A power distribution network operation state analysis method and system based on graph data

Abstract

The application provides a power distribution network operation state analysis method and system based on graph data, comprising: constructing a power distribution network graph database by converting the obtained technical parameters, historical and real-time operation data, user historical power consumption data, historical distributed power generation data, corresponding weather data and weather forecast data of the power distribution network into graph data by using a graph database construction method; based on the power distribution network graph database, using a power distribution network parallel power flow analysis method, a power supply path analysis method, a user power consumption curve analysis method and a distributed power generation prediction data analysis method to analyze the operation state of the power distribution network, and obtaining the analysis result of the operation state of the power distribution network. The method provided by the application can comprehensively and accurately analyze the operation state of the power distribution network at each time period.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of power distribution network diagram data application, and particularly relates to a power distribution network operation state analysis method and system based on diagram data. BACKGROUND

[0002] The existing technology center diagram data management and computing technology has been widely applied in the fields of Internet / Internet of Things, social networks, e-commerce, biological gene mapping, intelligent transportation, etc. In the power system, the data sharing mechanism based on diagram data management and computing is still in the exploratory stage. In recent years, in the process of data research for the power system, the "power grid map" data sharing platform has been explored based on the regulation and control cloud. The "power grid map" information index diagram has been initially realized, and the "power grid map" data sharing platform covering from 500KV to 10KV distribution transformers has been constructed. The fusion and sharing of regional EMS, ground DMS and the distribution automation system under their jurisdiction in the automation field have been initially realized. At the same time, a super-speed EMS system has been developed for the main grid of the power system, including typical scenarios and algorithms such as fast network topology analysis, state estimation, online power flow calculation, and contingency analysis. However, compared with the main grid, the network structure of the distribution network is more complex. Therefore, the analysis method for analyzing the operation state of the distribution network according to the diagram data is relatively one-sided, and it is difficult to comprehensively analyze the distribution network at each time period based on the diagram data. SUMMARY

[0003] In view of the problem in the prior art that the analysis method for analyzing the operation state of the distribution network according to the diagram data is relatively one-sided, and it is difficult to comprehensively analyze the distribution network at each time period based on the diagram data, the present application provides a power distribution network operation state analysis method based on diagram data, comprising:

[0004] The technical parameters, historical and real-time operation data, user historical power consumption data, historical distributed power generation data, corresponding weather data and weather forecast data of the acquired distribution network are converted into graph data by using a graph database construction method, and then a distribution network graph database is constructed.

[0005] Based on the distribution network graph database, the operation state of the distribution network is analyzed by using a distribution network parallel power flow analysis method, a power supply path analysis method, a user power consumption curve analysis method and a distributed power generation prediction data analysis method, and an analysis result of the operation state of the distribution network is obtained.

[0006] Preferably, based on the distribution network graph database, the operation state of the distribution network is analyzed by using a distribution network parallel power flow analysis method, a power supply path analysis method, a user power consumption curve analysis method and a distributed power generation prediction data analysis method, and an analysis result of the operation state of the distribution network is obtained, comprising:

[0007] Based on the technical parameters and real-time operation data of the power distribution network in the power distribution network graph database, the parallel power flow analysis method is used in combination with the graph data traversal algorithm to calculate the power flow analysis result of the power distribution network.

[0008] The power supply path analysis result of the power distribution network is obtained by using the graph data shortest path algorithm to analyze the power supply path of the power distribution network technical parameters in the power distribution network graph database.

[0009] The user power consumption analysis result of the power distribution network is obtained by using the clustering algorithm to analyze the user historical power consumption data of the power distribution network in the power distribution network graph database.

[0010] The distributed power generation prediction analysis result of the power distribution network under the corresponding weather forecast data is obtained by inputting the weather forecast data of the power distribution network into the pre-trained power distribution network distributed power generation data analysis prediction model.

[0011] The power flow analysis result of the power distribution network, the power supply path analysis result of the power distribution network, the user power consumption analysis result of the power distribution network, and the distributed power generation prediction analysis result of the power distribution network under the corresponding weather forecast data are used as the operation state analysis result of the power distribution network.

[0012] Preferably, the training of the power distribution network distributed power generation data analysis prediction model comprises:

[0013] The vertex feature vectors are obtained by using the graph vectorization technology and the graph neural network to process the vertex features of the historical distributed power generation data and the corresponding weather data of the power distribution network in the power distribution network graph database, and the vertex feature vectors are used as the training set of the power distribution network distributed power generation data analysis prediction model.

[0014] Based on the training set of the power distribution network distributed power generation data analysis prediction model, the vertex feature vectors of the weather data corresponding to the historical distributed power generation data are input, and the vertex feature vectors of the historical distributed power generation data are output to train the power distribution network distributed power generation data analysis prediction model, thereby obtaining the trained power distribution network distributed power generation data analysis prediction model.

[0015] Preferably, the power flow analysis result of the power distribution network is calculated based on the technical parameters and real-time operation data of the power distribution network in the power distribution network graph database, and the parallel power flow analysis method is used in combination with the graph data traversal algorithm, comprising:

[0016] The nonlinear equation of the power flow of the power distribution network is established based on the technical parameters and real-time operation data of the power distribution network.

[0017] The nonlinear equation is solved by using a graph data traversal algorithm combined with a Newton iteration algorithm to obtain a convergence value of the nonlinear equation satisfying a convergence condition, and the output power of a balanced node corresponding to the convergence value of the nonlinear equation satisfying the convergence condition and the power of each line are taken as the power flow analysis result of the power distribution network.

