Low-voltage distribution area topology identification method and system based on clustering analysis and graph theory

Abstract

The invention discloses a low-voltage power distribution area topology identification method and system based on clustering analysis and a graph theory, and relates to the technical field of low-voltage power distribution area topology classification. The system can accurately identify the topological structure of each partition through the combination of a clustering analysis model and a graph theory algorithm, and automatically calculate a topological characteristic coefficient Tp; therefore, the complex topological structure of the power distribution area can be classified and analyzed more carefully and comprehensively. According to a traditional method, accurate classification is often difficult when a complex branch structure is faced, and a topology recognition module ensures the accuracy of a classification result through a calculation mode of a graph theory. Through the feedback management module, when the power performance risk coefficient Lpjx of the partition exceeds a safety threshold R, a fault early warning signal can be sent out in time, abnormal nodes are automatically identified, faults are isolated, meanwhile, load transfer is executed, the power demand is rapidly transferred to the normally operating partition, and the power failure range and time are reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of low-voltage distribution station area topology classification, and in particular to a low-voltage distribution station area topology identification method and system based on cluster analysis and graph theory. Background Art

[0002] In the operation of modern power systems, the distribution network is an important link in power transmission, especially the operating stability of low-voltage distribution stations is directly related to the reliability of power supply to users.

[0003] Traditional topology identification and classification usually rely on manual inspection and maintenance, requiring technicians to identify the topology of the distribution network through on-site inspection and drawing analysis. This method is inefficient, especially in complex distribution networks, where identification and classification requires a lot of time and manpower.

[0004] Secondly, with the increase of distributed energy access and emerging power-consuming devices such as electric vehicles, the load distribution in low-voltage distribution areas has become increasingly complex, and traditional topology identification methods can no longer meet the needs of intelligent management. For example, with the increase of emerging power-consuming devices and the dynamic changes in the load of the distribution network, traditional load assessment is mainly based on the average value within a fixed period of time, which cannot accurately capture the load imbalance between nodes and the instantaneous changes in electrical parameters.

[0005] Therefore, there is an urgent need for a low-voltage distribution substation area topology identification method and system based on cluster analysis and graph theory to improve the management level and emergency response capability of distribution substation areas. Summary of the invention

[0006] In view of the deficiencies in the prior art, the present invention provides a low-voltage distribution station area topology identification method and system based on cluster analysis and graph theory to solve the problems mentioned in the background technology.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions: A low-voltage distribution station area topology identification system based on cluster analysis and graph theory, comprising:

[0008] The data acquisition module is used to inspect the distribution lines by installing infrared imaging equipment using drones, collect and identify the line length, number of branches and branch equipment, and collect the electrical parameters of each node to establish a node database;

[0009] The initial partitioning module is used to extract the geographical location and electrical parameters of the electrical equipment in the node database to perform node clustering, establish a clustering analysis model, and input the node database into the clustering analysis model to divide it into several initial partitions through the clustering analysis model;

[0010] A topology recognition module is used to establish a partition node network diagram based on a node database, and construct a topological characteristic coefficient Tp for each initial partition; and further mark the topological structure category of the initial partition according to the topological characteristic coefficient Tp;

[0011] The load analysis module extracts the connection node current phase angle difference Xwjc, node harmonic content Xbhl, node instantaneous current Ssdl, node switching voltage drop value Dyz, node fault frequency Gzpl and load balance coefficient Fh from the electrical parameters of each initial partition, predicts the load growth trend through model training, and calculates and generates the partition power performance risk coefficient Lpjx within the partition;

[0012] The feedback management module is used to compare and analyze the partition power performance risk coefficient Lpjx with the pre-set safety threshold R to determine whether there is any abnormality in the operation of equipment in each topological partition within the current low-voltage distribution station area. When the partition power performance risk coefficient Lpjx exceeds the safety threshold R, a risk warning signal is issued and displayed in real time on the user monitoring interface. At the same time, fault isolation and load transfer emergency processing plans are executed through remote monitoring means.

[0013] Preferably, the data acquisition module includes a first acquisition unit;

[0014] The first acquisition unit is used to inspect the distribution line by installing infrared imaging equipment through a drone, collect and identify the line length, the number of branches and branch equipment, obtain inspection data, and identify the inspection data acquisition with each mortgaged distribution transformer as the center, and identify and obtain the central node, branch node, intermediate node and terminal node;

[0015] The central node includes the location of the transformer or main switchgear, which serves as the power supply center for the entire substation area;

[0016] Branch nodes include branch switches or branch transformers to form new branches, which are important nodes in the distribution network;

[0017] Central nodes include connection points or support points distributed in the line. Connection points or support points include poles or brackets, which are used to connect different line sections and are key structural nodes of the line.

[0018] The end node includes the user's electricity meter and is the terminal node of the distribution line.

[0019] Preferably, the data acquisition module further includes a second acquisition unit, and the second acquisition unit is used to collect electrical parameters of each node;

[0020] The electrical parameters include: connection node current phase angle difference Xwjc, node harmonic content Xbhl, node instantaneous current Ssdl, node switching voltage drop value Dyz and node fault frequency Gzpl.

[0021] Preferably, the initial partitioning module includes a data preprocessing unit and a cluster partitioning unit;

[0022] The data preprocessing unit is used to preprocess the node database and standardize the processed related data information according to the dimensionless processing technology to unify the units;

[0023] The cluster partition unit is used to identify sub-areas within the station area through the initial partition tree and divide the monitoring area according to the preset hierarchical level. The specific steps are as follows:

[0024] S1. Determine whether to merge or split nodes by using Euclidean distance. After selecting the distance measurement method, calculate the distance between each node and other nodes to form a distance matrix D. The specific definition formula is:

[0025] D=[d ij ];

[0026]

[0027] Where D is an n×n distance matrix, which represents the distance matrix between nodes, and d ij represents the distance between node i and node j; x ik and x jk They represent the values ​​of node i and node j in the kth dimension, including voltage, current, longitude and latitude; p represents the number of dimensions of the feature;

[0028] S2, based on several distance matrices, each node is treated as a separate cluster, based on d ij Represents the distance between node i and node j, which is merged using the following distance measurement method:

[0029] S21, minimum distance method: select the distance between the nearest nodes of the clusters:

[0030]

[0031] S22, Maximum distance method: Select the distance of the farthest node between clusters:

[0032]

[0033] S23, average distance method: select the average distance between all nodes in the cluster:

[0034]

[0035] S3. Find the two clusters C with the smallest distance A and C B , merge them into a new cluster Repeat steps S2 and S3 to gradually merge the closest clusters until the preset number of clusters or distance threshold is reached, and finally several initial partitions are formed, each of which represents a relatively independent distribution station area topology substructure.

