal distribution characteristic information and the power grid flow distribution impact parameters, a matching analysis is performed with the line loss control strategy library to obtain a distribution network line loss matching control strategy; based on the distribution network line loss matching control strategy, the photovoltaic load spatiotemporal distribution characteristic information and the power grid flow distribution impact parameters are selected by threshold division to obtain a line loss control strategy parameter selection threshold; the line loss optimization effect fitness function is used to perform global optimization within the line loss control strategy parameter selection threshold to determine the distribution network line loss control strategy parameters.

[0023] Furthermore, the determining of the distribution network line loss control strategy parameters includes: A plurality of strategy parameters are randomly selected within the line loss control strategy parameter selection threshold, and the plurality of strategy parameters are evaluated using the line loss optimization effect fitness function to obtain a plurality of parameter fitnesses; the line loss control strategy parameter selection threshold is approximated in an optimization region based on the plurality of parameter fitnesses to determine a local optimization region for strategy parameters; a parameter search step is set according to the local optimization region for strategy parameters; a strategy parameter search and evaluation is performed in the local optimization region for strategy parameters according to the parameter search step, and an optimization region is iteratively approximated according to the parameter search and evaluation results until a preset number of iterations are reached, and the distribution network line loss control strategy parameters are determined by comparing the fitnesses.

[0024] Furthermore, the feedback optimization of the distribution network line loss control strategy parameters based on the distribution network loss reduction simulation effect includes: Based on the distribution network loss reduction simulation effect, the distribution network line loss control strategy parameters are optimized in the direction of analysis to obtain parameter variation optimization rules; according to the parameter variation optimization rules, the distribution network line loss control strategy parameters are mutated and expanded to obtain a line loss control strategy parameter cluster; parameter comparison and optimization are performed within the line loss control strategy parameter cluster to obtain line loss control optimization strategy parameters, and distribution network line loss optimization control is performed through the line loss control optimization strategy parameters.

[0025] Through the subsequent detailed description of the line loss optimization method based on distributed photovoltaic topology analysis in this specification, those skilled in the art can clearly understand the line loss optimization system based on distributed photovoltaic topology analysis in this embodiment. Since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.

[0026] Embodiment 2 is based on the same inventive concept as the line loss optimization system based on distributed photovoltaic topology analysis in the previous embodiment. Figure 2 As shown, the embodiment of the present application provides a line loss optimization method based on distributed photovoltaic topology analysis, the method comprising: The basic distribution data of the distribution network and the photovoltaic access data are collected and acquired, and graph connection attributes are defined. The basic distribution data of the distribution network and the photovoltaic access data are topologically connected based on the graph connection attributes to establish a distributed photovoltaic distribution network topology model.

[0027] The basic distribution data of the distribution network and the photovoltaic access data are collected and acquired through sensors, data acquisition systems, etc. Among them, the basic distribution data is the basic information related to the distribution network, such as the geographical location, connection relationship, capacity, load data, etc. of substations, lines, load points, branches, etc. Photovoltaic access data refers to the relevant information of the distributed photovoltaic system accessing the distribution network, including the geographical location, access point location, access power, installation capacity, power generation curve, etc. of the photovoltaic power generation system.

[0028] Define graph connection attributes. Graph connection attributes refer to the nodes (such as substation nodes, photovoltaic nodes, and load nodes) defined in topology modeling and the connection relationships between them, including entity node attributes and node connection edge attributes. Entity node attributes include entities in the distribution network, such as substations, user loads, and photovoltaic sites. Node connection edge attributes represent the connection relationship between nodes, usually representing lines.

[0029] Based on the graph connection properties, the basic distribution data of the distribution network and the photovoltaic access data are topologically connected, which means mapping the actual physical equipment and connection relationships into a graph structure, and connecting each node according to their geographical location or electrical connection method. Specifically, through the basic distribution data of the distribution network and the photovoltaic access data, each entity node in the network is identified, such as load nodes, power generation nodes, substation nodes, etc. Based on the line distribution of the distribution network, the photovoltaic access location, etc., the adjacency relationship between the nodes is constructed, and the corresponding connection edges are generated. The physical structure of the distribution network is abstracted into a topological graph. The nodes represent the physical entities in the network, and the edges represent the power lines. Based on the topological connection results, a distributed photovoltaic distribution network topology model is established. This model describes the connection method of each node and edge of the distribution network, which serves as the basis for subsequent analysis.

[0030] The photovoltaic output data set and the power grid load data set are associated by data mining technology, and the photovoltaic output data set and the power grid load data set are predicted and modeled by time series analysis to obtain a photovoltaic output-power grid load spatiotemporal prediction model.

[0031] Data mining technology refers to the technology of extracting useful information from a large amount of raw data. Here, the goal of data mining is to extract photovoltaic output data sets and power grid load data sets from the monitoring data and historical data of the distribution network. The photovoltaic output data set includes the photovoltaic power generation system's power generation, power generation, light intensity, temperature and other time series data related to photovoltaic power generation. The photovoltaic output data is affected by meteorological conditions, time (such as season, day length) and other factors; the power grid load data set includes the load demand time series data in the power grid, such as power consumption, load change rate, etc. The power grid load data is affected by multiple factors such as weather, time, and user demand.