[0018] Preferably, the nonlinear equation is solved by using a graph data traversal algorithm combined with a Newton iteration algorithm to obtain a convergence value of the nonlinear equation satisfying a convergence condition, and the output power of a balanced node corresponding to the convergence value of the nonlinear equation satisfying the convergence condition and the power of each line are taken as the power flow analysis result of the power distribution network, comprising:

[0019] S1 inputs the power of each bus supplied load in the technical parameters of the power distribution network and the power of each bus supplied load in the real-time running time into the nonlinear equation, and establishes an admittance matrix after adjusting the output power of each node in the technical parameters of the power distribution network to a set value, and then goes to S2;

[0020] S2 performs iterative power flow calculation by using a Newton iteration algorithm based on the admittance matrix, a set initial value of node voltage and a set number of iterations, and calculates the power imbalance rate of each node by using a graph data traversal method in the calculation process to obtain a convergence value of the power flow calculation and the power imbalance rate of each node, and then goes to S3;

[0021] S3 compares the convergence value with a convergence condition, if the convergence condition is satisfied, the iteration is ended, and goes to S5, otherwise goes to S4;

[0022] S4 corrects each node voltage in the nonlinear equation based on a Jacobian matrix and a correction equation, and obtains the corrected each node voltage by using a graph data traversal method, and takes the corrected each node voltage as the initial value of the node voltage to go to S2;

[0023] S5 outputs the output power of the balanced node and the power of each line as the power flow analysis result of the power distribution network.

[0024] Preferably, the clustering algorithm is used to perform clustering analysis on the user historical power consumption data of the power distribution network in the graph database of the power distribution network to obtain a user power consumption analysis result of the power distribution network, comprising:

[0025] The historical power consumption data of the power distribution network in the graph database is subjected to data cleaning, denoising and normalization processing to obtain processed historical power consumption data;

[0026] The processed historical power consumption data is sequentially input into an attraction degree matrix and a belonging degree matrix for calculation, and is subjected to attenuation by using an attenuation coefficient until a set number of iterations or a stability condition of the attraction degree matrix is reached to obtain a clustering result;

[0027] The clustering result is taken as an analysis result of power users of the power distribution network.

[0028] Preferably, the processed historical power consumption data is sequentially input into an attraction matrix and a belonging degree matrix for calculation, and an attenuation coefficient is used for attenuation until a set iteration number or a stable condition of the attraction matrix is reached, so as to obtain a clustering result, comprising:

[0029] S1 initializes the attraction matrix and the belonging degree matrix to 0, and enters S2;

[0030] S2 updates the attraction matrix and the belonging degree matrix by using each piece of processed historical power consumption data, and enters S3;

[0031] S3 attenuates the updated attraction matrix and the belonging degree matrix according to an attenuation coefficient, and enters S4;

[0032] S4 judges whether the attenuated attraction matrix is stable or whether a maximum iteration number is reached, if not, returns to S2, otherwise, outputs a clustering center, and distributes processed historical power consumption data of a non-clustering center to a corresponding cluster to obtain a clustering result.

[0033] Preferably, the graph data shortest path algorithm comprises a single source shortest path algorithm.

[0034] Based on the same inventive concept, the application further provides a power distribution network operation state analysis system based on graph data, comprising:

[0035] A graph database construction module is configured to convert the acquired technical parameters, historical and real-time operation data, user historical power consumption data, historical distributed power generation data, corresponding weather data and weather forecast data of the power distribution network into graph data by using a graph database construction method, and construct a power distribution network graph database.

[0036] A power distribution network operation state analysis module is configured to analyze the operation state of the power distribution network based on the power distribution network graph database by using a power distribution network parallel power flow analysis method, a power supply path analysis method, a user power consumption curve analysis method and a distributed power generation prediction data analysis method, and obtain an analysis result of the operation state of the power distribution network.

[0037] Preferably, the power distribution network operation state analysis module comprises:

[0038] A power flow analysis submodule is configured to calculate a power flow analysis result of the power distribution network by using the parallel power flow analysis method combined with a graph data traversal algorithm based on the technical parameters and real-time operation data of the power distribution network in the power distribution network graph database.

[0039] The power supply path analysis submodule is configured to perform power supply path analysis on the power distribution network technical parameters in the power distribution network graph database by using the graph data shortest path algorithm, and obtain power supply path analysis results of the power distribution network.

[0040] The user power consumption analysis submodule is configured to perform clustering analysis on the user historical power consumption data of the power distribution network in the power distribution network graph database by using a clustering algorithm, and obtain user power consumption analysis results of the power distribution network.

[0041] The distributed power generation prediction analysis submodule is configured to input the weather forecast data of the power distribution network into a pre-trained power distribution network distributed power generation data analysis prediction model, and obtain distributed power generation prediction analysis results of the power distribution network under corresponding weather forecast data.

[0042] The analysis result determination submodule is configured to determine the power flow analysis results of the power distribution network, the power supply path analysis results of the power distribution network, the user power consumption analysis results of the power distribution network, and the distributed power generation prediction analysis results of the power distribution network under corresponding weather forecast data as the operation state analysis results of the power distribution network.

[0043] Compared with the prior art, the present application has the following advantages:

[0044] 1. The present application provides a power distribution network operation state analysis method and system based on graph data, comprising: converting the obtained technical parameters, historical and real-time operation data, user historical power consumption data, historical distributed power generation data, and corresponding weather data and weather forecast data of the power distribution network into graph data by using a graph database construction method to construct a power distribution network graph database; based on the power distribution network graph database, using a power distribution network parallel power flow analysis method, a power supply path analysis method, a user power consumption curve analysis method, and a distributed power generation prediction data analysis method to analyze the operation state of the power distribution network, and obtaining the analysis results of the power distribution network operation state. The method provided by the present application can comprehensively and accurately analyze the operation state of the power distribution network at each time period.

