Novel power system load modeling method based on graph calculation

Abstract

The invention discloses a novel power system load modeling method based on graph calculation, and belongs to the field of power systems. The method comprises the steps that a graph model of a power system of a researched area is built, a directed weighted graph is built according to a real geographical connection relation and load modeling data, nodes in the graph are main transformers or electrical loads of a transformer substation, and the direction of the edge is the power flow direction; performing parallel calculation of a data structure on the graph model, and updating an electrical quantity parameter value of the node; performing community division on the nodes based on the improved modularity; and according to the obtained community reconstruction network, carrying out statistics on various load types and composition proportions of the nodes, obtaining a comprehensive load model of each load node through a weighting method, and further obtaining a comprehensive load model of the transformer substation. According to the method, the structure of the load model can be rapidly summarized in a short time, various load model parameters can be calculated, the accuracy of the load model parameters is improved, and therefore the accuracy of a system simulation result is improved.

Description

Technical Field

[0001] The present invention belongs to the field of power systems, and in particular relates to a novel power system load modeling method based on graph calculation. Background Art

[0002] Power system simulation calculation has important guiding significance for power system dispatching, operation and planning. As one of the four major model parameters of the power system, the accuracy of power system simulation results depends largely on the accuracy of the load model.

[0003] Load modeling has always been a difficult problem that is not easy to solve in power system research. Because there are many types of power loads, the characteristics of components vary greatly, they are widely distributed geographically, and are affected by factors such as the living standards, habits, climatic conditions, and industrial and agricultural production of the local people. The dynamic changes of loads are large, and the randomness of users' switching on and off of power equipment is strong. In this field, the work of Professor He Renmu of North China Electric Power University, Professor Tang Yong of China Electric Power Research Institute, Professor Ju Ping of Hohai University, and Professor Li Xinran of Hunan University have greatly promoted the theoretical development and practical application of load modeling. Professor He Renmu proposed a time-delay dynamic system to decouple node voltage and load power, developed a set of load dynamic characteristic modeling technology, actively promoted the application of synchronized phasor measurement units (PMUs) in power systems, and incorporated measurement-based load models into time-domain simulation programs, greatly improving the accuracy of simulation results.

[0004] Reference 1 (Tang Yong, He Renmu, Ju Ping, Li Xingyuan, et al. Dynamic Characteristics and Safety Evaluation of Power Receiving System [M]. Tsinghua University Press, 2010.) discloses the definition, mechanism, characteristics of power system voltage stability, generalized load characteristics and load modeling of power receiving system, etc., which greatly promotes the theoretical development and practical application of load modeling. Reference 2 (Tang Yong, Zhao Bing, Zhang Wenchao, et al. In-depth Research and Adaptability Analysis of Comprehensive Load Model Parameters [J]. Power System Technology, 2010(2):7.) introduces the process of modeling the comprehensive load model (SLM) of the actual power grid. Based on the statistical and survey data of the actual power grid load composition, it analyzes the distribution of various loads in my country's power grid, points out the necessity of establishing SLM parameters suitable for the actual power grid, and finally analyzes the impact of SLM parameters on the stability characteristics of the power grid. Reference 3 [Wang Xiaoming, Liu Dichen, Wu Jun, et al. Generalized Load Model for Transient Stability Analysis of Power System [J]. Power System Technology, 2011, 35(3):42-48. ] The general fan and motor loads are regarded as asynchronous motors, and the model structure of the asynchronous motor parallel static load is used to model the generalized power load containing distributed power sources. Reference 4 [Qu Xing, Li Xinran, Li Peiqiang, et al. Comprehensive load modeling considering the voltage regulation of distribution network [J]. Proceedings of the CSEE, 2018, 38(6):11.] Considering the voltage transformation link of the distribution network and the influence of voltage regulation, a "Π comprehensive load model" is proposed and established from the perspective of overall identification. Reference [Ju Ping, Guo Dezheng, Cao Lu, et al. A review and prospect of comprehensive power load modeling containing active loads [J]. Journal of Hohai University: Natural Science Edition, 2020, 48(4):10.] This paper summarizes the current research progress and the current situation of power load modeling, proposes a research framework for the problem of generalized comprehensive power load modeling containing active loads, and elaborates on the research plan in detail, focusing on solving the two key problems of coordinating the accuracy and simplicity of the active load model structure and coordinating the time-varying and robustness of the active load model parameters. Reference 6 [Sun Lingling, Gao Ciwei, Tan Jian, Cui Gaoying. Load aggregation technology and its application [J]. Automation of Electric Power Systems, 2017, 41(6):9.] reviewed the research results on active load aggregation object classification and control methods, active load aggregation modeling, and active load aggregation application scenarios, and discussed the direction of load aggregation that needs further research from three perspectives: uncertainty quantification of load aggregation, load dynamic aggregation modeling, and load aggregation model control instruction decomposition. Reference 7 [Cui Yiwei. User characteristic classification and comprehensive analysis method of substation composition characteristics for load modeling [D]. Hunan University. 2021.] Research was conducted on the comprehensive statistical load modeling method and application, and a set of comprehensive load composition characteristic analysis methods based on power big data was formed, which laid a certain theoretical foundation for the precision and practicality of load modeling.