[0032] Time series analysis is a technology used to analyze time series data. It uses the time characteristics of data to build models and predict. Time series analysis is performed on photovoltaic output data sets and grid load data sets respectively to capture their temporal and spatial variation patterns. Specifically, the photovoltaic output data sets and grid load data sets are arranged in chronological order to construct time series, for example, photovoltaic power generation or power load at different times of the day. The characteristics of these time series data are analyzed to identify influencing factors, such as seasonal trends, diurnal fluctuations, peak and valley power consumption periods, etc. Photovoltaic power generation and grid load may show periodic and seasonal fluctuations. According to the characteristics of the data, a time series prediction model is established to obtain the photovoltaic output prediction model and the grid load prediction model. The models are parallel integrated and verified and tuned to generate a photovoltaic output-grid load spatiotemporal prediction model.

[0033] Initialize particle swarm parameters based on the distributed photovoltaic distribution network topology model, optimize photovoltaic access positions for the particle swarm parameters according to the line loss optimization target, and iteratively obtain the distributed photovoltaic optimized distribution network topology model.

[0034] The PV access location is optimized using the particle swarm optimization algorithm, which is an optimization algorithm based on swarm intelligence. It simulates the collaborative behavior of bird flocks or fish schools in the process of searching for food. Each particle represents a potential solution to the problem. Through information exchange and update between particles, the optimal solution is gradually approached.

[0035] First, based on the distributed photovoltaic distribution network topology model, the particle swarm parameters are initialized, that is, a group of particles are randomly generated, each particle represents a candidate solution for a photovoltaic access location, and all particles together form a particle swarm. The particles in the group will continue to move in the solution space and gradually optimize the access location. Define the line loss optimization goal, that is, find the photovoltaic power generation access location so that the line loss of the distribution network is minimized after the photovoltaic system is connected.

[0036] The PV access position is optimized for the particle swarm parameters according to the line loss optimization objective. Specifically, the line loss of the distribution network is calculated according to the PV access position of each particle. The calculated fitness value represents the quality of the current solution of each particle. The goal is to minimize the line loss. The particle speed is updated according to the current speed and current optimal position (individual optimal position and global optimal position) of the particle, and the particle position is updated according to the new speed. The new fitness of the updated particle, i.e., the line loss of the new access position, is calculated, and the individual optimal position and global optimal position of each particle are updated.

[0037] The particle swarm algorithm continuously adjusts the access position of particles through multiple iterations and gradually approaches the optimal solution. After each iteration, the distribution network topology model is updated according to the current optimal photovoltaic access position. When the particle swarm algorithm converges or reaches the preset number of iterations, the optimal particle with the largest fitness is determined, that is, the optimal solution for the photovoltaic access position. At this time, the access position of distributed photovoltaics has been determined through optimization, and an optimized distributed photovoltaic distribution network topology model can be obtained. This model has better line loss performance than the initial model and can significantly reduce the loss in power transmission.

[0038] The photovoltaic load spatiotemporal distribution characteristic information is obtained according to the photovoltaic output-grid load spatiotemporal prediction model, and a strategy analysis is performed based on the photovoltaic load spatiotemporal distribution characteristic information and the distributed photovoltaic optimization distribution network topology model to determine the distribution network line loss control strategy parameters.

[0039] Based on the PV output-grid load spatiotemporal prediction model, the PV output and grid load at different time and space nodes in the future are predicted, and the spatiotemporal distribution characteristic information of PV load is generated, including the output capacity of PV power generation at different times (sequentiality), the distribution of grid load in different regions (spatiality), and the matching degree between PV power generation and grid load.

[0040] Conduct strategy analysis, specifically, analyze the changes in power flow in the power grid, that is, the power flow distribution in the power grid. After photovoltaic power generation is connected, the power distribution of the power grid will change. It is necessary to analyze how electricity is transmitted and exchanged in the power grid at different times and in different regions, especially the impact of photovoltaic power generation access on the power flow, including line load and line loss distribution. According to the analysis results, a matching analysis is performed in the line loss control strategy library to obtain the distribution network line loss matching control strategy. Based on the distribution network line loss matching control strategy, a global optimization is performed within the parameter selection threshold to determine the distribution network line loss control strategy parameters.

[0041] The distribution network line loss control strategy parameters are simulated and evaluated to obtain a distribution network loss reduction simulation effect, and the distribution network line loss control strategy parameters are feedback optimized based on the distribution network loss reduction simulation effect.

[0042] A simulation environment is constructed using the distributed photovoltaic distribution network topology model. The determined distribution network line loss control strategy parameters are imported into the simulation environment. Different load scenarios and photovoltaic power generation scenarios are run in the simulation environment to simulate the actual operation of the distribution network. The scenarios can include typical peak and valley power consumption, changes in photovoltaic power generation under different weather conditions, etc. During the simulation process, the power flow distribution, voltage conditions, photovoltaic power generation access capacity and other data of the distribution network are recorded in real time, and the line loss under simulation is calculated as the simulation result. Based on the simulation results, the effect of the line loss control strategy is evaluated. The evaluation indicators mainly include line loss changes, that is, the distribution network line loss under different scenarios is calculated through simulation, and the line loss conditions before and after the application of the line loss control strategy are compared to observe whether the line loss is significantly reduced. It also includes voltage stability, load matching, power flow distribution, etc.

[0043] If the simulation results show that the line loss has not achieved the expected reduction effect, or there are problems with voltage stability, the relevant control strategy parameters are optimized, such as adjusting the priority of photovoltaic access points to ensure that photovoltaic power generation can be consumed more locally and reduce long-distance power transmission until the simulation effect meets the expected requirements. By feeding back the optimization strategy parameters, the accuracy and effectiveness of the control strategy can be further improved, thereby ultimately achieving the effect of minimizing line losses.