[0045] 2. The present application fully utilizes the value of the power distribution system graph data, further improves the safe and economic operation level of the power distribution network, improves the customer service level, optimizes the level of clean energy consumption, and at the same time, lays a foundation for the marketization operation of the power distribution system and the construction of the comprehensive energy system. BRIEF DESCRIPTION OF DRAWINGS

[0046] Figure 1 FIG. 1 is a schematic diagram of a power distribution network operation state analysis method based on graph data according to the present application;

[0047] Figure 2 FIG. 2 is a structural diagram of a power distribution network operation state analysis system based on graph data according to the present application. DETAILED DESCRIPTION

[0048] Embodiment 1

[0049] In view of the fact that the matching degree of the graph database of the main network of the power system and the distribution network data is low in the prior art, the data correlation in the database corresponding to the distribution network is low, and the analysis method for analyzing the operation state of the distribution network according to the graph data is relatively one-sided, and it is difficult to comprehensively analyze the distribution network at each time period based on the graph data, the present application provides a distribution network operation state analysis method based on graph data, as shown in Figure 1 The method comprises the following steps:

[0050] Step 1: using a graph database construction method to convert the obtained technical parameters of the distribution network, historical and real-time operation data, user historical power consumption data, historical distributed power generation data and corresponding weather data and weather forecast data into graph data and then constructing a distribution network graph database;

[0051] Step 2: based on the distribution network graph database, using a distribution network parallel power flow analysis method, a power supply path analysis method, a user power consumption curve analysis method and a distributed power generation prediction data analysis method to analyze the operation state of the distribution network, and obtaining an analysis result of the operation state of the distribution network.

[0052] In step 1, the method comprises the following steps:

[0053] Obtaining the technical parameters of the distribution network, historical operation data, real-time operation data, user historical power consumption data, historical distributed power generation data and corresponding weather data and weather forecast data;

[0054] Using a graph database construction method to convert the obtained data into graph data and establishing a distribution network graph database;

[0055] In this embodiment, the device data table containing the technical parameters of the distribution network such as the substation, feeder, substation load, distribution network switch, distribution network bus, distribution network fuse, cable terminal, distribution network knife switch, distribution network transformer, cable section, distribution network grounding knife switch and compensator and the intermediate table containing the inter-table correlation relationship are exported from the database containing the related data of the distribution network, and the device table name is defined as the node type, each row of data in the device table is defined as the entity, the column in the device table is defined as the attribute, each field value in the device table is defined as the attribute value, and the related data in the intermediate table is defined as the connection edge between the entities, and the Neo4j database is used for data storage;

[0056] The data with time identifier such as the operation information of the distribution network equipment, the historical data of the power grid, the SCADA data, the intelligent electric meter and the user power consumption data after cleaning and denoising, and the subordinate relationship between the distribution network equipment are stored as the historical operation data of the distribution network;

[0057] Real-time device operation and smart meter data obtained by calling the real-time data interface of the device operation process in the power distribution network are stored as real-time operation data of the power distribution network.

[0058] Historical weather data obtained by the graph data processing technology and the weather data interface are stored as the historical distributed power generation data of the power distribution network and the corresponding weather data.

[0059] The weather forecast data are stored through the dynamic data storage mechanism of the graph data.

[0060] The graph database Neo4j is selected for the power distribution network data storage by virtue of the consistency in structure and the intuitiveness in display of the graph data structure and the actual power distribution network, thereby forming the power distribution network graph database.

[0061] In step 2, the operation state of the power distribution network is analyzed by using the power distribution network parallel power flow analysis method, the power supply path analysis method, the user power consumption curve analysis method and the distributed power generation prediction data analysis method based on the power distribution network graph database, and the analysis result of the operation state of the power distribution network is obtained.

[0062] Power flow analysis of the power distribution network: In the power distribution network, the power of the load supplied by each bus is known, and the voltage of each node except the balance node is unknown. The power flow calculation problem is converted into a nonlinear equation set problem. By using the real-time operation data of each device in the operation process of the power distribution network, the convergence value of the nonlinear equation set is solved by using the graph data traversal method and the Newton iteration method, thereby performing real-time power flow state calculation of the power distribution network, and obtaining the power flow analysis result of the power distribution network.

[0063] The specific analysis process is as follows:

[0064] S1 inputs the power of the load supplied by each bus in the real-time operation time and the power of the load supplied by each bus in the power distribution network technical parameters into the nonlinear equation, adjusts the output power of each node in the power distribution network technical parameters to a set value, establishes the admittance matrix, and enters S2;

[0065] S2 uses the admittance matrix, the set initial value of the node voltage as and f i (0) and the set iteration number k=0, performs iterative power flow calculation by using the Newton iteration algorithm, calculates the power imbalance rate of each node by using the graph data traversal method in the calculation process, obtains the convergence value of the power flow calculation and the power imbalance rate of each node, and enters S3;

[0066] S3 compares the convergence value with the convergence condition, if the convergence condition is met, the iteration is ended, and S5 is entered, otherwise S4 is entered;

[0067] S4, based on each element calculated by the Jacobian matrix, uses a correction equation to correct each node voltage in the nonlinear equation, and calculates the corrected node voltage by a graph data traversal method f i (j) and the corrected node voltage is taken as the initial value of the node voltage into S2.

[0068] S5 outputs the output power of the balanced node and the power of each line as the power flow analysis result of the distribution network.

[0069] Through the parallel power flow calculation of the distribution network based on the graph calculation method, the operating state of the entire distribution network system is determined, the load distribution of each line can be quickly understood, and further, the power flow analysis result can be used to prevent line overload or light load, to ensure the stable operation of the distribution network system, and through the adjustment of the operation mode, the entire distribution network resources can be fully utilized.

[0070] Supply path analysis: based on the topological connection mode and connection data of the technical parameters of the distribution network in the form of graph data, the shortest path algorithm based on vertex and edge query mode is used to analyze the supply path corresponding to each power supply point in the distribution network, and the supply path analysis result is obtained;

[0071] In this embodiment, the Single Source Shortest Path (SSSP) algorithm used is a single source shortest path algorithm implemented on the basis of the Dijkstra shortest path algorithm, which calculates the shortest path from the root node to all other nodes in the graph. Since there may be multiple supply paths for a power supply point in the distribution network, the SSSP algorithm can effectively find all the supply paths for a power supply point from the distribution network data model. In addition, for multiple different power supply points, the SSSP algorithm can also be used to obtain the supply path of each power supply point, realizing the full network tracking of important loads or key devices in the distribution transformer to the user power supply.

[0072] Based on the supply path analysis result, the power supply path of one or more distribution transformers or users in the distribution network can be determined, and further, solutions can be provided for grid fault monitoring, power outage area isolation, fault repair, power supply recovery and other applications.

