A simulation method of a combined weak power grid simulation device
By using a combined weak grid simulation device, current, voltage, and power data are collected in real time, abnormal nodes are screened, a weighted graph model is constructed, and the fault area is accurately located. This solves the problem of difficulty in determining the fault direction in combined weak grids and improves the accuracy and robustness of fault location.
Patent Information
- Application Number
- CN202510830294.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-20
- Publication Date
- 2025-11-18
- Estimated Expiration
- 2045-06-20
AI Technical Summary
In a combined weak network, fault current flows into or out of the fault point from multiple directions, making it difficult to accurately determine the fault direction, which in turn affects the accuracy of fault location and the stability of the combined weak network.
A combined weak grid simulation device is used to collect current, voltage and power data of key nodes in real time. Abnormal nodes are screened using a mutation point detection algorithm. Combined with non-negative matrix factorization and link analysis algorithms, a weighted graph model is constructed to analyze the correlation and fault contribution between nodes and accurately locate the fault area.
It improves the accuracy of fault location, avoids misjudgment and malfunction of relay protection devices, and enhances the robustness of fault location and the stability of combined weak power grids.
Smart Images

Figure CN120657860B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of combined weak grid simulation technology, and specifically to a simulation method for a combined weak grid simulation device. Background Technology
[0002] Combined weak grids refer to grid structures composed of multiple small, distributed power sources and loads in distributed energy, microgrids, and complex power systems. Currently, combined weak grid simulation devices are commonly used to simultaneously simulate various complex operating conditions such as the access of multiple distributed power sources, changes in different loads, and dynamic adjustments to the grid topology, providing a controllable and repeatable experimental platform for weak grid research.
[0003] However, combined weak current grids consist of multiple distributed power sources and loads, and their power direction and flow are multidirectional and dynamic. When simulating power supply, due to the diverse distribution and connection methods of the power sources, fault current may flow into or out of the fault point from multiple directions, making fault direction determination complex. In the operation of weak current grids, accurately determining the fault direction is a key link in quickly locating faults, isolating fault areas, and restoring power supply. If the fault direction cannot be accurately determined, it will lead to untimely fault isolation, causing the fault range to expand and affecting the power supply reliability of more users. Existing technologies may misjudge when locating faults in multi-power supply systems, which may cause relay protection devices to malfunction, and even affect the stability of combined weak current grids, increasing the operational risks of the grid. Summary of the Invention
[0004] In view of the above, it is necessary to provide a simulation method for a combined weak current grid simulation device, which improves the accuracy of fault location compared with the traditional simulation method for combined weak current grid simulation devices.
[0005] This application provides a simulation method for a combined weak current grid simulation device, the method comprising the following steps:
[0006] A combined weak grid simulation device was used to obtain a graphical model of the combined weak grid, and the current, voltage and power data of each key node in the graphical model were collected in real time.
[0007] By comprehensively analyzing the distribution of abrupt changes in current, voltage, and power data at each key node, abnormal nodes are extracted from all key nodes.
[0008] By analyzing the correlation of current changes, voltage changes, and power changes between any abnormal node and the other key nodes, and combining the differences in the complexity of current changes, voltage changes, and power changes between any abnormal node and the other key nodes, the correlation factors between any abnormal node and the other key nodes are obtained, and each related node of any abnormal node is selected from all key nodes.
[0009] The weight distribution of each relevant node in the low-dimensional features at the current acquisition time is evaluated by non-negative matrix factorization algorithm to obtain the fault contribution of each relevant node. Based on the connection relationship between key nodes in the graph model, and combined with the correlation factors between any abnormal node and its relevant nodes, weights are assigned to each relevant node. The fault connection weights of each relevant node are obtained by combining the correlation factors between any abnormal node and its relevant nodes, the fault contribution, and the weights.
[0010] Based on the fault connection weights, a weighted graph model is constructed. By analyzing the connection relationships between nodes in the weighted graph model, the fault region of any abnormal node is obtained.