[0044] Furthermore, the establishment of a distributed photovoltaic distribution network topology model includes: According to the graph connection attributes, entity node attributes and node connection edge attributes are determined; based on the entity node attributes, entity recognition and attribute identification are performed on the distribution network basic distribution data and photovoltaic access data to obtain a distribution network entity set and an entity node attribute set; connection relationship analysis is performed on the distribution network entity set through the distribution network basic distribution data and photovoltaic access data to construct a node neighbor connection relationship set; topological abstraction is performed on the node neighbor connection relationship set based on the node connection edge attributes to obtain an entity node topological adjacency matrix; topological identification connection is performed on the distribution network entity set based on the entity node topological adjacency matrix and the entity node attribute set to establish the distributed photovoltaic distribution network topology model.

[0045] Graph connection attributes are used to describe the characteristic attributes of each node in the distribution network and their connection relationships. The basic data of all relevant nodes are obtained from the distribution network, such as the geographical location and equipment characteristics of substations, photovoltaic access points, load points, switches, etc., and specific attributes are defined for each node, such as the type of node, the voltage level and capacity of the node, etc., as entity node attributes; for each pair of connected nodes, the attributes of the line between them are defined, such as length, impedance, maximum load current, etc., as node connection edge attributes.

[0046] By parsing the basic distribution data of the distribution network and the photovoltaic access data, all entity node types in the distribution network are identified, which usually include: substation nodes, which serve as the power source or voltage conversion point of the power grid; photovoltaic nodes, which represent the access points of distributed photovoltaic power generation systems; load nodes, which represent the points of power consumption; other nodes, such as switches, segmentation points, etc., depending on specific needs. For each identified entity node, attribute identification is performed according to its physical characteristics and power functions, such as voltage level, power, geographical location, etc. All identified and marked entity nodes are integrated to form a distribution network entity set, that is, a set of all nodes in the distribution network, and a set of entity node attributes, including detailed attributes of each node.

[0047] The basic distribution data of the distribution network and the photovoltaic access data are used to analyze the topological structure of the distribution network and identify the neighboring nodes of each node. The neighboring nodes refer to the nodes that are directly connected through the lines. For example, the photovoltaic node may be adjacent to the nearest substation or load node. Based on the identified neighboring nodes, the connection direction and distance analysis are performed. Specifically, the direction of power flow is determined. For example, in a typical distribution network, power usually flows from the substation node to the load node. The line length between the two nodes is calculated through the basic distribution data of the distribution network. This can be obtained through the geographic coordinates of the node or the pre-recorded line length information. The connection relationship between each node and its neighboring nodes is recorded to form a node neighbor connection relationship set, which describes the physical and electrical connection relationship between each node in the network and its neighboring nodes.

[0048] After constructing the node proximity connection relationship set, the connection edges between the nodes are topologically abstracted, that is, the nodes and connections in the physical power grid are represented by a graph theory model to more intuitively represent the network structure. Specifically, the connection direction and distance between each node are marked with connection edges, such as directed edges and connection edge lengths. For ease of identification, a physical node topological adjacency matrix of each node is constructed. The adjacency matrix is ​​a matrix that describes the connection relationship between nodes in a graph, and each element in it indicates the connection relationship between a node and the remaining nodes.

[0049] Exemplarily, for a network with n nodes, the adjacency matrix A is an n×n matrix, and each element A[i][j] represents the connection between node i and node j, specifically including: if there is a connection between node i and node j, then A[i][j]≠0, and the value can represent the properties of the connection edge, such as the length of the connection; if there is no connection between node i and node j, then A[i][j]=0; if the connection is directional, the values ​​of A[i][j] and A[j][i] can be different, indicating the direction of power flow.

[0050] Topological identification connection refers to the identification and abstraction of nodes and their connection relationships in the distribution network, mapping the actual physical connection relationship into the topological model so that it can be used for mathematical analysis and simulation calculations. Specifically, the entity node topological adjacency matrix and the entity node attribute set are combined to identify each node in the distribution network, and a complete network topology is constructed based on the connection relationship between nodes. The node attributes, adjacency matrix and connection edge attributes are combined to finally construct a distributed photovoltaic distribution network topology model, which reflects how electricity flows in the network and can be used for analysis and optimization.

[0051] Furthermore, the step of obtaining a photovoltaic output-grid load spatiotemporal prediction model includes: The photovoltaic output data set and the grid load data set are arranged in time series and analyzed for characteristics by using time series analysis to obtain photovoltaic output characteristic data sets and grid load characteristic data sets; the photovoltaic output characteristic data sets and the grid load characteristic data sets are respectively trained for prediction by using an LSTM network to obtain a photovoltaic output prediction model and a grid load prediction model; the photovoltaic output prediction model and the grid load prediction model are connected in parallel to fuse the photovoltaic output prediction model and the grid load prediction model to generate an initial photovoltaic load spatiotemporal prediction model; the initial photovoltaic load spatiotemporal prediction model is verified and optimized by using a model optimizer to obtain the photovoltaic output-grid load spatiotemporal prediction model.