[0005] Recently, with the construction of new power systems, various generalized loads, such as distributed power sources, energy storage, electric vehicles, electrolytic aluminum, and power electronic equipment, have been integrated into the power grid in large quantities, resulting in a more complex load structure, and the volatility, controllability, and sensitivity of the load have increased accordingly. However, how to establish an accurate load model has become a major theoretical and practical problem that power workers need to face and solve urgently. Summary of the invention

[0006] In the existing load modeling and parameter identification technologies, classification and integration are the key to solving the problem of time-varying loads, and are also a problem that has always existed. In this regard, the present invention provides a new power system load modeling method based on graph computing, which can quickly summarize the load model structure and calculate various load model parameters in a short time, thereby improving the accuracy of the load model parameters and thus improving the accuracy of the system simulation results.

[0007] A novel power system load modeling method based on graph calculation of the present invention comprises the following steps:

[0008] Step 1: Establish a graphical model of the power system in the studied area. The graphical model is a directed weighted graph. The nodes in the graph are the main transformers or power loads of the substation. The edges between the nodes are established according to the transmission and distribution lines, and the direction of the edges is the direction of the power flow.

[0009] Step 2: Perform parallel calculation of the data structure of the graph model of the power system in the studied area and update the electrical quantity parameter values ​​of the nodes; the parallel calculation means updating and obtaining the electrical quantity parameters of each power load node, and then sending the electrical quantity parameters of the node through the connected edges;

[0010] Step 3: Divide the nodes into communities based on the improved modularity;

[0011] The improved modularity Calculated as: ; Intermediate parameters ;in, represents the total number of edges in the graph model, Representation Node and nodes The number of edges between Respectively represent nodes degree, node The degree, Respectively represent nodes ,,node The community where the node is located. In the same community, ,otherwise ; Through the computing node The Euclidean distance between the electrical quantity parameters is obtained;

[0012] Step 4: Based on the community reconstruction network obtained in step 3, the various load types and composition ratios of the nodes are counted, and the comprehensive load model of each load node is obtained by a weighted method, and then the comprehensive load model of the substation is obtained.

[0013] In the step 1, the attributes of the main transformer node are set to include the equipment identification, the equipment to which it belongs, the capacity and the equipment load; the attributes of the power load node are set to include the equipment identification, the equipment to which it belongs, the load transfer object, the load limit and the actual load; the attributes of the edge are set to include the edge name, state and capacity, and the state attribute of the edge takes a value of 0 or 1, 0 indicates that the line is disconnected, and 1 indicates that the line is closed.

[0014] In step 2, each main transformer node is taken as the starting point, and each power load node is visited by traversing the edge with state 1 to obtain the electrical quantity parameters of the load node; the electrical quantity parameters x include static characteristic parameters and dynamic characteristic parameters, and the static characteristic parameters include the voltage characteristic parameters of the load active power. , Frequency characteristic parameters of load active power , Voltage characteristic parameters of load reactive power Frequency characteristic parameters of reactive power of load , the dynamic characteristic parameters include stator resistance , stator reactance , Excitation Reactance , rotor resistance , rotor reactance , Mechanical torque coefficient and , inertia time constant , Load rate And the proportion of motor load power consumption to total load power consumption .