[0073] User power consumption analysis: the historical power consumption data of the distribution network in the graph database is subjected to data cleaning, denoising and normalization processing to obtain the processed historical power consumption data;

[0074] The processed historical power consumption data is sequentially input into the attraction degree matrix and the attribution degree matrix for calculation, and an attenuation coefficient is used for attenuation until a set number of iterations or a stable condition of the attraction degree matrix is reached, and a clustering result is obtained.

[0075] The clustering result is taken as the power consumption user analysis result of the power distribution network.

[0076] In the embodiment, the user power consumption behavior analysis is performed using the affinity propagation (AP) based clustering algorithm. The basic idea of the algorithm is to regard all samples as nodes of a network, and then calculate the clustering center of each sample through message passing of each edge in the network. In the process, two kinds of messages are passed between nodes, namely, responsibility (attraction degree) and availability (affiliation degree). The algorithm continuously updates the attraction degree and affiliation degree values of each point in the iteration process until m high-quality centroids are generated, and the remaining data points are assigned to the corresponding clusters, and the clustering result is output.

[0077] The processed historical power consumption data are sequentially input to the attraction degree matrix and the affiliation degree matrix for calculation, and decay is performed using a decay coefficient until a set iteration number or a stable condition of the attraction degree matrix is reached, and the clustering result is obtained. The specific process is as follows:

[0078] S1 initializes the attraction degree matrix and the affiliation degree matrix to 0, and enters S2;

[0079] S2 updates the attraction degree matrix and the affiliation degree matrix using each of the processed historical power consumption data, and enters S3;

[0080] The attraction degree matrix is as follows:

[0081]

[0082] In the formula, r t+1 (i,k) represents the new attraction degree, r t (i,k) represents the attraction degree before update, a t (i,k) is the suitability of the i-th data selecting the k-th data as a clustering center, and S(i,k) represents the ability of the k-th data as a clustering center of the i-th data.

[0083] The affiliation degree matrix is as follows:

[0084]

[0085] In the formula, a t+1 (i,k) is the suitability of the i-th data selecting the k-th data as a clustering center at t+1, r t+1 (j,k) is the updated attraction degree, r t+1 (k,k) represents the suitability of the k-th data as a clustering center of the k-th data.

[0086] S3 attenuates the updated attraction degree matrix and the belonging degree matrix according to the attenuation coefficient, and enters S4; the updated attraction degree matrix and the belonging degree matrix are attenuated according to the attenuation coefficient, and the following formula is calculated:

[0087] r t+1 (i,k)=λ×r t (i,k)+(1-λ)×r t+1 (i,k)

[0088] a t+1 (i,k)=λ×a t (i,k)+(1-λ)×a t+1 (i,k)

[0089] In the formula, λ is an attenuation coefficient;

[0090] S4 judges whether the attenuated attraction degree matrix is stable or whether the maximum iteration number is reached, if not, returns to S2, otherwise, the data with the maximum sum of attraction degree and belonging degree is taken as the clustering center, and the remaining data points are assigned to the corresponding cluster to obtain the user power consumption curve clustering analysis result output clustering center, and the processed historical power consumption data of the non-clustering center is assigned to the corresponding cluster to obtain the clustering result.

[0091] Based on the power consumption user analysis result of the power distribution network, a plurality of different power consumption categories can be obtained, so as to obtain the power consumption type most similar to the power consumption behavior of each user, and the user power consumption load can be further predicted, thereby helping the operation and dispatch personnel to perform offline analysis on the power grid and laying a data foundation for accurate demand response.

[0092] Before the prediction and analysis of the distributed power generation, a distributed power generation data analysis prediction model of the power distribution network is pre-trained, and the specific steps are as follows:

[0093] The historical distributed power generation data and the corresponding weather data vertices of the power distribution network are converted into vector data by using the graph vectorization technology Graph Embedding;

[0094] The historical distributed power generation data and the corresponding weather vector data are input into the graph neural network GNN to automatically extract the vertex features, obtain the vertex feature vectors of the historical distributed power generation data and the corresponding weather data, and take the vertex feature vectors as the training set of the model;

[0095] Based on the training set of the model, the vertex feature vectors of the weather data corresponding to the historical distributed power generation data are taken as the input, and the vertex feature vectors of the historical distributed power generation data are taken as the output to train the model, so as to obtain the distributed power generation data analysis prediction model of the power distribution network.

[0096] Distributed power generation prediction analysis: input the weather forecast data of the power distribution network into a power distribution network distributed power generation data analysis and prediction model pre-trained based on historical distributed power generation data and corresponding weather data in a power distribution network graph database to obtain distributed power generation prediction analysis results of the power distribution network under corresponding weather forecast data;

[0097] In the embodiment, graph machine learning is used for distributed photovoltaic power station power generation prediction, the purpose is to train the model by using numerical weather forecast data, photovoltaic power station unit data, weather live data and AGC limit adjustment data and other indicators, so as to perform photovoltaic power station actual output prediction; the actual output value of the photovoltaic power station is predicted by combining the graph neural network (GNN) and the long short-term memory neural network (LSTM), the GNN is used to automatically extract the features of each node in the graph database, and then the LSTM is used for feature fusion and prediction;

[0098] The specific process of the distributed power generation prediction analysis is as follows:

[0099] The weather forecast data is converted into vector data by using the graph embedding technology;

[0100] The weather forecast vector data is input into the graph neural network GNN to automatically extract the vertex features, and the vertex feature vector X=(x1, x2,…, x n ) of the weather forecast vector data is obtained. n

[0101] The vertex feature vector of the weather forecast data is input into the pre-trained power distribution network distributed power generation data analysis and prediction model to perform feature fusion and result prediction, and the distributed power generation prediction analysis result is obtained.

[0102] The power distribution network distributed power generation data analysis and prediction model based on the LSTM includes an input gate, a forgetting gate and an output gate.