[0052] Both photovoltaic output data and grid load data have obvious time correlation. Time series arrangement is to organize the data in chronological order for subsequent time series analysis. Characteristic analysis refers to identifying and extracting key factors that affect photovoltaic output and grid load, and integrating them into characteristic data sets. Photovoltaic output characteristic analysis includes analysis of sunshine intensity, temperature, time period, etc. For example, sunshine intensity directly affects photovoltaic power generation, and high temperature usually reduces the efficiency of photovoltaic modules; grid load characteristic analysis includes analysis of time factors, weather factors, and social factors. For example, load demand is closely related to time, such as load increases during peak hours in the morning and evening, while the load is relatively low at night or during working days. According to the analysis results, photovoltaic output characteristic data sets and grid load characteristic data sets are obtained.

[0053] LSTM (Long Short-Term Memory) network is a special recursive neural network suitable for processing and predicting time series data. The structure of LSTM can capture long-term dependencies in data, and is particularly suitable for analyzing time-dependent sequence data such as photovoltaic output and grid load. Specifically, the LSTM network is used to train the photovoltaic output characteristic data set, with the aim of predicting future photovoltaic power generation through historical data and its influencing factors (such as sunshine, temperature, etc.) and obtaining a photovoltaic output prediction model; similarly, the LSTM network is used to train the grid load characteristic data set, with the aim of predicting future grid load based on historical electricity consumption data and its related characteristics (such as time, weather, etc.) and obtaining a grid load prediction model.

[0054] Parallel fusion is to integrate the output results of two prediction models so that the predictions of photovoltaic output and grid load can be processed simultaneously. Specifically, the photovoltaic output prediction and the grid load prediction are integrated according to the time step. Assuming that the prediction is for the power generation and load demand in the next 24 hours, the prediction results of photovoltaic output in each hour will correspond to the prediction results of the grid load in the same hour. Through parallel fusion, the initial photovoltaic load spatiotemporal prediction model is obtained. This model is a comprehensive model that can simultaneously predict photovoltaic power generation and grid load demand in both time and space dimensions.

[0055] Model optimizers are algorithms used to adjust model parameters, such as Adam optimizer, RMSprop, etc. The optimizer can automatically adjust the weight and bias parameters during the training process so that the model converges to a better solution. The selected optimizer automatically adjusts the weight parameters of the model through the gradient descent method to minimize the loss function. After each round of iteration, the model parameters are adjusted by feedback error until the verification error reaches the optimal value. After verification and optimization, the final photovoltaic output-grid load spatiotemporal prediction model can more accurately predict the photovoltaic power generation and grid load demand in different time periods and different spatial nodes in the future.

[0056] Furthermore, the iterative acquisition of the distributed photovoltaic optimized distribution network topology model includes: A line loss evaluation index is extracted for the line loss optimization target to obtain a line loss evaluation index set, and a line loss optimization effect fitness function is constructed according to the line loss evaluation index set; a parameter particle position and a parameter particle speed are determined according to the particle swarm parameters, and the fitness of the particle swarm parameters is evaluated using the line loss optimization effect fitness function to obtain a particle swarm fitness set; the parameter particle position and the parameter particle speed are iteratively updated based on the particle swarm fitness set until a preset termination condition is met, and the optimal parameter particle with the largest fitness is determined by optimization, and the optimal parameter particle includes a photovoltaic access optimization position; the distributed photovoltaic distribution network topology model is optimized and updated based on the photovoltaic access optimization position to obtain the distributed photovoltaic optimized distribution network topology model.

[0057] The goal of line loss optimization is to reduce power loss in the distribution network by adjusting photovoltaic access points, load distribution, power flow control, etc. The line loss in the distribution network is mainly determined by factors such as current loss in transmission and power transmission distance. The line loss assessment index is the key parameter used to evaluate the line loss of the power system, including line current, line impedance, power transmission distance, load demand, photovoltaic access location and other indicators, which constitute a set of line loss assessment indicators.

[0058] According to the set of line loss evaluation indicators, a fitness function for line loss optimization is constructed. The fitness function is a function used to evaluate the quality of each particle in particle swarm optimization. Here, the fitness function is used to evaluate the effect of a given photovoltaic access location on line loss optimization. The goal is to minimize the total line loss in the distribution network.

[0059] In the particle swarm optimization algorithm, each solution is called a particle. The particle swarm moves in the search space and gradually approaches the optimal solution through interaction with other particles. Here, each particle represents a candidate solution for a photovoltaic access location. The optimization goal is to minimize the line loss of the power grid. Each particle has two main parameters, namely the parameter particle position and the parameter particle speed. The parameter particle position corresponds to the location of the photovoltaic access point. The parameter particle speed indicates its moving direction and step size in the solution space, which determines how the particle updates its position in each iteration. Initially, the particle position can be randomly distributed on different nodes of the distribution network. Each particle corresponds to a photovoltaic access solution. The initial speed of the particle can be randomly set, which determines the initial moving step size of the particle in the solution space. For each particle, the corresponding line loss value is calculated using the constructed line loss optimization effect fitness function. The larger the fitness value, the better the solution is in reducing line loss. According to the output of the fitness function, the particle swarm fitness set is obtained.

[0060] The core of the particle swarm optimization algorithm is to gradually optimize the solution by continuously updating the position and speed of particles. The speed and position of each particle will be updated according to its own historical best position and global optimal position in each iteration. The speed update formula is as follows: ,in, is particle i at time The speed on is the inertia weight, which determines the influence of the particle's previous velocity; , are acceleration factors, representing the weights of individual cognition and group cognition respectively; , is a random number; is the historical optimal position found by particle i itself; is the global optimal position; is particle i at time The speed on is particle i at time The position update formula is as follows: ,in, is particle i at time The position on the particle, that is, the position of the particle is updated according to its velocity at each iteration.