[0015] The advantages and positive effects of the present invention are as follows: (1) The method of the present invention adopts a node-centric logical framework and calculation mode. The established graph model has stronger data storage, processing and analysis capabilities, and the load composition and parameter statistical results are more accurate and reliable, which is suitable for the load model and modeling method of large-scale interconnected systems at this stage. The present invention is efficient and fast, and the results are accurate. It can support and serve the load modeling and simulation calculation of the power system. (2) The method of the present invention can quickly calculate various load model parameters in a short time, improve the accuracy of load model parameters and the accuracy of system simulation results, lay a solid foundation for the safe and stable analysis and operation control decision-making of the power system, and has strong practical application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a flow chart for implementing the novel power system load modeling method based on graph calculation of the present invention;

[0017] Figure 2 It is an example diagram of a graph model established by the present invention for a load node of a power system;

[0018] Figure 3 is a flow chart of establishing a graphical model of an electric power system according to the present invention;

[0019] Figure 4 is the electrical quantity parameter value of the load node calculated by the embodiment of the present invention;

[0020] Figure 5 This is an example diagram of performing community division on a graph model according to an embodiment of the present invention;

[0021] Figure 6 It is a flow chart of using Louvain algorithm to perform community division in an embodiment of the present invention. DETAILED DESCRIPTION

[0022] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0023] The present invention combines the statistical synthesis method with the overall discrimination method to realize a new type of power system load modeling method based on graph calculation. The power system is a natural giant graph structure. According to the real geographical connection relationship and load modeling data between the areas under the jurisdiction of the State Grid Corporation, for nodes with detailed statistical data, modeling and load model parameters can be determined according to the load type data. Load type data such as the proportion of each load type in the total active power of the substation, the proportion of each load type, and the electrical characteristics of each type of electrical equipment. For nodes that lack detailed data or are difficult to count, especially loads that vary with time periods, regions, and seasons, the load model structure and parameters can be obtained by using historical data through the method of the present invention and the overall discrimination method, and can be further verified and corrected by the fault fitting method. Graph calculation can also be used to verify the correctness of the load model and parameters of the higher voltage level station nodes obtained by the statistical synthesis method.

[0024] The novel power system load modeling method based on graph calculation of the present invention first establishes a graph model of the power system, which refers to graph data with the main power load equipment in the power system, especially the distribution network, as vertices and the connection relationship or subordination relationship between loads as edges. Then, parallel calculation of the data structure is performed, and the calculation of each node only depends on the information of its own node, adjacent nodes and edges. It is efficient and fast when processing large-scale data, and can mine potential information from massive data. According to the correlation information between nodes, community attribute division is performed, that is, the similarity of the dynamic characteristic parameters of the load nodes is classified and marked. Then, the various load types and composition ratios of the nodes are counted, and the comprehensive load model of the load node is obtained by a weighted method, so as to continuously push up and weighted iterative calculation to obtain the comprehensive load model of the substation.

[0025] like Figure 1 As shown, the novel power system load modeling method based on graph calculation of the present invention includes four steps, which are described in turn below.

[0026] Step one, establish a directed weighted graph model of the power system. The nodes in the graph model are power load equipment and substations, and the edges represent the connection relationship or subordinate relationship between the equipment. The direction of the edge is set according to the power flow. The data information of the node is directly associated with each element of the graph model. In the embodiment of the present invention, the study area is treated according to the real geographical connection relationship and load modeling data between the areas under the jurisdiction of the State Grid Corporation, that is, the collected network topology data of the power system model, the outgoing line operation data and the historical load data of the power equipment in the power supply area of ​​the power system model, abstracted and associated to each node and edge, and a directed weighted graph model of the power system is established, and various types of loads and their adjacent load nodes are associated according to a directed graph. Since the power flow of the power system can flow in both directions, in order to more clearly show the node connection relationship in the graph model in the embodiment of the present invention, the directed graph is displayed as an undirected graph. The default direction is from source to load. The graph model involved in the accompanying drawings should be understood as a simplified directed graph.

[0027] like Figure 2 As shown in the figure, nodes represent individual devices in the real world, i.e., the main transformer or load device of a substation. Each node has attributes that describe the characteristics of the node. All nodes have the same attributes, all The attribute matrix of each node is marked as A, and each row in the matrix A corresponds to a node The attribute vector of ], . The relationship between nodes is represented by edges, which can have different directions or weights to describe the nature, strength and other information of the relationship. The state attribute matrix of the edge in the graph model is represented by matrix S, and the elements in matrix S represent the state of the edge between two nodes. The weight of the edge in the graph model is represented by matrix G, and the elements in matrix G represent the weight of the edge between two nodes. The power system is originally a complex and huge network, especially with the widespread penetration of distributed energy and the access of active loads, the scale and structure of the network are becoming increasingly complex. Its network topology plus the corresponding data can constitute graph data that can reflect the operating status of the power system. Specifically, power stations, loads, etc. are nodes of the graph data, and transmission and distribution lines are edges of the graph data. Each node and branch has its own attribute characteristics.