[0036] Preferably, the topology identification module includes a network graph theory node unit and a calculation unit;

[0037] The network graph theory node unit is used to establish a partition node network graph based on the node database set, and construct the topological characteristic coefficient Tp of each initial partition through the calculation unit. The topological characteristic coefficient Tp is calculated and obtained by the following formula:

[0038]

[0039] Where N is the total number of nodes in the network, M is the total number of edges in the network, F is the clustering coefficient, and E is i Indicates the actual number of edges between the i-th node and the nodes, K i represents the number of neighboring nodes connected to the i-th node, B represents the node betweenness, σ st represents the total number of shortest paths from node s to node t, σ st (i) represents the number of shortest paths from node s to node t through node i; L represents the average path length, d ij represents the distance between node i and node j; H i represents the number of edges at the i-th node, ZK J represents the line impedance value of the Jth edge, kdz J represents the line length of the Jth edge, P J represents the line load of the Jth edge, Y represents the redundant distance of the partition, sjlj represents the actual path length of the partition, zdlj represents the shortest path length of the partition; a1, a2, a3, a4, a5 and a6 are weight coefficients.

[0040] Preferably, the topology recognition module further includes a classification unit, which is used to preset a first threshold value Q1, a second threshold value Q2 and a third threshold value Q3, and compare the topological characteristic coefficient Tp of each initial partition with the first threshold value Q1, the second threshold value Q2 and the third threshold value Q3, respectively, to obtain corresponding classification results, including:

[0041] When the topological characteristic coefficient Tp> the first threshold Q1, the initial partition is divided into a ring network structure category;

[0042] When the second threshold Q2 < the topological characteristic coefficient Tp ≤ the first threshold Q1, the initial partition is divided into a tree structure category;

[0043] When the third threshold value Q3 < the topological characteristic coefficient Tp ≤ the second threshold value Q2, the initial partition is divided into a chain structure category;

[0044] When the topological characteristic coefficient Tp≤the third threshold Q3, the initial partition is divided into a radial structure category.

[0045] Preferably, the load analysis module includes a load collection unit and a prediction unit;

[0046] The load collection unit is used to collect the actual load value P of each initial partition. actual and rated load value P rated , the load balance coefficient Fh is calculated by the following formula:

[0047]

[0048] The prediction unit is used to extract the connection node current phase angle difference Xwjc, node harmonic content Xbhl, node instantaneous current Ssdl, node switching voltage drop value Dyz and node fault frequency Gzpl in the electrical parameters, and associate the load balance coefficient Fh. After dimensionless processing, the partition power performance risk coefficient Lpjx is calculated by the following formula:

[0049]

[0050] Wherein, a1, a2, a3, a4, a5 and a6 represent the weight coefficients of the connection node current phase angle difference Xwjc, the node harmonic content Xbhl, the node instantaneous current Ssdl, the node switching voltage drop value Dyz, the node fault frequency Gzpl and the load balance coefficient Fh respectively, and V represents the first correction constant.

[0051] Preferably, the feedback management module includes an evaluation unit and a strategy unit;

[0052] The evaluation unit is used to preset the safety threshold R, and compare and analyze the partition power performance risk coefficient Lpjx generated by the load analysis module with the preset safety threshold R to determine whether there is any abnormality in the operation of the equipment in each topological partition in the current low-voltage distribution station area, including:

[0053] When the partition power performance risk coefficient Lpjx>safety threshold R, it indicates that the current partition load is abnormal and the node equipment is operating abnormally, and a risk warning signal is issued and displayed in real time on the user monitoring interface;

[0054] When the partition power performance risk coefficient Lpjx ≤ the safety threshold R, it means that the current partition load is normal, the node equipment operates normally, and continuous monitoring is performed.

[0055] Preferably, the strategy unit is used to receive risk warning signals, mark the abnormal risk of the current partition load node equipment, start the fault isolation program, identify and isolate the abnormal equipment or lines, perform load transfer operations, transfer the load to the normal partition, and perform emergency repairs on the power equipment of the abnormal node.

[0056] The low voltage distribution substation area topology identification method based on cluster analysis and graph theory includes the following steps:

[0057] Step 1: Use an infrared imaging device installed on a drone to inspect the distribution line, collect and identify the line length, number of branches and branch equipment, and collect the electrical parameters of each node to establish a node database;

[0058] Step 2: extracting the geographical location and electrical parameters of electrical equipment in the node database to cluster nodes, and establishing a cluster analysis model, and inputting the node database into the cluster analysis model, and dividing it into several initial partitions through the cluster analysis model;

[0059] Step 3: Establish a partition node network diagram based on the node database, and construct a topological characteristic coefficient Tp for each initial partition; and further mark the topological structure category of the initial partition according to the topological characteristic coefficient Tp;

[0060] Step 4: extract the connection node current phase angle difference Xwjc, node harmonic content Xbhl, node instantaneous current Ssdl, node switching voltage drop value Dyz, node fault frequency Gzpl and load balance coefficient Fh from the electrical parameters of each initial partition, predict the load growth trend through model training, and calculate and generate the partition power performance risk coefficient Lpjx within the partition;

[0061] Step 5. Compare and analyze the partition power performance risk coefficient Lpjx with the pre-set safety threshold R to determine whether there is any abnormality in the operation of equipment in each topological partition within the current low-voltage distribution station area. When the partition power performance risk coefficient Lpjx exceeds the safety threshold R, a risk warning signal is issued and displayed in real time on the user monitoring interface. At the same time, fault isolation and load transfer emergency processing plans are executed through remote monitoring means.