[0103] The forgetting gate is as follows:

[0104] f t =σ(W f ·[h t-1 ,x t ]+b f )

[0105] In the formula, f t is the forgetting gate, W f ​is a bias term for the forget gate, h f is a bias term for the forget gate, h t-1 is the hidden layer output at time t-1, x t is the input at current time t, and σ is a nonlinear activation function,

[0106] The input gate is shown as follows:

[0107] i t = σ(W i · [h t-1 , x t ] + b i )

[0108]

[0109] In the formula, i t is the input gate, W i is a weight term for the input gate, b i is a bias term for the input gate, and tanh is a nonlinear activation function, is the cell state at current time t, W C is a weight for the cell state, b C is a bias for the cell state,

[0110] The output gate is shown as follows:

[0111] o t = σ(W o · [h t-1 , x t ] + b o )

[0112] h t = o t · tanh(C t )

[0113] In the formula, o t is the output gate, W o is a weight term for the output gate, b o is a bias term for the input gate, C t is the new cell state at current time t, h t is the output information of the hidden layer at current time.

[0114] Based on the prediction and analysis results of distributed power generation, it is helpful for the power system dispatching department to arrange the coordination of conventional energy and photovoltaic power generation, and timely adjust the dispatching plan, and reasonably and safely operate the power grid.

[0115] To fully explore data information and better utilize the value of data assets, this invention proposes a distribution network operation status analysis method based on graph data. Addressing issues such as disjointed cross-disciplinary processes, weak real-time data sharing, and insufficient data value mining in distribution systems, this method leverages the inherent graphical nature of the distribution network, formed by the interconnection of distribution equipment and the distribution network itself. It transforms the network topology and operational data into a graph structure, uses a graph database for data storage, and constructs a graph database for the distribution network. This database comprehensively links various distribution equipment, management elements, and historical measurement data. Based on this, it utilizes graph data-based methods for parallel power flow analysis, power supply path analysis, user electricity consumption curve analysis, and distributed generation prediction data analysis to provide a more comprehensive and accurate analysis of the distribution network's operation status at different time periods. The analysis results can then be applied in practice to improve the service and management level of regional power supply companies.

[0116] Example 2:

[0117] Based on the same inventive concept, this invention also provides a distribution network operation status analysis system based on graph data, such as... Figure 2 As shown, it includes:

[0118] The graph data construction module is used to convert the acquired technical parameters, historical and real-time operation data, user historical electricity consumption data, historical distributed generation data, and corresponding weather data and weather forecast data of the distribution network into graph data to construct the distribution network graph database using graph database construction methods.

[0119] The distribution network operation status analysis submodule is used to analyze the operation status of the distribution network based on the distribution network graph database, using methods such as distribution network parallel power flow analysis, power supply path analysis, user electricity consumption curve analysis, and distributed generation prediction data analysis, and to obtain the analysis results of the distribution network operation status.

[0120] The graph database construction module includes: a data acquisition submodule and a database construction submodule;

[0121] The data acquisition submodule is configured to acquire technical parameters of the power distribution network, historical operation data, real-time operation data, user historical power consumption data, historical distributed power generation data, corresponding weather data, and weather forecast data; export, from a database in which relevant data of the power distribution network are stored, equipment data tables of substations, feeder lines, substation loads, power distribution network switches, power distribution network busbars, power distribution network fuses, cable terminals, power distribution network switches, power distribution network transformers, cable sections, power distribution network grounding switches, compensators, and the like, which contain technical parameters of the power distribution network, and intermediate tables containing inter-table association relationships, and define the equipment table name as a node type, each row of data in the equipment table as an entity, columns in the equipment table as attributes, and each field value in the equipment table as an attribute value, and the relevant data in the intermediate table as a connection edge between entities, and use a Neo4j database to store data;

[0122] The data acquisition submodule is configured to acquire technical parameters of the power distribution network, historical operation data, real-time operation data, user historical power consumption data, historical distributed power generation data, corresponding weather data, and weather forecast data; export, from a database in which relevant data of the power distribution network are stored, equipment data tables of substations, feeder lines, substation loads, power distribution network switches, power distribution network busbars, power distribution network fuses, cable terminals, power distribution network switches, power distribution network transformers, cable sections, power distribution network grounding switches, compensators, and the like, which contain technical parameters of the power distribution network, and intermediate tables containing inter-table association relationships, and define the equipment table name as a node type, each row of data in the equipment table as an entity, columns in the equipment table as attributes, and each field value in the equipment table as an attribute value, and the relevant data in the intermediate table as a connection edge between entities, and use a Neo4j database to store data;

[0123] The data acquisition submodule is configured to acquire technical parameters of the power distribution network, historical operation data, real-time operation data, user historical power consumption data, historical distributed power generation data, corresponding weather data, and weather forecast data; export, from a database in which relevant data of the power distribution network are stored, equipment data tables of substations, feeder lines, substation loads, power distribution network switches, power distribution network busbars, power distribution network fuses, cable terminals, power distribution network switches, power distribution network transformers, cable sections, power distribution network grounding switches, compensators, and the like, which contain technical parameters of the power distribution network, and intermediate tables containing inter-table association relationships, and define the equipment table name as a node type, each row of data in the equipment table as an entity, columns in the equipment table as attributes, and each field value in the equipment table as an attribute value, and the relevant data in the intermediate table as a connection edge between entities, and use a Neo4j database to store data;

[0124] The data acquisition submodule is configured to acquire technical parameters of the power distribution network, historical operation data, real-time operation data, user historical power consumption data, historical distributed power generation data, corresponding weather data, and weather forecast data; export, from a database in which relevant data of the power distribution network are stored, equipment data tables of substations, feeder lines, substation loads, power distribution network switches, power distribution network busbars, power distribution network fuses, cable terminals, power distribution network switches, power distribution network transformers, cable sections, power distribution network grounding switches, compensators, and the like, which contain technical parameters of the power distribution network, and intermediate tables containing inter-table association relationships, and define the equipment table name as a node type, each row of data in the equipment table as an entity, columns in the equipment table as attributes, and each field value in the equipment table as an attribute value, and the relevant data in the intermediate table as a connection edge between entities, and use a Neo4j database to store data;