[0011] The combined weak current grid simulation device includes a power supply module, an impedance matching module, a load simulation module, and a control and monitoring system. The transformer in the power supply module is designed with dual voltage levels; the reactor in the impedance matching module is used to adjust the impedance to achieve a specific short-circuit ratio; the load simulation module is used to adjust the load size and type; and the control and monitoring system uses power system simulation software to simulate the combined weak current grid and obtain a graphical model of the combined weak current grid.
[0012] In one embodiment, the extraction process for each abnormal node is as follows:
[0013] All current, voltage, and power data collected at each key node are arranged in time sequence to form the current sequence, voltage sequence, and power sequence of each key node.
[0014] A mutation point detection algorithm was used to obtain each mutation point in each current sequence, voltage sequence, and power sequence.
[0015] The occurrence times of all abrupt changes in the current, voltage, and power sequences of any key node are statistically analyzed. If abrupt changes are detected in all three sequences of the key node, and the dispersion of the time interval between any two adjacent abrupt changes is less than a preset threshold, then the key node is considered an abnormal node.
[0016] In one embodiment, the process of obtaining the relevant factors is as follows:
[0017] Calculate the correlation coefficients of the current sequence, voltage sequence, and power sequence between any anomalous node and each of the other key nodes; calculate the fractal dimension of the current sequence, voltage sequence, and power sequence of each key node.
[0018] Calculate the fractal dimension difference of the current sequence, the fractal dimension difference of the voltage sequence, and the fractal dimension difference of the power sequence between any of the abnormal nodes and the other key nodes respectively.
[0019] The correlation factors are positively correlated with the correlation coefficients of the current series, the voltage series, and the power series, respectively; and negatively correlated with the fractal dimension differences of the current series, the voltage series, and the power series, respectively.
[0020] In one embodiment, the expression for the correlation factor is:
[0021] In the formula, represents the correlation factor between the v-th anomalous node and the i-th critical node; w represents the number of data types collected at each critical node; , , Let represent the correlation coefficients of the current sequence, voltage sequence, and power sequence between the v-th anomalous node and the i-th critical node, respectively. , , denoted as the fractal dimension difference of the current sequence, the fractal dimension difference of the voltage sequence, and the fractal dimension difference of the power sequence between the v-th anomalous node and the i-th critical node, respectively; e represents an exponential function with the natural constant as the base.
[0022] In one embodiment, the method for selecting the relevant nodes is as follows: each key node whose correlation factor with any abnormal node is greater than a preset correlation factor threshold is selected as a relevant node of any abnormal node.
[0023] In one embodiment, the method for obtaining the fault contribution is as follows:
[0024] A feature matrix is constructed by using key nodes as column indicators and current, voltage, and power as row indicators. The data of all related nodes corresponding to any abnormal node in the feature matrix are retained, and the data of the remaining columns are set to 0 to obtain the reset feature matrix.
[0025] The nonnegative matrix factorization algorithm is used to decompose and reset the feature matrix to obtain the coefficient matrix. The sum of the data in each column of the coefficient matrix is used as the fault contribution of the key node corresponding to each column.
[0026] In one embodiment, assigning weights to the relevant nodes includes:
[0027] Any anomalous node and its related nodes in the graph model are marked. The marked graph model is then input into a link analysis algorithm, which outputs the weight vectors of the related nodes. The entropy of the weight vectors of the related nodes is used as the weight of the related nodes. The normalized value of the correlation factor between any anomalous node and its related nodes is used as the initial weight of the related nodes.
[0028] In one embodiment, the expression for the faulty connection weight is:
[0029] In the formula, This represents the fault connection weight of the j-th related node of the v-th anomalous node; This represents the correlation factor between the v-th anomalous node and its j-th related node; , ϵ represents the fault contribution degree and weight of the j-th related node of the v-th abnormal node, respectively; ϵ represents a preset positive number.
[0030] In one embodiment, the method for obtaining the fault area is as follows:
[0031] In the weighted graph model, the node weights of each related node of any abnormal node are its fault connection weights, and the node weights of the remaining key nodes are 0; a graph traversal algorithm is used to obtain each fault propagation path in the weighted graph model with any abnormal node as the root node.