[0062] The present invention provides a method and system for identifying low-voltage distribution station area topology based on cluster analysis and graph theory. It has the following beneficial effects:

[0063] (1) Traditional topology identification relies on manual inspections and drawing analysis, which is inefficient and prone to errors. Through the data acquisition module in the present invention, the drone is equipped with infrared imaging equipment to inspect the distribution lines, combined with the automatic collection of node electrical parameters and database management, which greatly improves the efficiency of topology structure identification. Compared with the traditional manual method, the present invention can automatically complete the global identification of the distribution network in a shorter time, which is particularly suitable for complex low-voltage substations, avoiding delays and errors caused by manual inspections. Through the combination of cluster analysis models and graph theory algorithms, the present invention can accurately identify the topological structure of each partition (such as ring network, tree, chain, radial, etc.), and automatically calculate the topological characteristic coefficient Tp, so that the complex topological structure of the distribution substation can be more carefully and comprehensively classified and analyzed. Traditional methods are often difficult to accurately classify when faced with complex branch structures. The topology identification module of the present invention ensures the accuracy of the classification results through graph theory calculation methods.

[0064] (2) Traditional load assessment usually relies on the average value within a fixed time period, which makes it difficult to capture the load imbalance between nodes and the instantaneous changes in electrical parameters. The load analysis module of the present invention can analyze the load status in the distribution station area in real time by extracting the connection node current phase angle difference Xwjc, node harmonic content Xbhl, node instantaneous current Ssdl, node switching voltage drop value Dyz, node fault frequency Gzpl and load balance coefficient Fh dynamic electrical parameters. Especially in the context of the increase in distributed energy access and emerging power-consuming equipment such as electric vehicles, this real-time load analysis method is more suitable for the management needs of modern distribution networks.

[0065] (3) Through model training, the load growth trend is predicted, and possible future load overload situations are identified in advance to avoid equipment failure and power supply instability caused by sudden load increases. Compared with traditional static assessments, the present invention can dynamically monitor and predict possible future load changes, thus improving the foresight of load management.

[0066] (4) The present invention uses a feedback management module to compare and analyze the partition power performance risk coefficient Lpjx and the safety threshold R. When the partition power performance risk coefficient Lpjx exceeds the safety threshold R, a fault warning signal can be issued in time to avoid further deterioration of the equipment abnormality. This real-time monitoring and warning function enables the operating status of the distribution station area to be continuously tracked. Once an abnormal situation occurs, the system can respond immediately, notify the technical staff, and perform fault isolation and load transfer through remote means. Fault handling in traditional technologies often relies on manual judgment and operation, and processing delays may lead to larger-scale power outages. On the contrary, after detecting an abnormality, the present invention can automatically identify abnormal nodes and isolate faults, and at the same time perform load transfer, quickly transferring power demand to normally operating partitions, reducing the scope and time of power outages. This automated emergency handling greatly improves the power supply reliability of the distribution network and ensures the stability of power consumption for users. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 It is a schematic diagram of a block diagram of a low-voltage distribution station area topology identification system based on cluster analysis and graph theory of the present invention;

[0068] Figure 2 The figure is a schematic diagram of the steps of the low-voltage distribution station area topology identification method based on cluster analysis and graph theory of the present invention. DETAILED DESCRIPTION

[0069] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0070] Example 1

[0071] See also Figure 1 , a low voltage distribution substation area topology identification system based on cluster analysis and graph theory, including:

[0072] The data acquisition module is used to inspect the distribution lines by installing infrared imaging equipment using drones, collect and identify the line length, number of branches and branch equipment, and collect the electrical parameters of each node to establish a node database;

[0073] The initial partitioning module is used to extract the geographical location and electrical parameters of the electrical equipment in the node database to perform node clustering, establish a clustering analysis model, and input the node database into the clustering analysis model to divide it into several initial partitions through the clustering analysis model;

[0074] A topology recognition module is used to establish a partition node network diagram based on a node database, and construct a topological characteristic coefficient Tp for each initial partition; and further mark the topological structure category of the initial partition according to the topological characteristic coefficient Tp;

[0075] The load analysis module extracts the connection node current phase angle difference Xwjc, node harmonic content Xbhl, node instantaneous current Ssdl, node switching voltage drop value Dyz, node fault frequency Gzpl and load balance coefficient Fh from the electrical parameters of each initial partition, predicts the load growth trend through model training, and calculates and generates the partition power performance risk coefficient Lpjx within the partition;

[0076] The specific steps for predicting load growth trends through model training are:

[0077] The electrical parameter data of each node in the distribution area is collected in real time through the second acquisition unit (sensors installed at each node). These parameters include but are not limited to: connection node current phase angle difference Xwjc, node harmonic content Xbhl, node instantaneous current Ssdl, node switching voltage drop value Dyz, node fault frequency Gzpl, actual load value P of the initial partition actual and rated load value P rated .

[0078] Collect historical load data in the distribution substation area, including long-term load records, and conduct detailed time series analysis based on data at different time scales of minutes, hours or days. Clean the collected electrical parameter data and deal with noise, outliers, and missing data in the collection. Abnormal data points can be corrected through statistical analysis or data interpolation technology to ensure the accuracy and integrity of the data. According to the actual needs of the distribution substation area, extract key features from the original data, including the time series characteristics of the load (daily / hourly load changes), the change trend of node electrical parameters, and load fluctuation characteristics. For example, calculate the daily, weekly, and monthly load mean, variance and other statistical values ​​to extract the periodicity and volatility characteristics of the load. Normalize the electrical parameters and load data to make the dimensions of each data consistent, so as to facilitate model processing and avoid a feature that is too large or too small to affect the model training effect. According to the load prediction needs and data characteristics of the distribution substation area, select ARIMA, SARIMA model or deep learning model to predict the load growth trend. Divide the historical data into training set, validation set and test set. Generally, 70% of the data is used for training, 15% for verification, and the remaining 15% for testing the prediction effect of the model. By continuously inputting the latest node electrical parameters, the model can make real-time rolling predictions on the load change trend for a period of time in the future.