[0125] The data acquisition submodule is configured to acquire technical parameters of the power distribution network, historical operation data, real-time operation data, user historical power consumption data, historical distributed power generation data, corresponding weather data, and weather forecast data; export, from a database in which relevant data of the power distribution network are stored, equipment data tables of substations, feeder lines, substation loads, power distribution network switches, power distribution network busbars, power distribution network fuses, cable terminals, power distribution network switches, power distribution network transformers, cable sections, power distribution network grounding switches, compensators, and the like, which contain technical parameters of the power distribution network, and intermediate tables containing inter-table association relationships, and define the equipment table name as a node type, each row of data in the equipment table as an entity, columns in the equipment table as attributes, and each field value in the equipment table as an attribute value, and the relevant data in the intermediate table as a connection edge between entities, and use a Neo4j database to store data;

[0126] The database construction submodule is configured to convert the acquired data into graph data and establish a power distribution network graph database by using a graph database construction method; and use the graph data structure and the consistency in structure and the intuitiveness in display of the actual power distribution network to select a graph database Neo4j for power distribution network data storage, thereby forming the power distribution network graph database.

[0127] The power distribution network operation state analysis module includes a power flow analysis submodule, a power supply path analysis submodule, a user power consumption analysis submodule, a distributed power generation prediction analysis submodule, and an analysis result determination submodule.

[0128] The power flow analysis submodule is configured to calculate power flow analysis results of the power distribution network by using the parallel power flow analysis method combined with a graph data traversal algorithm based on the technical parameters and real-time operation data of the power distribution network in the power distribution network graph database.

[0129] The power flow analysis submodule has the following specific working process:

[0130] S1 inputs the load supplied by each bus in the technical parameters of the power distribution network and the power of the load supplied by each bus in the real-time running time into the non-linear equation, adjusts the output power of each node in the technical parameters of the power distribution network to a set value, and then establishes an admittance matrix to enter S2;

[0131] S2 sets the initial value of the node voltage as and f i (0) and the iteration number as k = 0, performs iterative power flow calculation by using the Newton iteration algorithm, calculates the power imbalance rate of each node by using the graph data traversal method in the calculation process, and enters S3 after obtaining the convergence value of the power flow calculation and the power imbalance rate of each node;

[0132] S3 compares the convergence value with the convergence condition, if the convergence condition is met, the iteration is ended, and S5 is entered, otherwise S4 is entered;

[0133] The correction unit is configured to correct each node voltage in the non-linear equation by using a correction equation based on each element calculated by the Jacobian matrix, and calculate the corrected each node voltage by using the graph data traversal method f i (j) , and enter S2 with the corrected each node voltage as the initial value of the node voltage;

[0134] S5 outputs the output power of the balanced node and the power of each line as the power flow analysis result of the power distribution network.

[0135] The power supply path analysis submodule is configured to perform power supply path analysis of each power supply point in the power distribution network by using a shortest path algorithm based on vertex and edge query based on the topological connection mode and connection data in the technical parameters of the power distribution network in the form of the graph data, and obtain a power supply path analysis result;

[0136] The user power consumption analysis submodule includes a normalization processing unit, a clustering unit, and a power consumption user analysis result determination unit.

[0137] The normalization processing unit is configured to perform data cleaning, denoising, and normalization processing on the historical power consumption data of the power distribution network in the graph database to obtain processed historical power consumption data.

[0138] The clustering unit is configured to input the processed historical power consumption data into an attraction degree matrix and a belonging degree matrix in sequence for calculation, and perform attenuation by using an attenuation coefficient until a set iteration number or a stability condition of the attraction degree matrix is reached, to obtain a clustering result.

[0139] The power consumption user analysis result determination unit is configured to take the clustering result as the power consumption user analysis result of the power distribution network.

[0140] The specific working process of the clustering unit is as follows:

[0141] S1 initializes the attraction degree matrix and the belonging degree matrix to 0, and enters S2;

[0142] S2 updates the attraction degree matrix and the belonging degree matrix by using each of the processed historical power consumption data, and enters S3;

[0143] The attraction degree matrix is as follows:

[0144]

[0145] In the formula, r t+1 (i,k) represents the new attraction degree, r t (i,k) represents the attraction degree before updating, a t (i,k) is the suitability of the i-th data selecting the k-th data as a clustering center, and S(i,k) represents the ability of the k-th data as the clustering center of the i-th data;

[0146] The belonging degree matrix is as follows:

[0147]

[0148] In the formula, a t+1 (i,k) is the suitability of the i-th data selecting the k-th data as a clustering center at the t+1 time, r t+1 (j,k) is the updated attraction degree, r t+1 (k,k) represents the suitability of the k-th data as the clustering center of the k-th data;

[0149] S3 attenuates the updated attraction degree matrix and the belonging degree matrix according to the attenuation coefficient, and enters S4;

[0150] The updated attraction degree matrix and the belonging degree matrix are attenuated according to the attenuation coefficient, and are calculated according to the following formula:

[0151] r t+1 (i,k) = λ × r t (i,k) + (1-λ) × r t+1 (i,k)

[0152] a t+1 (i,k) = λ × a t (i,k) + (1-λ) × a t+1 (i,k)

[0153] In the formula, λ is the attenuation coefficient;

[0154] S4 judges whether the decayed attraction degree matrix is stable or whether the maximum iteration number is reached, if not, returns to S2, otherwise, the data with the maximum sum of attraction degree and belonging degree is taken as the clustering center, and the remaining data points are assigned to the corresponding clusters to obtain the user electricity consumption curve clustering analysis result output clustering center, and the processed historical electricity consumption data of non-clustering center is assigned to the corresponding cluster to obtain the clustering result.

[0155] The distributed power generation prediction analysis submodule pre-trains a distributed power generation data analysis prediction model of the power distribution network before analyzing the power distribution network, and the specific steps are as follows:

[0156] The historical distributed power generation data and corresponding weather data vertices of the power distribution network are converted into vector data by using a graph embedding technology.

[0157] The historical distributed power generation data and corresponding weather vector data are input into a graph neural network GNN to automatically extract vertex features, obtain vertex feature vectors of the historical distributed power generation data and corresponding weather data, and take the vertex feature vectors as a training set of the model.