[0061] Based on the above speed update formula and position update formula, in each iteration, after updating the new position of each particle, the line loss optimization effect of the new position is evaluated through the fitness function to obtain a new fitness value. Each particle updates its individual optimal solution. If the position of the current iteration is better than the particle's historical position, it is replaced with the new position. At the same time, the particle with the best fitness among all particles is updated to the global optimal solution, which is the current best solution in the entire particle swarm. The particle swarm optimization algorithm repeatedly updates the position and speed of the particles and searches for the optimal solution based on the fitness function. Each iteration updates the optimal solution based on the position in the new solution space, and iterates continuously until a preset termination condition is met. For example, when the number of iterations reaches the upper limit and the fitness change is not obvious, the global optimal particle in the particle swarm is the final optimization result, representing the optimized position of photovoltaic access.

[0062] In the distributed photovoltaic distribution network topology model, the photovoltaic power generation system is connected to the node at the photovoltaic access optimization location to obtain a distributed photovoltaic optimized distribution network topology model, which not only reduces line losses but also improves the overall performance of the power grid.

[0063] Furthermore, the determining of the distribution network line loss control strategy parameters includes: Based on the distributed photovoltaic optimized distribution network topology model, a power grid flow distribution impact analysis is performed to obtain power grid flow distribution impact parameters, wherein the power grid flow distribution impact parameters include line load impact parameters and power grid line loss impact parameters; a line loss control strategy library is constructed, and based on the photovoltaic load spatiotemporal distribution characteristic information and the power grid flow distribution impact parameters, a matching analysis is performed with the line loss control strategy library to obtain a distribution network line loss matching control strategy; based on the distribution network line loss matching control strategy, the photovoltaic load spatiotemporal distribution characteristic information and the power grid flow distribution impact parameters are selected by threshold division to obtain a line loss control strategy parameter selection threshold; the line loss optimization effect fitness function is used to perform global optimization within the line loss control strategy parameter selection threshold to determine the distribution network line loss control strategy parameters.

[0064] The power grid flow distribution impact analysis evaluates the power grid operation status by calculating the flow of electricity between various nodes in the distribution network. The results of the power flow analysis are mainly used to evaluate the operation status of different lines and nodes in the power grid. The key influencing parameters include line load impact parameters and power grid line loss impact parameters.

[0065] Specifically, the line load influencing parameters include line load rate and line flow distribution. The line load rate refers to the ratio of the actual load of the power line to its rated load. The higher the load rate, the greater the load pressure of the line, which may lead to overload or increased losses. The reasonable access of photovoltaic power generation can effectively reduce the load of certain high-load lines. The line flow distribution indicates the flow of electricity carried on each line, including the direction and flow of electricity. Reasonable photovoltaic access can guide electricity to flow to the load through a shorter path and reduce unnecessary long-distance transmission.

[0066] The parameters affecting grid line loss mainly include total line loss. The line loss in the grid is mainly determined by line impedance and current. After the power flow distribution is optimized, it is necessary to calculate the total line loss of the grid after connecting to photovoltaic power generation and measure the line loss changes caused by power flow adjustment, which is mainly determined by the current, line impedance and line length of each line.

[0067] A line loss control strategy library is developed based on different power flow distributions, photovoltaic power generation outputs and load demand conditions, including photovoltaic power generation scheduling strategies, load transfer strategies, and voltage regulation strategies. For example, during peak load periods, photovoltaic power generation gives priority to supplying power to high-load areas, reducing the demand for long-distance transmission, thereby reducing line losses; through load transfer (such as demand response), part of the load on high-load lines is transferred to low-load lines to relieve line pressure and reduce line losses.

[0068] According to the parameters influencing the power grid flow distribution and the spatial and temporal distribution characteristics of the photovoltaic load, the line loss control strategy library is matched to find the most suitable strategy. Specifically, the flow distribution parameters are compared with the strategy parameters in the line loss control strategy library to find a control strategy that meets the current flow state. For example, if the load rate of some lines is too high, the load transfer strategy is selected; if the voltage deviation of some nodes is too large, the voltage regulation strategy is selected. According to the spatial and temporal distribution characteristics of photovoltaic power generation and load, it is determined which strategy can most effectively reduce line losses. For example, in a high-load period, the power generation scheduling strategy is preferred to supply photovoltaic power to local loads to reduce long-distance transmission; in the low-load period, the voltage regulation or flow optimization strategy is adopted to avoid excessive grid voltage.

[0069] If there are multiple strategies that can match the current power flow distribution and photovoltaic load characteristics, the strategies are sorted according to the line loss reduction effect, and the strategy with the best effect is selected first, and finally a distribution network line loss matching control strategy suitable for the current power grid state is obtained.

[0070] By analyzing the temporal and spatial changes of photovoltaic power generation and load demand, the strategy parameter range suitable for each time period or load point is determined. For example, photovoltaic power generation is usually higher during the day, so the parameter range of power generation scheduling can be set as high as 70%-90% during the day; by analyzing the power grid flow and line load conditions, the parameter range of the load rate and voltage fluctuation of each line is determined. If the load rate of a line is high, the threshold of the load transfer strategy is set within a lower load rate range. According to key factors such as photovoltaic power generation, load demand, and flow distribution, the parameter value range of each control strategy is divided. The result of the division is the upper and lower limits of each parameter, and the line loss control strategy parameter selection threshold is obtained, that is, the parameter range allowed for each control strategy.