[0079] The feedback management module is used to compare and analyze the partition power performance risk coefficient Lpjx with the pre-set safety threshold R to determine whether there is any abnormality in the operation of equipment in each topological partition within the current low-voltage distribution station area. When the partition power performance risk coefficient Lpjx exceeds the safety threshold R, a risk warning signal is issued and displayed in real time on the user monitoring interface. At the same time, fault isolation and load transfer emergency processing plans are executed through remote monitoring means.

[0080] In this embodiment, traditional topology identification relies on manual inspection and drawing analysis, which is inefficient and prone to errors. Through the data acquisition module in the present invention, the drone is equipped with infrared imaging equipment to inspect the distribution lines, combined with the automatic collection of node electrical parameters and database management, which greatly improves the efficiency of topology structure identification. Compared with the traditional manual method, the present invention can automatically complete the global identification of the distribution network in a shorter time, which is particularly suitable for complex low-voltage areas, avoiding delays and errors caused by manual inspections. Through the combination of cluster analysis models and graph theory algorithms, the present invention can accurately identify the topological structure of each partition (such as ring network, tree, chain, radial, etc.), and automatically calculate the topological characteristic coefficient Tp, so that the complex topological structure of the distribution area can be more carefully and comprehensively classified and analyzed. Traditional methods are often difficult to accurately classify when facing complex branch structures. The topology identification module of the present invention ensures the accuracy of the classification results through the calculation method of graph theory.

[0081] Traditional load assessment usually relies on the average value within a fixed time period, which makes it difficult to capture the load imbalance between nodes and the instantaneous changes in electrical parameters. The load analysis module of the present invention can analyze the load status in the distribution station area in real time by extracting dynamic electrical parameters such as the current phase angle difference, harmonic content, instantaneous current, and switching voltage drop of the connected nodes. Especially in the context of the increase in distributed energy access and emerging power-consuming devices such as electric vehicles, this real-time load analysis method is more suitable for the management needs of modern distribution networks.

[0082] Through model training, the load growth trend is predicted, and possible future load overload situations are identified in advance to avoid equipment failure and power supply instability caused by sudden load increases. Compared with traditional static assessments, this invention can dynamically monitor and predict possible future load changes, improving the foresight of load management.

[0083] The present invention compares and analyzes the partition power performance risk coefficient Lpjx and the safety threshold R through the feedback management module. When the partition power performance risk coefficient Lpjx exceeds the safety threshold R, a fault warning signal can be issued in time to avoid further deterioration of the abnormal situation of the equipment. This real-time monitoring and early warning function enables the operating status of the distribution station area to be continuously tracked. Once an abnormal situation occurs, the system can respond immediately, notify the technical staff, and perform fault isolation and load transfer by remote means. Fault handling in traditional technologies often relies on manual judgment and operation, and processing delays may lead to power outages on a larger scale. On the contrary, after detecting an abnormality, the present invention can automatically identify abnormal nodes and isolate faults, and at the same time perform load transfer, quickly transfer power demand to the normal operating partition, and reduce the scope and time of power outages. This automated emergency handling greatly improves the power supply reliability of the distribution network and ensures the stability of power consumption for users.

[0084] The structure of modern power distribution network is becoming increasingly complex, especially in the context of distributed energy access (such as photovoltaic power generation, wind power generation, etc.) and electric vehicle charging station construction, the traditional static topology identification and management can no longer meet the needs. The present invention can adapt to complex power demand through a dynamic monitoring system, update the topology structure and load status in time, and ensure the stable operation of the distribution area under multiple power sources.

[0085] Example 2

[0086] See also Figure 1 , the data acquisition module includes a first acquisition unit;

[0087] The first acquisition unit is used to install infrared imaging equipment through drones to inspect distribution lines, collect and identify line length, number of branches and branch equipment, obtain inspection data, and identify that the inspection data is obtained with each mortgaged distribution transformer as the center, and identify and obtain central nodes, branch nodes, intermediate nodes and terminal nodes; these devices use infrared technology to image temperature differences, detect potential hot spots in the lines, and help identify temperature anomalies of electrical equipment, especially overheating under high load.

[0088] Through image processing technology and infrared imaging data analysis, four key nodes in the distribution line are automatically identified:

[0089] The central node includes the location of the transformer or main switchgear, which serves as the power supply center for the entire substation area;

[0090] Branch nodes include branch switches or branch transformers to form new branches, which are important nodes in the distribution network;

[0091] Central nodes include connection points or support points distributed in the line. Connection points or support points include poles or brackets, which are used to connect different line sections and are key structural nodes of the line.

[0092] The end node includes the user's electricity meter and is the terminal node of the distribution line.

[0093] The data acquisition module also includes a second acquisition unit, which is used to collect electrical parameters of each node;

[0094] The electrical parameters include: connection node current phase angle difference Xwjc, node harmonic content Xbhl, node instantaneous current Ssdl, node switching voltage drop value Dyz and node fault frequency Gzpl.

[0095] The electrical parameters are collected by the following sensor network deployed at each node of the initial partition:

[0096] Current phase angle sensor: used to measure the current phase angle difference (Xwjc) between connected nodes to help analyze load imbalance.

[0097] Harmonic sensors: used to detect the harmonic content (Xbhl) of a node and identify power quality issues in the grid, especially when the load changes dynamically.

[0098] Instantaneous current sensor: used to measure the instantaneous current (Ssdl) of the node, which helps to capture the transient phenomenon of current change and analyze the instantaneous fluctuation of node load.

[0099] Voltage drop detector: monitors the voltage changes at the node, especially the voltage drop value (Dyz) during switching, which is used to evaluate the stability of power transmission during the switching process.

[0100] Fault frequency sensor: used to count the fault frequency (Gzpl) of nodes, which can help identify the device health status of the node and predict the possibility of fault occurrence.