[0032] Calculate the sum of the node weights of all critical nodes along each fault propagation path;
[0033] For the fault propagation path with the largest sum, count the number of edges connected to each key node on the fault propagation path, calculate the product of the node weight of each key node on the fault propagation path and the number of edges, and form the fault region of any abnormal node by combining all the key nodes and edges traversed from any abnormal node to the key node with the largest product.
[0034] This application has at least the following beneficial effects:
[0035] This application uses a mutation point detection algorithm to detect mutation points in current, voltage, and power sequences. Based on the distribution of mutation point occurrence times, it filters abnormal nodes using a time threshold. By combining the weights of linear correlation and fractal dimension differences, it solves the problem of traditional single correlation coefficients being susceptible to noise interference, accurately reflecting the electrical correlation between nodes. For the interference problem of unrelated nodes, it extracts low-dimensional features from the reset feature matrix through non-negative matrix decomposition, designs fault connection weights using a link analysis algorithm, and eliminates the influence of unrelated nodes and redundant paths by comprehensively considering node correlation and fault propagation contribution, thus filtering out key propagation paths strongly correlated with the fault. For the ambiguity of fault location, it generates fault propagation paths based on a weighted graph model and graph traversal algorithm. By multiplying the fault connection weights by the electrical connection density, it solves the problem of traditional methods ignoring the dynamic propagation characteristics of the topology, accurately locating the fault area. This improves fault location from "single-dimensional detection" to "multi-dimensional correlation analysis," significantly enhancing the accuracy and robustness of fault location and avoiding the expansion of the fault range due to misjudgment. Attached Figure Description
[0036] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0037] Figure 1 A block diagram of a combined weak current grid simulation device provided in one embodiment of this application;
[0038] Figure 2 A flowchart illustrating the steps of a simulation method for a combined weak current grid simulation device, as provided in one embodiment of this application;
[0039] Figure 3 This is a flowchart for obtaining the fault area. Detailed Implementation
[0040] In the description of the embodiments in this application, the words "exemplary," "or," and "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design scheme described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design schemes. Specifically, the use of the words "exemplary," "or," and "for example" is intended to present the relevant concepts in a specific manner.
[0041] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. It should be understood that, unless otherwise stated, " / " in this application means "or".
[0042] It should also be noted that the terms "first" and "second" in this application are used to distinguish similar objects, rather than to describe a specific order or sequence.
[0043] The components of a combined weak grid simulation device provided in this application are described in detail below with reference to the accompanying drawings.
[0044] Please see Figure 1 This document illustrates a block diagram of a combined weak current grid simulation device according to an embodiment of this application. To enhance the functionality of the combined weak current grid simulation device, a combined weak current grid simulation generator is provided. Specifically, the combined weak current grid simulation generator mainly consists of a power supply module, an impedance matching module, a load simulation module, and a control and monitoring system. In this embodiment, the initial output power of the power supply module is configured to be 3MW. The transformer in the power supply module is designed with dual voltage levels, 10KV and 35kV, with a withstand voltage capacity of 45.5kV, to simulate grid interconnection scenarios at different voltage levels. The reactor in the impedance matching module is used to adjust the impedance to achieve the simulated short-circuit ratio (SCR) of 1.0, effectively suppressing current fluctuations and simulating the low short-circuit capacity characteristics of a weak grid. The load simulation module is used to adjust the load size and type to match the output of the power supply module, simulating the load conditions in an actual power grid. The control and monitoring system uses Simulink to simulate the combined weak grid. Simulink is a widely used tool in power system simulation, providing a rich library of power system modules. It can build the topology of combined weak grids and simulate the dynamic behavior of distributed power sources (such as photovoltaic and wind power), loads, and the power grid. During the simulation, the power grid is abstracted as a graph model. The nodes in the graph model are the power source connection points, load connection points, and line intersections of three or more lines. The edges in the graph model represent the lines connecting these nodes. The construction of the graph model reflects the topology of the power grid, intuitively showing the connection relationships between power sources, loads, and lines, and providing a basic framework for subsequent data acquisition, fault analysis, and protection strategy design. By setting the above parameters, the combined weak grid simulation generator can achieve the simulation function of short-circuit ratio SCR=1.0 for 3MW and above.