[0028] The embodiment of the present invention uses the graph database TigerGraph as a platform to establish a power system graph model. Compared with the traditional power grid model, it not only reflects the connection relationship between physical devices, but also can directly perform graph calculations based on the power grid topology. The modeling process of the power grid graph model is as follows: Figure 3 As shown in the figure, the logical view schema of the original distribution network model and the distribution network graph model is shown. The graph model can be established according to the type and scale of the power grid. The graph can be divided into layers and blocks to correspond to the power system with multiple voltage levels. The nodes can be specific load nodes or substations with different load components. Figure 3 As shown, the node sub represents the main transformer of the substation, which is represented by a blue node, and the node Loda represents the load device node, which is represented by a brown node. In the embodiment of the present invention, the attributes of the main transformer node are set to include the device identification ID, the device Belong, the power capacity Content, the device load Load, etc., the attributes of the power load node are set to include the device identification ID, the device Belong, the load transfer object Transfer target, the load limitation Load limitation, the actual load Actual load, etc., and the attributes of the edge are set to include the edge name Name, the status Status and the capacity Content, etc. The value of the edge status attribute status is 0 or 1, 0 means the line is disconnected, and 1 means the line is closed.

[0029] Step 2: Perform parallel calculation of the data structure of the graph model of the power model of the studied area.

[0030] The most prominent feature of the present invention is the logical framework and calculation mode centered on the load node. Compared with the traditional matrix-based method, the node and edge information in the present invention is directly represented and calculated using graph data.

[0031] Figure 3The distribution network graph model in can describe the network topology and the connection relationship between each load vertex. Based on the characteristics of power system, especially the closed-loop design and open-loop operation of distribution network, each main transformer node is used as the starting point to traverse and access each load node through the edge with status 1, and the number of the load node and all electrical quantity parameters of each load node are obtained. Then the data structure is parallelly calculated, and the calculation of each node only depends on the information of its own node, adjacent nodes and edges.

[0032] At present, graph computing systems generally adopt node-centric programming models or interfaces. The method of the present invention customizes a node update function update() to describe the operations that a node needs to perform in one round of iteration. The graph computing process will go through multiple rounds of iterations, and the graph computing system will complete the computing task by calling the update function on each node. The typical node update function update() follows the following logic:

[0033]

[0034] The logic process of the above pseudo code can be divided into three steps: the node vertex obtains information x through its in-edge and out-edge; uses the function f(x) to transform and update its own node value; and sends the updated value through its adjacent edge. Function g(vertex.value) represents the way to transmit information, such as writing to the adjacent edge or generating a message queue.

[0035] For each load node, parameter identification is performed, and data information of each node and adjacent edges is read and collected to obtain the electrical quantity parameter information x of the node vertex, including static characteristic parameters and dynamic characteristic parameters. The dynamic characteristic parameters of various load nodes include stator resistance , stator reactance , Excitation Reactance , rotor resistance , rotor reactance , Mechanical torque coefficient and , inertia time constant , Load rate , the proportion of motor load power consumption to total load power consumption etc.; static characteristic parameters include: voltage characteristic parameters of load active power , Frequency characteristic parameters of load active power , Voltage characteristic parameters of load reactive power , frequency characteristic parameters of load reactive power Then, to update and transfer data, the overall synchronous parallel computing model (BSP) in the Prege graph computing mode is mainly used to solve and update based on the distributed storage of graph data. Among them, each round of iteration is called a superstep, and all processes are executed in parallel. Figure 4 As shown in the figure, the left side is the adjacency matrix, and the matrix is ​​updated by the function f(x). What is updated is the electrical quantity parameters of each node after the simplified topology. The function f(x) corresponds to the circuit equivalent transformation and power flow calculation in the power system. For the circuits of the same voltage level, the equivalent transformation is performed. For different voltage levels, the calculation is done step by step. That is, if the starting voltage and the end load are known, the calculation is started from the end based on a certain feeder, and the power and node parameters of the branch head end are calculated, and so on until the power supply node is reached.

[0036] With the construction of new power systems, photovoltaic, wind power, electric vehicles and other generalized loads are connected in large quantities, and the parameters of various loads are changing dynamically. Taking the air conditioning load, which has been more prominent in recent years, as an example, the working mechanism of the air conditioner is the same as that of the heat pump working in the cooling mode. The air conditioner is generally composed of one or more fans and a compression induction motor. The real-time voltage of the air conditioner is , Current and active power Data, calculate the reactive power and power factor as follows:

[0037]

[0038] Its static model parameters and dynamic model parameters are shown in Tables 1 and 2 below.