[0158] Based on the training set of the model, the vertex feature vectors of the weather data corresponding to the historical distributed power generation data are taken as input, and the vertex feature vectors of the historical distributed power generation data are taken as output to train the model, and a distributed power generation data analysis prediction model of the power distribution network is obtained.

[0159] The distributed power generation prediction analysis submodule is used to input the weather forecast data of the power distribution network into the distributed power generation data analysis prediction model of the power distribution network which is pre-trained based on the historical distributed power generation data and corresponding weather data in the power distribution network graph database, to obtain a distributed power generation prediction analysis result of the power distribution network under the corresponding weather forecast data.

[0160] The distributed power generation prediction analysis submodule comprises a vectorization unit, a feature extraction unit, and a prediction unit.

[0161] The vectorization unit is used to convert the weather forecast data into vector data by using a graph embedding technology.

[0162] The feature extraction unit is used to input the weather forecast vector data into a graph neural network GNN to automatically extract vertex features, and obtain vertex feature vectors X=(x1, x2, …, x n ) of the weather forecast vector data. n

[0163] ​The prediction unit is configured to input the vertex feature vector of the weather forecast data into a pre-trained distribution power grid distributed power generation data analysis prediction model, perform feature fusion and result prediction, and obtain a distributed power generation prediction analysis result.

[0164] The LSTM-based distribution power grid distributed power generation data analysis prediction model comprises an input gate, a forgetting gate and an output gate.

[0165] The forgetting gate is shown in the following formula:

[0166] f t = σ (W f · [h t-1 , x t ] + b f )

[0167] In the formula, f t is the forgetting gate, W f is the weight term of the forgetting gate, b f is the bias term of the forgetting gate, h t-1 is the output of the hidden layer at t-1, x t is the input at the current t, and σ is a nonlinear activation function.

[0168] The input gate is shown in the following formula:

[0169] i t = σ (W i · [h t-1 , x t ] + b i )

[0170]

[0171] In the formula, i t is the input gate, W i is the weight term of the input gate, b i is the bias term of the input gate, tanh is a nonlinear activation function, is the cell state at the current t, W C is the weight of the cell state, and b C is the bias of the cell state.

[0172] The output gate is shown in the following formula:

[0173] o t = σ (W o · [h t-1 , x t ] + b o )

[0174] h t = o t · tanh (Ct )

[0175] wherein o t is an output gate, W o is a weight term for the output gate, b o is a bias term for the input gate, C t is the new cell state at the current time t, h t is the output information of the hidden layer at the current time.

[0176] Obviously, the described embodiments are only part of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present application.

[0177] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage, etc.) containing computer-usable program code.

[0178] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing apparatus to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing apparatus produce a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in a flow or multiple flows and / or blocks Figure 1 The functions specified in a flow or multiple flows and / or blocks

[0179] These computer program instructions can also be stored in a computer-readable memory that can direct the computer or other programmable data processing apparatus to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including instruction devices that implement the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in a flow or multiple flows and / or blocks Figure 1 The functions specified in a flow or multiple flows and / or blocks

[0180] These computer program instructions can also be loaded into a computer or other programmable data processing devices, so that a series of operational steps are performed on the computer or other programmable data processing devices to generate computer-implemented processes, thus the instructions executed on the computer or other programmable data processing devices provide processes for implementing the functions specified in the flowchart Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0181] The above merely describes the embodiments of the present application, and is not intended to limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the scope of the claims of the present application.

Claims

1. A power distribution network operating state analysis method based on graph data, characterized by, The method comprises the following steps: The technical parameters, historical and real-time operation data, user historical power consumption data, historical distributed power generation data, corresponding weather data and weather forecast data of the power distribution network are converted into graph data by using a graph database construction method, and a power distribution network graph database is constructed; Based on the power distribution network graph database, the operation state of the power distribution network is analyzed by using a power distribution network parallel power flow analysis method, a power supply path analysis method, a user power consumption curve analysis method and a distributed power generation prediction data analysis method, and an analysis result of the operation state of the power distribution network is obtained; The analysis result of the operation state of the power distribution network is obtained by analyzing the operation state of the power distribution network based on the power distribution network graph database by using the power distribution network parallel power flow analysis method, the power supply path analysis method, the user power consumption curve analysis method and the distributed power generation prediction data analysis method, and the analysis result of the operation state of the power distribution network comprises: Based on the technical parameters and real-time operation data of the power distribution network in the power distribution network graph database, the parallel power flow analysis method is used in combination with a graph data traversal algorithm to calculate a power flow analysis result of the power distribution network; The power supply path analysis result of the power distribution network is obtained by using a graph data shortest path algorithm to analyze the power supply path of the power distribution network technical parameters in the power distribution network graph database; The user power consumption analysis result of the power distribution network is obtained by using a clustering algorithm to analyze the user historical power consumption data of the power distribution network in the power distribution network graph database; The distributed power generation prediction analysis result of the power distribution network under corresponding weather forecast data is obtained by inputting the weather forecast data of the power distribution network into a pre-trained power distribution network distributed power generation data analysis prediction model; The power flow analysis result of the power distribution network, the power supply path analysis result of the power distribution network, the user power consumption analysis result of the power distribution network and the distributed power generation prediction analysis result of the power distribution network under corresponding weather forecast data are taken as the operation state analysis result of the power distribution network; The power flow analysis result of the power distribution network is calculated based on the technical parameters and real-time operation data of the power distribution network in the power distribution network graph database by using the parallel power flow analysis method in combination with a graph data traversal algorithm, and the power flow analysis result of the power distribution network comprises: A nonlinear equation of power flow of the power distribution network is established based on the technical parameters and real-time operation data of the power distribution network; The nonlinear equation is solved by using a graph data traversal algorithm in combination with a Newton iteration algorithm, a convergence value of the nonlinear equation satisfying a convergence condition is obtained, and the balanced node output power and the power of each line corresponding to the convergence value of the nonlinear equation satisfying the convergence condition are taken as the power flow analysis result of the power distribution network; The nonlinear equation is solved by using a graph data traversal algorithm in combination with a Newton iteration algorithm, a convergence value of the nonlinear equation satisfying a convergence condition is obtained, and the balanced node output power and the power of each line corresponding to the convergence value of the nonlinear equation satisfying the convergence condition are taken as the power flow analysis result of the power distribution network, and the method comprises the following steps: S1: The power of each bus supplied load in the technical parameters of the power distribution network and the real-time operation time is input into the nonlinear equation, and the output power of each node in the power distribution network technical parameters is adjusted to a set value, and then a conductance matrix is established, and S2 is entered; S2 performs iterative power flow calculation by using a Newton iterative algorithm based on the admittance matrix, the set initial value of the node voltage, and the set number of iterations, and calculates the power imbalance rate of each node by using a graph data traversal method in the calculation process, and enters S3 after obtaining the convergence value of the power flow calculation and the power imbalance rate of each node; S3 compares the convergence value with a convergence condition, and if the convergence condition is met, ends the iteration and enters S5, otherwise enters S4; S4 corrects each node voltage in the nonlinear equation based on the Jacobian matrix and the correction equation, calculates the corrected node voltage by using the graph data traversal method, and enters S2 with the corrected node voltage as the initial value of the node voltage; S5 outputs the output power of the balanced node and the power of each line as the power flow analysis result of the distribution network; The clustering algorithm is used to analyze the user historical power consumption data of the distribution network in the distribution network graph database to obtain the user power consumption analysis result of the distribution network, including: The historical power consumption data in the graph database is subjected to data cleaning, denoising, and normalization processing to obtain processed historical power consumption data; The processed historical power consumption data is sequentially input into an attraction degree matrix and a belonging degree matrix for calculation, and is attenuated by using an attenuation coefficient until a set number of iterations or a stable condition of the attraction degree matrix is reached to obtain a clustering result; The clustering result is taken as the power consumption user analysis result of the distribution network.