[0045] The following description, in conjunction with the accompanying drawings, details the specific scheme of the simulation method for a combined weak grid simulation device provided in this application.
[0046] like Figure 2 The diagram illustrates a flowchart of a simulation method for a combined weak current grid simulation device according to an embodiment of this application. The method includes the following steps:
[0047] Step S1: Collect current data, voltage data, and power data of each key node in the graphical model in real time.
[0048] In a combined weak grid, nodes that are important for the grid's operating status and fault location are defined as critical nodes. Current sensors, voltage sensors, and power sensors are installed at each critical node to measure the current, voltage, and power data at each critical node in real time. All current, voltage, and power data collected at each critical node are normalized to eliminate the influence of dimensions.
[0049] In this embodiment, key nodes include: each power access point of the distributed power source, each load access point, and the intersection of three or more lines.
[0050] In this embodiment, the acquisition frequency of current data, voltage data and power data is 10kHz. The acquisition frequency value is preset by the user and can be set by the implementer. This application does not impose any special restrictions.
[0051] In this embodiment, the softmax function is used to normalize the current data, voltage data, and power data respectively.
[0052] All current, voltage, and power data at each key node are arranged in time sequence to form the current sequence, voltage sequence, and power sequence for each key node. A feature matrix is constructed by using the key nodes as column indicators and the current, voltage, and power as row indicators.
[0053] In this embodiment, the first row of the feature matrix contains current data, the second row contains voltage data, and the third row contains power data. The number of columns in the feature matrix is equal to the number of critical nodes.
[0054] Step S2: By comprehensively analyzing the distribution of abrupt changes in the current, voltage, and power data of each key node, extract each abnormal node from all key nodes.
[0055] Real-time fault monitoring of combined low-voltage power grids by measuring current, voltage, and power at key nodes can provide a basis for subsequent fault location and isolation. Delayed fault monitoring may lead to an expansion of the fault area, increasing repair difficulty and power outage time, and even triggering more serious power accidents.
[0056] For the current, voltage, and power sequences of key nodes in a combined weak current network, a mutation point detection algorithm is used to obtain each mutation point in the current, voltage, and power sequences. The occurrence times of all mutation points in the current, voltage, and power sequences of any key node are statistically analyzed. If mutation points are detected in all three sequences of any key node, and the standard deviation of the time interval between any two adjacent mutation point occurrence times is less than a preset threshold, then that key node is determined to be connected to a fault point and is designated as an abnormal node. For example, a sudden increase in current or a sudden drop in voltage at an abnormal node may indicate a short-circuit fault, and a sudden change in power may indicate an abnormal power supply or load.
[0057] In this embodiment, the Cumulative Sum Control Chart (CUSUM) algorithm is used to obtain each abrupt change point in each current sequence, voltage sequence, and power sequence. The CUSUM algorithm is a well-known technology and will not be described in detail in this application. As other implementation methods, based on the ability to obtain each abrupt change point in each current sequence, voltage sequence, and power sequence, implementers may use other existing technologies, such as Bayesian abrupt change point detection algorithm, Mann-Kendall abrupt change point detection algorithm, etc. This application does not impose any special restrictions.
[0058] In this embodiment, the preset threshold value is 0.02. The preset threshold value is preset by the user and can be set by the implementer. This application does not impose any special restrictions.
[0059] Step S3: By analyzing the correlation of current changes, voltage changes, and power changes between any abnormal node and the other key nodes, and combining the differences in the complexity of current changes, voltage changes, and power changes between any abnormal node and the other key nodes, the correlation factors between any abnormal node and the other key nodes are obtained, and the relevant nodes of any abnormal node are selected from all key nodes.
[0060] In combined low-voltage power grids, fault currents may flow into or out of the fault point from multiple directions, making it difficult to determine the fault direction. By screening nodes directly connected to the fault point, the fault range can be narrowed down, avoiding misjudgments and malfunctions of relay protection devices. In a circuit, current at any node follows Kirchhoff's Current Law, meaning the sum of the currents flowing into a node equals the sum of the currents flowing out of the node. If there is a direct electrical connection between two nodes, such as through a conductor or transformer, a change in the current at one node will cause a change in the current at the connected node. For example, if the current at one node suddenly increases, the current at the next connected node will also increase accordingly. Therefore, when the current, voltage, or power of a node changes, nodes that are electrically close to it or directly connected will also be affected, exhibiting a correlation.