[0101] Example 3

[0102] See also Figure 1 ,The initial partitioning module includes a data preprocessing unit and a cluster partitioning unit;

[0103] The data preprocessing unit is used to preprocess the node database and standardize the processed related data information according to the dimensionless processing technology to unify the units;

[0104] The cluster partition unit is used to identify sub-areas within the station area through the initial partition tree and divide the monitoring area according to the preset hierarchical level. The specific steps are as follows:

[0105] S1. Determine whether to merge or split nodes by using Euclidean distance. After selecting the distance measurement method, calculate the distance between each node and other nodes to form a distance matrix D. The specific definition formula is:

[0106] D=[d ij ];

[0107]

[0108] Where D is an n×n distance matrix, which represents the distance matrix between nodes, and d ij represents the distance between node i and node j; x ik and x jk They represent the values ​​of node i and node j in the kth dimension, including voltage, current, longitude and latitude; p represents the number of dimensions of the feature;

[0109] S2, based on several distance matrices, each node is treated as a separate cluster, based on d ij Represents the distance between node i and node j, which is merged using the following distance measurement method:

[0110] S21, minimum distance method: select the distance between the nearest nodes of the clusters:

[0111]

[0112] S22, Maximum distance method: Select the distance of the farthest node between clusters:

[0113]

[0114] S23, average distance method: select the average distance between all nodes in the cluster:

[0115]

[0116] S3. Find the two clusters C with the smallest distance A and C B , merge them into a new cluster Repeat steps S2 and S3 to gradually merge the closest clusters until the preset number of clusters or distance threshold is reached, and finally several initial partitions are formed, each of which represents a relatively independent distribution station area topology substructure.

[0117] Preferably, the topology identification module includes a network graph theory node unit and a calculation unit;

[0118] The network graph theory node unit is used to establish a partition node network graph based on the node database set, and construct the topological characteristic coefficient Tp of each initial partition through the calculation unit. The topological characteristic coefficient Tp is calculated and obtained by the following formula:

[0119]

[0120]

[0121] In the formula, N is the total number of nodes in the network, M is the total number of edges in the network, F represents the clustering coefficient. The larger the F value, the closer the neighbors of the nodes in the network are connected to each other, and it has a strong local clustering characteristic. The clustering coefficient can be used to analyze the connectivity between the internal devices of the substation area and judge the local clustering characteristics of the devices, E i represents the actual number of connected edges between the other nodes of the i-th node, K i represents the number of neighbor nodes connected to the i-th node, B represents the node betweenness, σ st represents the total number of shortest paths between node s and node t, σ st (i) represents the number of shortest paths from node s to node t passing through node i; the node betweenness represents how many paths pass through node i among the shortest paths between any pair of nodes s and t in the network. The higher the node betweenness, the more crucial the node is in the network and the more important it is for the overall connectivity;

[0122] L represents the average path length, d ij represents the distance between node i and node j; the average path length is used to evaluate the transmission efficiency of the power supply network; a smaller L indicates a high transmission efficiency between the nodes in the network and a shorter system response time; a larger L indicates a longer connection between the nodes and a more complex and inefficient transmission process, H i represents the number of edges of the i-th node, ZK J represents the line impedance value of the J-th edge, kdz J represents the line length of the J-th edge, P J represents the line load of the J-th edge, Y represents the partition redundancy distance, sjlj represents the actual path length of the partition, zdlj represents the shortest path length of the partition; the partition redundancy distance provides a comparison between the actual path length and the shortest path length of the connection between the nodes in the network, and provides a basis for judging the topological characteristics (such as loop network, tree-like, etc.) of the network. By analyzing the redundancy distance, different types of topological structures can be effectively distinguished. a1, a2, a3, a4, a5, and a6 are weight coefficients, which are adjusted and set by the user, 0 < a1 < 1, 0 < a2 < 1, 0 < a3 < 1, 0 < a4 < 1, 0 < a5 < 1, 0 < a6 < 1, and a1 + a2 + a3 + a4 + a5 + a6 = 1.

[0123] In this embodiment, the cluster partitioning unit provides a variety of distance calculation methods such as the minimum distance method, the maximum distance method and the average distance method, so that when dealing with different types of distribution networks, the appropriate clustering strategy can be flexibly selected to ensure the accuracy and effectiveness of clustering. By continuously merging clusters with the smallest distance, multiple initial partitions are formed. This mechanism can adapt to the complexity of the distribution network and ultimately generate relatively independent topological substructures, which is convenient for subsequent management and analysis. By establishing a partition node network diagram and using the topological characteristic coefficient Tp for analysis, the connectivity and transmission efficiency of the network can be quantified. A high clustering coefficient indicates a close connection between devices, which helps to discover potential fault points and their impact range. This embodiment not only improves the management efficiency of the distribution network by integrating data preprocessing, cluster analysis and topology recognition, but also enhances the system's adaptability to complex situations, providing important support for the construction of intelligent power systems.

[0124] Example 4

[0125] Please refer to Figure 1 The topology recognition module also includes a classification unit, which is used to preset a first threshold value Q1, a second threshold value Q2 and a third threshold value Q3, and compare the topological characteristic coefficient Tp of each initial partition with the first threshold value Q1, the second threshold value Q2 and the third threshold value Q3 respectively to obtain corresponding classification results, including:

[0126] When the topological characteristic coefficient Tp> the first threshold Q1, the initial partition is divided into a ring network structure category;

[0127] When the second threshold Q2 < the topological characteristic coefficient Tp ≤ the first threshold Q1, the initial partition is divided into a tree structure category;

[0128] When the third threshold value Q3 < the topological characteristic coefficient Tp ≤ the second threshold value Q2, the initial partition is divided into a chain structure category;

[0129] When the topological characteristic coefficient Tp≤the third threshold Q3, the initial partition is divided into a radial structure category.