2. The method of claim 1, wherein, The training of the distribution network distributed power generation data analysis and prediction model includes: The vertex feature vectors are obtained by using graph vectorization technology and graph neural networks to process the vertex features of the historical distributed power generation data and the corresponding weather data of the distribution network in the distribution network graph database, and the vertex feature vectors are taken as the training set of the distribution network distributed power generation data analysis and prediction model; The distribution network distributed power generation data analysis and prediction model is trained based on the training set of the distribution network distributed power generation data analysis and prediction model, with the vertex feature vectors of the weather data corresponding to the historical distributed power generation data as the input and the vertex feature vectors of the historical distributed power generation data as the output, to obtain the trained distribution network distributed power generation data analysis and prediction model.

3. The method of claim 1, wherein, The processed historical power consumption data is sequentially input into an attraction degree matrix and a belonging degree matrix for calculation, and is attenuated by using an attenuation coefficient until a set number of iterations or a stable condition of the attraction degree matrix is reached to obtain a clustering result, including: S1 initializes the attraction degree matrix and the belonging degree matrix to 0, and enters S2; S2 updates the attraction degree matrix and the belonging degree matrix by using each processed historical power consumption data, and enters S3; S3 attenuates the updated attraction degree matrix and the belonging degree matrix according to the attenuation coefficient, and enters S4; S4 judges whether the attenuated attraction degree matrix is stable or whether the maximum number of iterations is reached, and if not, returns to S2, otherwise, outputs the clustering center and distributes the processed historical power consumption data of the non-clustering center to the corresponding cluster to obtain the clustering result.

4. The method of claim 1, wherein, The graph data shortest path algorithm comprises a single-source shortest path algorithm.

5. A power distribution network operating state analysis system based on graph data, for the method of claim 1, characterized in that, Comprise: The graph database construction module is configured to convert the technical parameters, historical and real-time operation data, user historical power consumption data, historical distributed power generation data, corresponding weather data, and weather forecast data of the power distribution network into graph data, and then construct a power distribution network graph database using a graph database construction method. The power distribution network operation state analysis module is configured to analyze the operation state of the power distribution network based on the power distribution network graph database and using a power distribution network parallel power flow analysis method, a power supply path analysis method, a user power consumption curve analysis method, and a distributed power generation prediction data analysis method, to obtain an analysis result of the operation state of the power distribution network.

6. The system of claim 5, wherein, The power distribution network operation state analysis module comprises: The power flow analysis submodule is configured to calculate a power flow analysis result of the power distribution network based on the technical parameters and real-time operation data of the power distribution network in the power distribution network graph database and using the parallel power flow analysis method combined with a graph data traversal algorithm. The power supply path analysis submodule is configured to analyze the power supply path of the power distribution network based on the technical parameters of the power distribution network in the power distribution network graph database and using a graph data shortest path algorithm, to obtain a power supply path analysis result of the power distribution network. The user power consumption analysis submodule is configured to analyze the user historical power consumption data of the power distribution network in the power distribution network graph database using a clustering algorithm, to obtain a user power consumption analysis result of the power distribution network. The distributed power generation prediction analysis submodule is configured to input the weather forecast data of the power distribution network into a pre-trained power distribution network distributed power generation data analysis and prediction model, to obtain a distributed power generation prediction analysis result of the power distribution network under the corresponding weather forecast data. The analysis result determination submodule is configured to determine the power flow analysis result of the power distribution network, the power supply path analysis result of the power distribution network, the user power consumption analysis result of the power distribution network, and the distributed power generation prediction analysis result of the power distribution network under the corresponding weather forecast data as the operation state analysis result of the power distribution network. The user power consumption analysis submodule is configured to analyze the user historical power consumption data of the power distribution network in the power distribution network graph database using a clustering algorithm, to obtain a user power consumption analysis result of the power distribution network. The historical power consumption data in the graph database is subjected to data cleaning, denoising, and normalization processing to obtain processed historical power consumption data. The processed historical power consumption data is sequentially input into an attraction degree matrix and a belonging degree matrix for calculation, and an attenuation coefficient is used for attenuation until a set number of iterations or a stable condition of the attraction degree matrix is reached, to obtain a clustering result. The clustering result is used as the power consumption user analysis result of the power distribution network.

Images