[0061] Taking any abnormal node as an example, calculate the correlation coefficient of the current sequence between the abnormal node and each of the other key nodes, calculate the correlation coefficient of the voltage sequence between the abnormal node and each of the other key nodes, calculate the correlation coefficient of the power sequence between the abnormal node and each of the other key nodes, and calculate the fractal dimension of the current sequence, the fractal dimension of the voltage sequence, and the fractal dimension of the power sequence of each key node.
[0062] In this embodiment, the correlation coefficients between current sequences, voltage sequences, and power sequences are all Pearson correlation coefficients. Pearson correlation coefficients are well-known technologies and will not be described in detail here. As other implementation methods, based on the ability to calculate the correlation coefficients between current sequences, voltage sequences, and power sequences, implementers may use other existing technologies, such as Spearman correlation coefficients, Kendall rank correlation coefficients, etc. This application does not impose any special restrictions.
[0063] In this embodiment, the box counting method is used to calculate the fractal dimension of the current sequence, the fractal dimension of the voltage sequence, and the fractal dimension of the power sequence, respectively. The box counting method is a well-known technology and will not be described in detail in this application. As other implementation methods, based on the ability to calculate the fractal dimension of the current sequence, the fractal dimension of the voltage sequence, and the fractal dimension of the power sequence, the implementer may use other existing technologies, such as the Hurst exponent method, the Higuchi algorithm, etc. This application does not impose any special restrictions.
[0064] Based on the above analysis, by using the correlation coefficients of the current sequence, voltage sequence, and power sequence between any anomalous node and the other key nodes, and combining the fractal dimension differences of the current sequence, voltage sequence, and power sequence between any anomalous node and the other key nodes, the correlation factor between any anomalous node and the other key nodes is obtained, and its expression is:
[0065] In the formula, represents the correlation factor between the v-th anomalous node and the i-th critical node; w represents the number of data types collected for each critical node; , , Let represent the correlation coefficients of the current sequence, voltage sequence, and power sequence between the v-th anomalous node and the i-th critical node, respectively. , , Let represent the fractal dimension differences of the current sequence, voltage sequence, and power sequence between the v-th anomalous node and the i-th critical node, respectively; e represents an exponential function with the natural constant as its base, used to... , , The mapping is to positive numbers. In this embodiment, the value of w is 3.
[0066] In this embodiment, the differences involved in the calculation of relevant factors are all absolute values of the differences. As other implementation methods, based on the ability to measure the differences between fractal dimensions, the implementer may use other calculation methods for measurement, such as the square of the difference, the ratio, etc. This application does not impose any special restrictions.
[0067] It should be noted that: This reflects the synchronicity of current changes between the v-th anomalous node and the i-th critical node. This reflects the difference in current change between the v-th anomalous node and the i-th critical node. As the value increases, the complexity of current changes between the v-th anomalous node and the i-th critical node becomes more distinct, and the correlation between their current changes is weakened; when... The larger the value, the stronger the consistency in current change trends between the v-th abnormal node and the i-th critical node, and the more similar the complexity of current changes. This indicates a stronger current pair. The greater the contribution, the better. Based on the above analysis, for and right Analyze the degree of contribution. When The larger the value, the stronger the correlation between the v-th abnormal node and the i-th critical node, indicating that the i-th critical node is more likely to be affected by the fault point or directly connected to the fault point.
[0068] Since the correlation factor can directly reflect the electrical connection relationship between any abnormal node and the other key nodes, a correlation coefficient threshold is set. Key nodes whose correlation factor with any abnormal node is greater than the correlation factor threshold are regarded as the relevant nodes of any abnormal node. The relevant nodes have a strong correlation with any abnormal node in electrical connection and are more likely to be directly affected by the fault point or directly connected to the fault point.