[0130] In this embodiment, by setting different thresholds, the distribution power grid area can be more finely divided into ring network, tree-shaped, chain-shaped, and radial structures, so as to provide more detailed network topology information, which helps to understand and analyze the layout and characteristics of the network. Different types of topological structures have different requirements in management and maintenance. The classification results can help operators formulate targeted maintenance strategies and improve management efficiency. For example, the ring network structure may require more redundant power supply strategies, while the tree-shaped structure may focus more on the reliability of nodes. By clarifying the topological categories of each initial partition, the speed of fault diagnosis and handling can be accelerated. For example, faults in the ring network structure can be quickly bypassed through other paths, while faults in the chain-shaped structure may affect the entire chain, and the classification results help to take appropriate emergency response measures. In the face of emerging electrical equipment such as distributed energy and electric vehicles, clear topological structure classification helps to understand the impact of these devices on the distribution network and timely adjust the network layout to adapt to the changing load demand.

[0131] Embodiment 5

[0132] Please refer to Figure 1 , the load analysis module includes a load acquisition unit and a prediction unit;

[0133] The load acquisition unit is used to collect the actual load value P actual and the rated load value P rated of each initial partition, and calculate and obtain the load balance coefficient Fh through the following formula:

[0134]

[0135] The prediction unit is used to extract the connection node current phase angle difference Xwjc, node harmonic content Xbhl, node instantaneous current Ssdl, node switching voltage drop value Dyz, and node fault frequency Gzpl in the electrical parameters, and associate the load balance coefficient Fh. After dimensionless processing, calculate and obtain the partition power performance risk coefficient Lpjx through the following formula:

[0136]

[0137] In the formula, a1, a2, a3, a4, a5, and a6 respectively represent the weight coefficients of the connection node current phase angle difference Xwjc, node harmonic content Xbhl, node instantaneous current Ssdl, node switching voltage drop value Dyz, node fault frequency Gzpl, and load balance coefficient Fh, which are adjusted and set by the user, 0 < a1 < 1, 0 < a2 < 1, 0 < a3 < 1, 0 < a4 < 1, 0 < a5 < 1, 0 < a6 < 1, and a1 + a2 + a3 + a4 + a5 + a6 = 1. V represents the first correction constant.

[0138] The phase angle difference of the current at the connected nodes can reveal the current phase difference between the nodes and help identify load imbalance problems. This feature has an important improvement effect on the load imbalance problem that is difficult to detect in time due to manual inspection in traditional methods; harmonic content is an important indicator for evaluating power quality. Excessive harmonic content may cause equipment damage and reduce system efficiency. Through automatic monitoring, it is possible to timely discover and take measures to improve power quality, replacing the time-consuming manual inspection in traditional methods. Through real-time monitoring of instantaneous current, the power dispatching plan can be optimized to ensure the continuity and reliability of power supply. The node switching voltage drop value can reflect the responsiveness of the equipment when the load changes, helping to evaluate the stability of the power grid. In traditional methods, there is often a lack of systematic monitoring of this parameter. Timely monitoring of voltage drop can quickly take measures when the equipment switches, reduce power outage time, and improve the recovery capability of the system. Through real-time monitoring of fault frequency, it is possible to respond to fault risks in a timely manner and ensure the safe operation of the system.

[0139] The feedback management module includes an evaluation unit and a strategy unit;

[0140] The evaluation unit is used to preset the safety threshold R, and compare and analyze the partition power performance risk coefficient Lpjx generated by the load analysis module with the preset safety threshold R to determine whether there is any abnormality in the operation of the equipment in each topological partition in the current low-voltage distribution station area, including:

[0141] When the partition power performance risk coefficient Lpjx>safety threshold R, it indicates that the current partition load is abnormal and the node equipment is operating abnormally, and a risk warning signal is issued and displayed in real time on the user monitoring interface;

[0142] When the partition power performance risk coefficient Lpjx ≤ the safety threshold R, it means that the current partition load is normal, the node equipment operates normally, and continuous monitoring is performed.

[0143] The strategy unit is used to receive risk warning signals, mark the abnormal risk of the current partition load node equipment, start the fault isolation program, identify and isolate the abnormal equipment or lines, perform load transfer operations, transfer the load to the normal partition, and perform emergency repairs on the power equipment at the abnormal node.

[0144] In this embodiment, the load collection unit obtains the actual load value and rated load value of each initial partition in real time, so that the distribution network can dynamically monitor the load status, thereby promptly discovering potential load imbalance problems. By calculating the load balance coefficient Fh, managers can effectively evaluate the load distribution of each partition, ensure the stability of power supply, and avoid power equipment failures caused by uneven load. The prediction unit integrates multiple electrical parameters and calculates the partition power performance risk coefficient Lpjx. Through multi-dimensional data analysis, it helps to comprehensively evaluate the power performance of the partition, thereby providing a more scientific decision-making basis. The evaluation unit compares the risk coefficient with the preset safety threshold RRR. When a load abnormality occurs, the system will automatically issue a warning signal to notify managers in a timely manner, respond quickly to potential problems, and avoid larger-scale failures.

[0145] Example 6

[0146] Please refer to Figure 2 ,The low voltage distribution station area topology identification method based on cluster analysis and graph theory includes the following steps:

[0147] Step 1: Use an infrared imaging device installed on a drone to inspect the distribution line, collect and identify the line length, number of branches and branch equipment, and collect the electrical parameters of each node to establish a node database;

[0148] Step 2: extracting the geographical location and electrical parameters of electrical equipment in the node database to cluster nodes, and establishing a cluster analysis model, and inputting the node database into the cluster analysis model, and dividing it into several initial partitions through the cluster analysis model;

[0149] Step 3: Establish a partition node network diagram based on the node database, and construct a topological characteristic coefficient Tp for each initial partition; and further mark the topological structure category of the initial partition according to the topological characteristic coefficient Tp;

[0150] Step 4: extract the connection node current phase angle difference Xwjc, node harmonic content Xbhl, node instantaneous current Ssdl, node switching voltage drop value Dyz, node fault frequency Gzpl and load balance coefficient Fh from the electrical parameters of each initial partition, predict the load growth trend through model training, and calculate and generate the partition power performance risk coefficient Lpjx within the partition;

[0151] Step 5. Compare and analyze the partition power performance risk coefficient Lpjx with the pre-set safety threshold R to determine whether there is any abnormality in the operation of equipment in each topological partition within the current low-voltage distribution station area. When the partition power performance risk coefficient Lpjx exceeds the safety threshold R, a risk warning signal is issued and displayed in real time on the user monitoring interface. At the same time, fault isolation and load transfer emergency processing plans are executed through remote monitoring means.