[0039] Table 1 Static characteristic parameters

[0040]

[0041] It is the proportion of asynchronous motors, and here is the parameter of motor active load proportion corresponding to the heat pump air-conditioning load.

[0042] Table 2 Dynamic characteristic parameters

[0043]

[0044] Abstract the power system into a A weighted network of nodes , determine the relevant parameter attributes of each node in the graph. For load nodes with complex, unknown or variable load types, the power system performs flow calculations, and uses identification algorithms to identify parameters based on the flow calculation results and the uploaded data of typical sites (typical data from PMU and roadblock recorders). Therefore, the power system can be represented as a vertex and edge data set for the graph database and stored in the graph database. The data set of each physical node forms an independent data file for loading. By integrating data management with core computing, the BMP parallel computing model is adopted, all processes are executed in parallel, and message queues are used between nodes. The nodes are given computing power, making the power node status update and information transmission process more convenient and efficient, and convenient and timely to describe the evolution and changes of the power topology map and related status information.

[0045] Step three: divide the community attributes according to the correlation information between nodes.

[0046] The method of the present invention analyzes the similarity of load parameter indicators according to the improved modularity index, and performs typical classification of power loads at the same time based on the Louvain algorithm, thereby determining the typical load parameters of the node and the composition ratio of each type of power equipment, such as Figure 5 shown.

[0047] The main content of the community detection method based on modularity is: For a weighted network G with 10 nodes, by checking all possible node groupings in the graph, we find the result with the largest modularity as the optimal community division method. For the same data set, a good community division can ensure that the node association within the community is high and the node association outside the community is low. If the modularity value is relatively large, it means that the community cohesion is high and the community discovery result is good. Among them, modularity The calculation formula is as follows:

[0048] In the formula Representation Node and nodes The number of edges between Representation Node The degree, Representation Node The degree, represents the sum of all edges in the network, Representation Node The community where you live, Representation Node If the community In the same community, , otherwise 0. Set the intermediate parameter , indicating the node in the empty model and nodes The probability that there are edges between them is reflected by introducing a null model, that is, a network with some of the same structural features as the original network and randomly connected edges, to reflect the expected value of the proportion of edges within the community. is a probability matrix. This invention improves it by calculating the Euclidean distance (Euclidean distance) to evaluate the similarity between the electrical quantity parameters of each node. and Similarity as follows:

[0049]

[0050] in, The nodes are and No. electrical quantity parameters. Then the improved probability matrix , and change the intermediate parameters to = Substituting into the modularity calculation formula, it can fully reflect that the higher the similarity between node parameters, the easier it is to be classified into the same load type. Then the Louvain algorithm is used to calculate the modularity, which is divided into two stages and iterated repeatedly until the modularity function no longer increases, and the iteration is stopped to obtain the final community division result, such as Figure 6 As shown, it can be divided into two stages.

[0051] The first stage: node allocation, including: first, each node is regarded as a separate community, at which point the number of communities is the same as the number of nodes, and then each node , consider adding it to each neighbor node The community and calculate the increment of modularity , if the maximum modularity increment is max If the modularity increment is greater than zero, it will be added to the community of neighboring nodes that increases the modularity increment the most. Otherwise, if the modularity increment is not greater than zero, the node No movement is performed. Repeat the above process until all nodes no longer move, and the first stage stops. Join the community Its modularity increment It can be calculated as follows:

[0052] ;

[0053] in, Representing the community The sum of the internal edge weights, Representation and Community The sum of the edge weights connecting the nodes, For the node The sum of the weights of the connected edges, i.e. the nodes The degree, Representation Node With the community The sum of the weights of the edges connecting the node.

[0054] The second stage: compressing the graph, including: reconstructing the network using the communities discovered in the first stage, treating each community in the first stage as a new node. At this time, the weight of the new node is the sum of the weights of the nodes within the community, the sum of the weights of the edges between the nodes in the community is used as the weight of the ring of the new node, and the weight of the edge between communities is converted into the weight of the edge between two new nodes. Through the above steps, a new graph can be obtained.

[0055] Repeat the above two-stage process until the modularity of the entire community no longer changes, stop iteration, and obtain the final community reconstruction network.

[0056] According to the results of community division, that is, the similarity of dynamic characteristic parameters of load nodes, they are classified and marked.