[0069] In this embodiment, the correlation coefficient threshold is 0.6. The value of the correlation coefficient threshold is preset by the user and can be set by the implementer. This application does not impose any special restrictions.
[0070] Step S4: Evaluate the weight distribution of each relevant node in the low-dimensional features at the current acquisition time using a non-negative matrix factorization algorithm to obtain the fault contribution of each relevant node. Based on the connection relationship between key nodes in the graph model, and combined with the correlation factors between any abnormal node and its relevant nodes, assign weights to each relevant node. Combine the correlation factors between any abnormal node and its relevant nodes, the fault contribution, and the weights to obtain the fault connection weights of each relevant node.
[0071] To more intuitively analyze the electrical connection relationship between any abnormal node and its related nodes, the data in the columns corresponding to all related nodes of any abnormal node in the feature matrix are retained, and the data in the remaining columns are assigned a value of 0, thus obtaining a reset feature matrix.
[0072] Using the reset eigenvalue matrix as input to the nonnegative matrix factorization algorithm, setting the rank r=2, the output consists of two nonnegative matrices, including one of size r. The basis matrix and its size are The coefficient matrix is defined as follows: w represents the number of data types collected by each key node, and N represents the number of key nodes. The sum of the data in each column of the coefficient matrix is used as the fault contribution of the key node corresponding to each column. Each column of the coefficient matrix corresponds to a key node, and each row corresponds to a low-dimensional feature. The low-dimensional feature can reflect the similarity and differences between key nodes. The magnitude of the key node on the low-dimensional feature can reflect the correlation between the key node and the fault signal. If the value of any key node on the low-dimensional feature is larger, it indicates that the key node may have played a more important role in the propagation of the fault signal. Summing the data in each column of the coefficient matrix can quantify the total contribution of the key node corresponding to each column on all low-dimensional features.
[0073] Furthermore, any anomalous node and its related nodes in the graph model are labeled. The labeled graph model is then input into a link analysis algorithm, which outputs a weight vector for each related node. The entropy of the weight vector of each related node is used as the weight of each related node. The normalized value of the correlation factor between any anomalous node and its related nodes is used as the initial weight of each related node. The method for calculating entropy is a well-known technique and will not be elaborated upon in this application.
[0074] In this embodiment, the PageRank algorithm is used to output the weight vector of each relevant node. The PageRank algorithm is a well-known technology and will not be described in detail in this application. The damping factor of the PageRank algorithm is 0.85. As another implementation method, based on the ability to measure the importance of each key node in the graph model, the implementer can choose other feasible methods.
[0075] In this embodiment, the softmax function is used to normalize the relevant factors.
[0076] Based on the above analysis, and combining the correlation factors between any abnormal node and its related nodes, the fault contribution, and the weights, the fault connection weights of each related node of any abnormal node are obtained, expressed as follows:
[0077] In the formula, This represents the fault connection weight of the j-th related node of the v-th anomalous node; This represents the correlation factor between the v-th anomalous node and its j-th related node; , ϵ represents the fault contribution degree and weight of the j-th related node of the v-th abnormal node, respectively; ϵ represents a preset positive number to avoid the denominator being 0. The value of ϵ is preset by the user and can be set by the implementer. In this embodiment, the value of ϵ is 0.01.
[0078] It should be noted that: It reflects the overall correlation between the v-th abnormal node and its j-th related node in terms of current, voltage, and power variation trends, and measures the electrical connection strength and synchronicity between the j-th related node and the fault point. The larger the value, the stronger the correlation between the j-th related node and the fault point, and the more likely it is to be directly connected to the fault point or directly affected by the fault. The larger; The importance of the j-th relevant node in the fault propagation process was measured. The larger the value, the more important the j-th related node is in fault propagation, and the more likely the fault is to occur between the v-th anomalous node and its j-th related node. The larger; The influence and stability of the j-th relevant node in the power grid topology were measured. The smaller the value, the more concentrated the weight distribution of the j-th related node, the more likely it is to be a critical node in the power grid, and the stronger its influence on fault propagation. The larger. Used to quantify the degree of correlation between the j-th related node and the fault point. The larger the value, the stronger the connection between the j-th related node and the fault point, and the more important its role in fault propagation.