[0071] A global optimization is performed within the threshold for selecting the parameters of the line loss control strategy. The purpose is to find the optimal solution within the allowable range of these parameters. The effects of different strategy parameter combinations are evaluated through the fitness function. The impact of each parameter combination on the power grid line loss is calculated through the fitness function. The goal is to minimize the line loss. Through global optimization, a set of parameter combinations is found to maximize the fitness value as the distribution network line loss control strategy parameters.

[0072] Furthermore, the determining of the distribution network line loss control strategy parameters includes: A plurality of strategy parameters are randomly selected within the line loss control strategy parameter selection threshold, and the plurality of strategy parameters are evaluated using the line loss optimization effect fitness function to obtain a plurality of parameter fitnesses; the line loss control strategy parameter selection threshold is approximated in an optimization region based on the plurality of parameter fitnesses to determine a local optimization region for strategy parameters; a parameter search step is set according to the local optimization region for strategy parameters; a strategy parameter search and evaluation is performed in the local optimization region for strategy parameters according to the parameter search step, and an optimization region is iteratively approximated according to the parameter search and evaluation results until a preset number of iterations are reached, and the distribution network line loss control strategy parameters are determined by comparing the fitnesses.

[0073] Within the line loss control strategy parameter selection threshold, multiple strategy parameters are generated by random sampling. For each randomly selected strategy parameter, the line loss optimization effect fitness function is used to evaluate the optimization effect of each set of strategy parameters on the distribution network line loss. The calculation result is the fitness value of each strategy parameter. The larger the fitness value, the better the line loss optimization effect of the combination.

[0074] Sort the fitness of multiple parameters of all strategy parameters by quality, find the strategy parameters with larger fitness values, that is, better optimization effects, and narrow the value range of strategy parameters by analyzing the fitness value distribution, gradually approaching the local area where the optimal solution is located, and determining the local optimization area of ​​the strategy parameters. In this way, a more refined search can be concentrated in the better area.

[0075] The search step size refers to the amplitude of adjusting the parameters each time during the optimization process. The step size determines the search accuracy during the optimization process. The smaller the step size, the more precise the search, but the amount of calculation will also increase; the larger the step size, the rougher the search, but it may approach the solution faster. According to the local optimization area of ​​the strategy parameter, set the parameter search step size, that is, when the local optimization area is small, you need to set a smaller step size to accurately search for the optimal solution; when the area is large, you can use a slightly larger step size for preliminary search and gradually reduce the step size.

[0076] In the local optimization area of ​​the strategy parameters, multiple strategy parameters are generated again according to the search step size, and each strategy parameter is evaluated using the line loss optimization fitness function. The line loss optimization effect is calculated to obtain the corresponding fitness value. The fitness values ​​of all parameters are compared to find the parameter with the larger fitness value, and the interval where the parameter is located is obtained. The interval is used as the new local optimization area, and the search step size is further reduced. A second round of search is performed in the new smaller interval. In each iteration, the search step size is gradually reduced, and the optimal solution is gradually approached. This process is repeated until the preset number of iterations is reached, and the optimal distribution network line loss control strategy parameters are finally determined.

[0077] Furthermore, the feedback optimization of the distribution network line loss control strategy parameters based on the distribution network loss reduction simulation effect includes: Based on the distribution network loss reduction simulation effect, the distribution network line loss control strategy parameters are optimized in the direction of analysis to obtain parameter variation optimization rules; according to the parameter variation optimization rules, the distribution network line loss control strategy parameters are mutated and expanded to obtain a line loss control strategy parameter cluster; parameter comparison and optimization are performed within the line loss control strategy parameter cluster to obtain line loss control optimization strategy parameters, and distribution network line loss optimization control is performed through the line loss control optimization strategy parameters.

[0078] Distribution network loss reduction simulation is to simulate the operating status of the distribution network under different line loss control strategy parameters through simulation software to evaluate the optimization effect of the strategy on line loss, including line loss changes under different loads and power generation conditions, line load rate, voltage deviation, etc. By analyzing the simulation results, it can be determined whether the current line loss control strategy has achieved the expected loss reduction effect, and identify where there is still room for optimization.

[0079] Through the simulation results, the specific impact of each control strategy parameter on line loss optimization is determined, and how to adjust the parameters can further reduce line losses. For example, if the simulation results show that the load transfer strategy fails to effectively reduce the load pressure of the line during certain load peaks, it can be inferred that the load transfer parameters need to be further adjusted; if the scheduling strategy of photovoltaic power generation fails to effectively reduce the demand for long-distance power transmission, it can be analyzed whether the priority power supply area or scheduling time of photovoltaic power generation needs to be adjusted. According to the results of the optimization direction analysis, the parameter variation optimization rule is formulated, which defines how the parameters change during the optimization process to further reduce line losses.

[0080] Mutation expansion refers to making a certain degree of change to the current control strategy parameters based on the optimization rules to generate a set of new parameter combinations. Specifically, each control strategy parameter is mutated using the obtained parameter variation optimization rules. For example, if the variation rule of the photovoltaic power generation scheduling parameters is ±5%, then several different scheduling parameters are generated on the original basis, such as +5%, -5%. The mutated parameters form a new parameter cluster, namely the line loss control strategy parameter cluster, covering multiple possible configuration schemes.