[0152] By using drones equipped with infrared imaging equipment for inspection, the system can quickly collect line length, number of branches and electrical parameters to establish a comprehensive node database. This process reduces the time and cost of manual inspections and improves the accuracy of data collection.

[0153] Next, through node clustering analysis, the system divides the nodes into several initial partitions based on geographical location and electrical characteristics to form a clear network structure. The topological characteristic coefficient Tp of each initial partition is calculated and marked as different topological structure types (such as ring network, tree, etc.), which provides strong support for subsequent load analysis and fault diagnosis. By extracting the electrical parameters of each partition and training the model, the system can effectively predict the load growth trend and calculate the risk coefficient of the partition power performance. This real-time monitoring capability enables operation managers to identify potential risks in a timely manner and take corresponding measures to ensure power supply safety.

[0154] Finally, when the partition power performance risk factor Lpjx exceeds the preset safety threshold R, the system will automatically issue an early warning signal and quickly perform fault isolation and load transfer through remote monitoring. This series of measures not only improves the response speed and safety of the distribution network, but also enhances the ability of intelligent management, providing users with reliable power supply guarantees.

[0155] The threshold is set to facilitate comparison. The size of the threshold depends on the amount of sample data and the number of bases set by technicians in this field for each group of sample data; as long as it does not affect the proportional relationship between the parameter and the quantized value.

[0156] The above formulas are obtained by collecting a large amount of data for software simulation and selecting a formula that is close to the actual value. The coefficients in the formula are set by technical personnel in this field according to actual conditions. The above is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited to this. Any technical personnel familiar with the technical field within the technical scope disclosed by the present invention, according to the technical solution and the inventive concept of the present invention, make equivalent replacement or change, which should be covered within the protection scope of the present invention.

Claims

1. A low voltage distribution area topology identification system based on cluster analysis and graph theory, characterized by: include: The data acquisition module is used to inspect the distribution lines by installing infrared imaging equipment using drones, collect and identify the line length, number of branches and branch equipment, and collect the electrical parameters of each node to establish a node database; The initial partitioning module is used to extract the geographical location and electrical parameters of the electrical equipment in the node database to perform node clustering, establish a clustering analysis model, and input the node database into the clustering analysis model to divide it into several initial partitions through the clustering analysis model; A topology recognition module is used to establish a partition node network diagram based on the node database and construct a topological characteristic coefficient Tp of each initial partition; And further mark the topological structure category of the initial partition according to the topological characteristic coefficient Tp; The load analysis module extracts the connection node current phase angle difference Xwjc, node harmonic content Xbhl, node instantaneous current Ssdl, node switching voltage drop value Dyz, node fault frequency Gzpl and load balance coefficient Fh from the electrical parameters of each initial partition, predicts the load growth trend through model training, and calculates and generates the partition power performance risk coefficient Lpjx within the partition; The feedback management module is used to compare and analyze the partition power performance risk coefficient Lpjx with the pre-set safety threshold R to determine whether there is any abnormality in the operation of equipment in each topological partition within the current low-voltage distribution station area. When the partition power performance risk coefficient Lpjx exceeds the safety threshold R, a risk warning signal is issued and displayed in real time on the user monitoring interface. At the same time, fault isolation and load transfer emergency processing plans are executed through remote monitoring means.

2. The low voltage distribution station area topology identification system based on cluster analysis and graph theory according to claim 1 is characterized in that: The data acquisition module includes a first acquisition unit; The first acquisition unit is used to inspect the distribution line by installing infrared imaging equipment through a drone, collect and identify the line length, the number of branches and branch equipment, obtain inspection data, and identify the inspection data acquisition with each mortgaged distribution transformer as the center, identify and obtain the central node, branch node, intermediate node and terminal node; The central node includes the location of the transformer or main switchgear, which serves as the power supply center for the entire substation area; Branch nodes include branch switches or branch transformers to form new branches, which are important nodes in the distribution network; Central nodes include connection points or support points distributed in the line. Connection points or support points include poles or brackets, which are used to connect different line sections and are key structural nodes of the line. The end node includes the user's electricity meter and is the terminal node of the distribution line.

3. The low voltage distribution station area topology identification system based on cluster analysis and graph theory according to claim 2 is characterized in that: The data acquisition module further includes a second acquisition unit, and the second acquisition unit is used to acquire electrical parameters of each node; The electrical parameters include: connection node current phase angle difference Xwjc, node harmonic content Xbhl, node instantaneous current Ssdl, node switching voltage drop value Dyz and node fault frequency Gzpl.

4. The low voltage distribution station area topology identification system based on cluster analysis and graph theory according to claim 1 is characterized in that: The initial partitioning module includes a data preprocessing unit and a cluster partitioning unit; The data preprocessing unit is used to preprocess the node database, and standardize the processed related data information according to the dimensionless processing technology to unify the units; The cluster partition unit is used to identify sub-areas within the station area through an initial partition tree, and divide the monitoring area according to a preset hierarchical level. The specific steps are: S1. Determine whether to merge or split nodes by using Euclidean distance. After selecting the distance measurement method, calculate the distance between each node and other nodes to form a distance matrix D. The specific definition formula is: Where D is an n×n distance matrix, which represents the distance matrix between nodes, and d ij represents the distance between node i and node j; x ik and x jk They represent the values ​​of node i and node j in the kth dimension, including voltage, current, longitude and latitude; p represents the number of dimensions of the feature; S2, based on several distance matrices, each node is treated as a separate cluster, based on d ij Represents the distance between node i and node j, which is merged using the following distance measurement method: S21, minimum distance method: select the distance between the nearest nodes of the clusters: S22, Maximum distance method: Select the distance of the farthest node between clusters: S23, average distance method: select the average distance between all nodes in the cluster: S3. Find the two clusters C with the smallest distance A and C B , merge them into a new cluster Repeat steps S2 and S3 to gradually merge the closest clusters until the preset number of clusters or distance threshold is reached, and finally several initial partitions are formed, each of which represents a relatively independent distribution station area topology substructure.