[0079] Step S5: Based on the fault connection weights, construct a weighted graph model, and obtain the fault region of any abnormal node by analyzing the connection relationships between nodes in the weighted graph model.
[0080] In a combined weak grid, an abnormal node only represents a node related to a fault. However, the exact location of the fault may be located on a line or a critical node near the abnormal node. Therefore, it is necessary to accurately locate the fault. A weighted graph model is constructed based on a graph model. In the weighted graph model, the node weight of each related node of any abnormal node is its fault connection weight, and the node weight of each other critical node is 0.
[0081] Since faults typically propagate from the fault point to surrounding critical nodes, the propagation path can be represented by edges in a weighted graph model. The intensity of fault propagation is related to the node weights of relevant nodes; critical nodes with higher weights are more likely to be important nodes in the fault propagation. Taking any anomalous node as the root node, a graph traversal algorithm is used to obtain each fault propagation path in the weighted graph model with the anomalous node as the root node. The sum of the node weights of all critical nodes on each fault propagation path is calculated. For the fault propagation path with the largest sum, the number of edges connected to each critical node on the fault propagation path is counted, and the product of the node weight of each critical node on the fault propagation path and the number of edges is calculated. All critical nodes and edges traversed from the anomalous node to the critical node with the largest product constitute the fault region of the anomalous node. The flowchart for obtaining the fault region is as follows: Figure 3 As shown.
[0082] It should be noted that: node weight is used to reflect the importance of critical nodes in fault propagation. The larger the node weight, the more likely the critical node is to be directly connected to the fault point or directly affected by the fault point, and the more important its role in the fault propagation process. The number of edges connected to a critical node represents the number of electrical connections of the critical node in the power grid. The larger the number of edges, the more other critical nodes the critical node is connected to, and the higher the probability that the fault signal will propagate through these connections.
[0083] In this embodiment, the fault propagation paths in the weighted graph model with any abnormal node as the root node are obtained by the breadth-first search algorithm. As another implementation method, based on the ability to obtain the fault propagation paths in the weighted graph model with any abnormal node as the root node, the implementer can choose other feasible algorithms.
[0084] In summary, this application uses a mutation point detection algorithm to detect mutation points in current, voltage, and power sequences. Based on the distribution of mutation point occurrence times, it filters abnormal nodes using a time threshold. By combining the weights of linear correlation and fractal dimension differences, it solves the problem of traditional single correlation coefficients being susceptible to noise interference, accurately reflecting the electrical correlation between nodes. For the interference problem of irrelevant nodes, it extracts low-dimensional features from the reset feature matrix through non-negative matrix decomposition, designs fault connection weights using a link analysis algorithm, and eliminates the influence of irrelevant nodes and redundant paths by comprehensively considering node correlation and fault propagation contribution, thus filtering out key propagation paths strongly correlated with the fault. For the ambiguity of fault location, it generates fault propagation paths based on a weighted graph model and graph traversal algorithm. By multiplying the fault connection weights by the electrical connection density, it solves the problem of traditional methods ignoring the dynamic propagation characteristics of the topology, accurately locating the fault area. This improves fault location from "single-dimensional detection" to "multi-dimensional correlation analysis," significantly enhancing the accuracy and robustness of fault location and avoiding the expansion of the fault range due to misjudgment.
[0085] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions marked in the blocks may occur in a different order than that shown in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. In the descriptions corresponding to the flowcharts and block diagrams in the accompanying drawings, the operations or steps corresponding to different blocks may also occur in a different order than disclosed in the description, and sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. Each block in a block diagram and / or flowchart, and combinations of blocks in a block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.
[0086] It will be apparent to those skilled in the art that this application is not limited to the details of the exemplary embodiments described above, and that this application can be implemented in other specific forms without departing from its essential characteristics. Therefore, the embodiments described above should be considered exemplary and non-limiting in all respects.