[0081] Comparative optimization refers to finding the optimal strategy parameters for line loss by evaluating the performance of multiple parameters in the parameter cluster. Specifically, the influence of each group of parameters on line loss under different working conditions is obtained through simulation, and the optimization effect of each group of parameters is measured by the fitness function to obtain the fitness value of each group of parameters. The fitness values ​​of all parameters are compared to find the combination with the largest fitness value, that is, the line loss control optimization strategy parameters. The obtained line loss control optimization strategy parameters are used in the actual distribution network control operation to achieve the optimal line loss control effect.

[0082] In summary, the line loss optimization method based on distributed photovoltaic topology analysis provided in the embodiments of the present application has the following technical effects: A distributed photovoltaic distribution network topology model is established. This model effectively integrates the basic distribution data of the distribution network and the photovoltaic access data to form a clear network structure. The graph theory topology connection method is used to accurately reflect the connection relationship between each node in the power grid, so that the impact of photovoltaic access on the power grid flow can be visualized and analyzed; data mining and time series analysis technology are used to construct a photovoltaic output-grid load spatiotemporal prediction model. This model integrates the spatiotemporal characteristics of photovoltaic power generation and the dynamic changes of grid load demand, and realizes the matching prediction of photovoltaic power generation and grid load demand; the particle swarm algorithm is used to optimize the photovoltaic access location, and the optimal location of the photovoltaic access point is iteratively adjusted based on the line loss optimization target, and the optimized distributed photovoltaic optimized distribution network topology model is formed. This process combines the grid topology structure and the prediction model, so that Photovoltaic access points can minimize line losses during grid operation, reduce losses during power transmission, and achieve reasonable allocation of grid resources and maximize power efficiency. Through comprehensive analysis of the spatiotemporal distribution characteristics of photovoltaic loads and combined with the distributed photovoltaic optimization distribution network topology model, effective distribution network line loss control strategy parameters can be formulated to respond to dynamic changes in photovoltaic output and grid load in real time, and further reduce line losses by reasonably scheduling photovoltaic power generation and adjusting power flow paths. Simulation evaluation and feedback optimization are carried out to verify the effectiveness of the formulated line loss control strategy through simulation, and optimize the parameters based on the simulation results. This process ensures the effectiveness and pertinence of the line loss control strategy, so that the strategy can be continuously adjusted and improved according to actual conditions, realizing closed-loop optimization and ensuring that line losses are minimized during grid operation.

[0083] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A line loss optimization system based on distributed photovoltaic topology analysis, characterized in that: The system comprises: A topological connection module, which is used to collect and obtain basic distribution data of the distribution network and photovoltaic access data, and define graph connection attributes, and topologically connect the basic distribution data of the distribution network and the photovoltaic access data based on the graph connection attributes to establish a distributed photovoltaic distribution network topology model; A prediction modeling module, wherein the prediction modeling module is used to obtain a photovoltaic output data set and a power grid load data set by associating them through data mining technology, and to perform prediction modeling on the photovoltaic output data set and the power grid load data set using time series analysis to obtain a photovoltaic output-power grid load spatiotemporal prediction model; An access location optimization module, the access location optimization module is used to initialize particle swarm parameters based on the distributed photovoltaic distribution network topology model, optimize the photovoltaic access location of the particle swarm parameters according to the line loss optimization target, and iteratively obtain the distributed photovoltaic optimized distribution network topology model; A strategy analysis module, the strategy analysis module is used to obtain the spatiotemporal distribution characteristic information of photovoltaic load according to the spatiotemporal prediction model of photovoltaic output-grid load, perform strategy analysis based on the spatiotemporal distribution characteristic information of photovoltaic load and the distributed photovoltaic optimization distribution network topology model, and determine the distribution network line loss control strategy parameters; A feedback optimization module, the feedback optimization module is used to simulate and evaluate the distribution network line loss control strategy parameters, obtain the distribution network loss reduction simulation effect, and feedback optimize the distribution network line loss control strategy parameters based on the distribution network loss reduction simulation effect.

2. The line loss optimization system based on distributed photovoltaic topology analysis according to claim 1, characterized in that: The establishment of a distributed photovoltaic distribution network topology model includes: Determine entity node attributes and node connection edge attributes according to the graph connection attributes; Based on the entity node attributes, entity recognition and attribute identification are performed on the basic distribution data of the distribution network and the photovoltaic access data to obtain a distribution network entity set and an entity node attribute set; Performing connection relationship analysis on the distribution network entity set through the distribution network basic distribution data and photovoltaic access data to construct a node proximity connection relationship set; Performing topological abstraction on the node neighbor connection relationship set based on the node connection edge attributes to obtain a physical node topological adjacency matrix; The distribution network entity set is connected by topological identification based on the entity node topological adjacency matrix and the entity node attribute set to establish the distributed photovoltaic distribution network topology model.

3. The line loss optimization system based on distributed photovoltaic topology analysis according to claim 1, characterized in that: The step of obtaining a photovoltaic output-grid load spatiotemporal prediction model includes: Using time series analysis to perform time series arrangement and characteristic analysis on the photovoltaic output data set and the power grid load data set to obtain a photovoltaic output characteristic data set and a power grid load characteristic data set; The photovoltaic output characteristic data set and the power grid load characteristic data set are respectively trained for prediction by using an LSTM network to obtain a photovoltaic output prediction model and a power grid load prediction model; The photovoltaic output prediction model and the power grid load prediction model are connected in parallel to generate an initial photovoltaic load spatiotemporal prediction model; The initial photovoltaic load spatiotemporal prediction model is verified and optimized using a model optimizer to obtain the photovoltaic output-grid load spatiotemporal prediction model.