5. The low voltage distribution station area topology identification system based on cluster analysis and graph theory according to claim 1 is characterized in that: The topology identification module includes a network graph theory node unit and a calculation unit; The network graph theory node unit is used to establish a partition node network graph based on a node database set, and construct a topological characteristic coefficient Tp of each initial partition through a calculation unit. The topological characteristic coefficient Tp is calculated and obtained by the following formula: Where N is the total number of nodes in the network, M is the total number of edges in the network, F is the clustering coefficient, and E is i Indicates the actual number of edges between the i-th node and the nodes, K i represents the number of neighboring nodes connected to the i-th node, B represents the node betweenness, σ st represents the total number of shortest paths from node s to node t, σ st (i) represents the number of shortest paths from node s to node t through node i; L represents the average path length, d ij represents the distance between node i and node j; H i represents the number of edges at the i-th node, ZK J represents the line impedance value of the Jth edge, kdz J represents the line length of the Jth edge, P J represents the line load of the Jth edge, Y represents the redundant distance of the partition, sjlj represents the actual path length of the partition, zdlj represents the shortest path length of the partition; a1, a2, a3, a4, a5 and a6 are weight coefficients.

6. The low voltage distribution station area topology identification system based on cluster analysis and graph theory according to claim 5 is characterized in that: The topology recognition module also includes a classification unit, which is used to preset a first threshold value Q1, a second threshold value Q2 and a third threshold value Q3, and compare the topological characteristic coefficient Tp of each initial partition with the first threshold value Q1, the second threshold value Q2 and the third threshold value Q3 respectively to obtain corresponding classification results, including: When the topological characteristic coefficient Tp> the first threshold Q1, the initial partition is divided into a ring network structure category; When the second threshold Q2 < the topological characteristic coefficient Tp ≤ the first threshold Q1, the initial partition is divided into a tree structure category; When the third threshold value Q3 < the topological characteristic coefficient Tp ≤ the second threshold value Q2, the initial partition is divided into a chain structure category; When the topological characteristic coefficient Tp≤the third threshold Q3, the initial partition is divided into a radial structure category.

7. The low voltage distribution substation area topology identification system based on cluster analysis and graph theory according to claim 1 is characterized in that: The load analysis module includes a load collection unit and a prediction unit; The load collection unit is used to collect the actual load value P of each initial partition. actual and rated load value P rated , the load balance coefficient Fh is calculated by the following formula: The prediction unit is used to extract the connection node current phase angle difference Xwjc, node harmonic content Xbhl, node instantaneous current Ssdl, node switching voltage drop value Dyz and node fault frequency Gzpl in the electrical parameters, and associate the load balance coefficient Fh. After dimensionless processing, the partition power performance risk coefficient Lpjx is calculated by the following formula: Wherein, a1, a2, a3, a4, a5 and a6 represent the weight coefficients of the connection node current phase angle difference Xwjc, the node harmonic content Xbhl, the node instantaneous current Ssdl, the node switching voltage drop value Dyz, the node fault frequency Gzpl and the load balance coefficient Fh respectively, and V represents the first correction constant.

8. The low voltage distribution station area topology identification system based on cluster analysis and graph theory according to claim 7 is characterized in that: The feedback management module includes an evaluation unit and a strategy unit; The evaluation unit is used to preset a safety threshold R, and compare and analyze the partition power performance risk coefficient Lpjx generated by the load analysis module with the preset safety threshold R, to determine whether there is any abnormality in the operation of the equipment in each topological partition in the current low-voltage distribution station area, including: When the partition power performance risk coefficient Lpjx>safety threshold R, it indicates that the current partition load is abnormal and the node equipment is operating abnormally, and a risk warning signal is issued and displayed in real time on the user monitoring interface; When the partition power performance risk coefficient Lpjx ≤ the safety threshold R, it means that the current partition load is normal, the node equipment operates normally, and continuous monitoring is performed.

9. The low voltage distribution station area topology identification system based on cluster analysis and graph theory according to claim 8 is characterized in that: The strategy unit is used to receive risk warning signals, mark the abnormal risk of the current partition load node equipment, start the fault isolation program, identify and isolate abnormal equipment or lines, perform load transfer operations, transfer the load to the normal partition, and perform emergency repairs on the power equipment at the abnormal node.

10. A low-voltage distribution substation area topology identification method based on cluster analysis and graph theory, applied to a low-voltage distribution substation area topology identification system based on cluster analysis and graph theory as claimed in any one of claims 1 to 9, characterized in that: The following steps are involved: Step 1: Use an infrared imaging device installed on a drone to inspect the distribution line, collect and identify the line length, number of branches and branch equipment, and collect the electrical parameters of each node to establish a node database; Step 2: extracting the geographical location and electrical parameters of electrical equipment in the node database to cluster nodes, and establishing a cluster analysis model, and inputting the node database into the cluster analysis model, and dividing it into several initial partitions through the cluster analysis model; Step 3: Establish a partition node network diagram based on the node database, and construct a topological characteristic coefficient Tp for each initial partition; And further mark the topological structure category of the initial partition according to the topological characteristic coefficient Tp; Step 4: extract the connection node current phase angle difference Xwjc, node harmonic content Xbhl, node instantaneous current Ssdl, node switching voltage drop value Dyz, node fault frequency Gzpl and load balance coefficient Fh from the electrical parameters of each initial partition, predict the load growth trend through model training, and calculate and generate the partition power performance risk coefficient Lpjx within the partition; Step 5. Compare and analyze the partition power performance risk coefficient Lpjx with the pre-set safety threshold R to determine whether there is any abnormality in the operation of equipment in each topological partition within the current low-voltage distribution station area. When the partition power performance risk coefficient Lpjx exceeds the safety threshold R, a risk warning signal is issued and displayed in real time on the user monitoring interface. At the same time, fault isolation and load transfer emergency processing plans are executed through remote monitoring means.

Images