[0057] Step 4: Statistically calculate the various load types and composition ratios of the node, and obtain the comprehensive load model of the load node through a weighted method. By continuously pushing up and weighted iterative calculations, the comprehensive load model corresponding to the substation is obtained.

[0058] Taking advantage of the layered and zoned characteristics of the power system load, the loads under the same voltage level (bus) are aggregated and directly calculated, and the equivalent models of the power flow calculation at different levels are pushed forward and back. Through continuous upward push and weighted iteration, the final comprehensive load model is obtained.

[0059] Except for the technical features described in the specification, all of them are known technologies to those skilled in the art. The present invention omits the description of known components and known technologies to avoid redundancy and unnecessary limitation of the present invention. The implementation methods described in the above embodiments do not represent all implementation methods consistent with the present application. On the basis of the technical solution of the present invention, various modifications or variations that can be made by those skilled in the art without creative labor are still within the protection scope of the present invention.

Claims

1. A new power system load modeling method based on graph computing, characterized in that: The steps include: Step 1: Establish a graphical model of the power system in the studied area. The graphical model is a directed weighted graph. The nodes in the graph are the main transformers or power loads of the substation. The edges between the nodes are established according to the transmission and distribution lines, and the direction of the edges is the direction of the power flow. Step 2: Perform parallel calculation of the data structure of the graph model of the power system in the studied area and update the electrical quantity parameter values ​​of the nodes; the parallel calculation means updating and obtaining the electrical quantity parameters of each power load node, and then sending the electrical quantity parameters of the node through the connected edges; Step 3: Divide the nodes into communities based on the improved modularity; The improved modularity Calculated as: ; Intermediate parameters ;in, represents the total number of edges in the graph model, Representation Node and nodes The number of edges between Respectively represent nodes degree, node The degree, Respectively represent nodes ,,node The community where the node is located. In the same community, ,otherwise ; Through the computing node The Euclidean distance between the electrical quantity parameters is obtained; Step 4: Based on the community reconstruction network obtained in step 3, the various load types and composition ratios of the nodes are counted, and the comprehensive load model of each load node is obtained by a weighted method, and then the comprehensive load model of the substation is obtained.

2. The method according to claim 1, characterized in that In the step 1, the attributes of the main transformer node are set to include the equipment identification, the equipment to which it belongs, the capacity and the equipment load; the attributes of the power load node are set to include the equipment identification, the equipment to which it belongs, the load transfer object, the load limit and the actual load; the attributes of the edge are set to include the edge name, state and capacity, and the state attribute of the edge takes a value of 0 or 1, 0 indicates that the line is disconnected, and 1 indicates that the line is closed.

3. The method according to claim 1 or 2, characterized in that: In step 2, each main transformer node is taken as the starting point, and each power load node is visited by traversing the edge with state 1 to obtain the electrical quantity parameters of the load node; the electrical quantity parameter x includes static characteristic parameters and dynamic characteristic parameters, and the static characteristic parameters include the voltage characteristic parameters of the load active power. , Frequency characteristic parameters of load active power , Voltage characteristic parameters of load reactive power Frequency characteristic parameters of reactive power of load , the dynamic characteristic parameters include stator resistance , stator reactance , Excitation Reactance , rotor resistance , rotor reactance , Mechanical torque coefficient and , inertia time constant , Load rate And the proportion of motor load power consumption to total load power consumption .

4. The method according to claim 1, characterized in that: In the step 3, when dividing the nodes into communities, it includes: Phase 1: Treat each node as a separate community. , calculate and add the node to the surrounding neighbor nodes The modularity increment after the community If the maximum modularity increment is greater than zero, the node is added to the community of neighboring nodes that maximizes the modularity increment. If the maximum modularity increment is not greater than zero, the node No movement is performed; the above process is repeated until all nodes no longer move, and the first stage stops; Join the community Its modularity increment It can be calculated as follows: ;in, Representing the community The sum of the weights of the internal edges, Representation and Community The sum of the edge weights connecting the nodes, Representation Node With the community The sum of the weights of the edges connecting the node; Phase 2: Reconstruct the network using the communities found in Phase 1. Treat each community in Phase 1 as a new node. The weight of the new node is the sum of the weights of the nodes within the community, the weights of the edges between the nodes within the community, and the weight of the ring as the new node. The weight of the edge between communities is converted into the weight of the edge between two new nodes. Repeat the above two-stage process until the entire community modularity no longer changes, stop iteration, and obtain the final community reconstruction network.

Images