4. The line loss optimization system based on distributed photovoltaic topology analysis according to claim 1, characterized in that: The iterative method for obtaining a distributed photovoltaic optimized distribution network topology model includes: Extracting line loss evaluation indicators for the line loss optimization target to obtain a line loss evaluation indicator set, and constructing a line loss optimization effect fitness function based on the line loss evaluation indicator set; According to the particle swarm parameters, the parameter particle position and the parameter particle speed are determined, and the fitness of the particle swarm parameters is evaluated using the line loss optimization effect fitness function to obtain a particle swarm fitness set; Iteratively updating the parameter particle position and parameter particle speed based on the particle swarm fitness set until a preset termination condition is met, and optimizing and determining the optimal parameter particle with the maximum fitness, wherein the optimal parameter particle includes the photovoltaic access optimization position; The distributed photovoltaic distribution network topology model is optimized and updated based on the photovoltaic access optimization position to obtain the distributed photovoltaic optimization distribution network topology model.

5. The line loss optimization system based on distributed photovoltaic topology analysis according to claim 4, characterized in that: Determining the distribution network line loss control strategy parameters includes: Based on the distributed photovoltaic optimized distribution network topology model, a power grid power flow distribution impact analysis is performed to obtain power grid power flow distribution impact parameters, wherein the power grid power flow distribution impact parameters include line load impact parameters and power grid line loss impact parameters; Constructing a line loss control strategy library, and performing matching analysis with the line loss control strategy library based on the spatiotemporal distribution characteristic information of the photovoltaic load and the influencing parameters of the power grid flow distribution, to obtain a distribution network line loss matching control strategy; Based on the distribution network line loss matching control strategy, the photovoltaic load spatiotemporal distribution characteristic information and the power grid power flow distribution influencing parameters are divided into thresholds to obtain a line loss control strategy parameter selection threshold; The line loss optimization effect fitness function is used to perform global optimization within the line loss control strategy parameter selection threshold to determine the distribution network line loss control strategy parameters.

6. The line loss optimization system based on distributed photovoltaic topology analysis according to claim 5, characterized in that: The determining of the distribution network line loss control strategy parameters comprises: Randomly selecting multiple strategy parameters within the line loss control strategy parameter selection threshold, and evaluating the multiple strategy parameters using the line loss optimization effect fitness function to obtain multiple parameter fitness; Based on the multiple parameter fitnesses, the thresholds for selecting the line loss control strategy parameters are approximated in an optimization region to determine a local optimization region for the strategy parameters; According to the local optimization area of ​​the strategy parameters, setting the parameter search step size; A strategy parameter search and evaluation is performed in the local optimization area of ​​the strategy parameters according to the parameter search step, and an optimization area iterative approximation is performed according to the parameter search and evaluation result until a preset number of iterations is reached, and the distribution network line loss control strategy parameters are determined by fitness comparison.

7. The line loss optimization system based on distributed photovoltaic topology analysis according to claim 1, characterized in that: The performing feedback optimization on the distribution network line loss control strategy parameters based on the distribution network loss reduction simulation effect includes: Based on the distribution network loss reduction simulation effect, the distribution network line loss control strategy parameters are optimized and analyzed to obtain parameter variation optimization rules; According to the parameter variation optimization rule, the distribution network line loss control strategy parameters are mutated and expanded to obtain a line loss control strategy parameter cluster; Parameter comparison and optimization are performed within the line loss control strategy parameter cluster to obtain line loss control optimization strategy parameters, and distribution network line loss optimization control is performed using the line loss control optimization strategy parameters.

8. A line loss optimization method based on distributed photovoltaic topology analysis, characterized in that: Based on the implementation of the line loss optimization system based on distributed photovoltaic topology analysis according to any one of claims 1 to 7, the method comprises: Collect and obtain basic distribution data of the distribution network and photovoltaic access data, and define graph connection attributes, and topologically connect the basic distribution data of the distribution network and photovoltaic access data based on the graph connection attributes to establish a distributed photovoltaic distribution network topology model; The photovoltaic output data set and the grid load data set are obtained by associating them through data mining technology, and the photovoltaic output data set and the grid load data set are predicted and modeled using time series analysis to obtain a photovoltaic output-grid load spatiotemporal prediction model; Initialize particle swarm parameters based on the distributed photovoltaic distribution network topology model, optimize photovoltaic access positions for the particle swarm parameters according to the line loss optimization target, and iteratively obtain the distributed photovoltaic optimized distribution network topology model; According to the photovoltaic output-grid load spatiotemporal prediction model, the spatiotemporal distribution characteristic information of photovoltaic load is obtained, and based on the spatiotemporal distribution characteristic information of photovoltaic load and the distributed photovoltaic optimization distribution network topology model, a strategy analysis is performed to determine the distribution network line loss control strategy parameters; The distribution network line loss control strategy parameters are simulated and evaluated to obtain a distribution network loss reduction simulation effect, and the distribution network line loss control strategy parameters are feedback optimized based on the distribution network loss reduction simulation effect.

Images