Claims
1. A simulation method for a combined weak current grid simulation device, characterized in that, The method includes the following steps: A combined weak grid simulation device was used to obtain a graphical model of the combined weak grid, and the current, voltage and power data of each key node in the graphical model were collected in real time. By comprehensively analyzing the distribution of abrupt changes in current, voltage, and power data at each key node, abnormal nodes are extracted from all key nodes. By analyzing the correlation of current changes, voltage changes, and power changes between any abnormal node and the other key nodes, and combining the differences in the complexity of current changes, voltage changes, and power changes between any abnormal node and the other key nodes, the correlation factors between any abnormal node and the other key nodes are obtained, and each related node of any abnormal node is selected from all key nodes. The weight distribution of each relevant node in the low-dimensional features at the current acquisition time is evaluated by non-negative matrix factorization algorithm to obtain the fault contribution of each relevant node. Based on the connection relationship between key nodes in the graph model, and combined with the correlation factors between any abnormal node and its relevant nodes, weights are assigned to each relevant node. The fault connection weights of each relevant node are obtained by combining the correlation factors between any abnormal node and its relevant nodes, the fault contribution, and the weights. Based on the fault connection weights, a weighted graph model is constructed. By analyzing the connection relationships between nodes in the weighted graph model, the fault region of any abnormal node is obtained. The combined weak current grid simulation device includes a power supply module, an impedance matching module, a load simulation module, and a control and monitoring system. The transformer in the power supply module is designed with dual voltage levels; the reactor in the impedance matching module is used to adjust the impedance to achieve a specific short-circuit ratio; the load simulation module is used to adjust the load size and type; and the control and monitoring system uses power system simulation software to simulate the combined weak current grid and obtain a graphical model of the combined weak current grid.
2. The simulation method of the combined weak current grid simulation device as described in claim 1, characterized in that, The extraction process for each abnormal node is as follows: All current, voltage, and power data collected at each key node are arranged in time sequence to form the current sequence, voltage sequence, and power sequence of each key node. A mutation point detection algorithm was used to obtain each mutation point in each current sequence, voltage sequence, and power sequence. The occurrence times of all abrupt changes in the current, voltage, and power sequences of any key node are statistically analyzed. If abrupt changes are detected in all three sequences of the key node, and the dispersion of the time interval between any two adjacent abrupt changes is less than a preset threshold, then the key node is considered an abnormal node.
3. The simulation method of the combined weak current grid simulation device as described in claim 1, characterized in that, The process of obtaining the relevant factors is as follows: Calculate the correlation coefficients of the current sequence, voltage sequence, and power sequence between any anomalous node and each of the other key nodes; calculate the fractal dimension of the current sequence, voltage sequence, and power sequence of each key node. Calculate the fractal dimension difference of the current sequence, the fractal dimension difference of the voltage sequence, and the fractal dimension difference of the power sequence between any of the abnormal nodes and the other key nodes respectively. The correlation factors are positively correlated with the correlation coefficients of the current series, the voltage series, and the power series, respectively. It is negatively correlated with the fractal dimension differences of current series, voltage series, and power series, respectively.
4. The simulation method of the combined weak current grid simulation device as described in claim 3, characterized in that, The expression for the relevant factor is: In the formula, represents the correlation factor between the v-th anomalous node and the i-th critical node; w represents the number of data types collected at each critical node; , , Let represent the correlation coefficients of the current sequence, voltage sequence, and power sequence between the v-th anomalous node and the i-th critical node, respectively. , , denoted as the fractal dimension difference of the current sequence, the fractal dimension difference of the voltage sequence, and the fractal dimension difference of the power sequence between the v-th anomalous node and the i-th critical node, respectively; e represents an exponential function with the natural constant as the base.
5. The simulation method of the combined weak current grid simulation device as described in claim 1, characterized in that, The method for selecting each relevant node is as follows: each key node whose correlation factor with any abnormal node is greater than a preset correlation factor threshold is selected as a relevant node of any abnormal node.
6. The simulation method of the combined weak current grid simulation device as described in claim 1, characterized in that, The method for obtaining the fault contribution is as follows: By using key nodes as column indicators and current, voltage, and power as row indicators, a feature matrix is constructed. The data in the columns corresponding to all related nodes of any abnormal node in the feature matrix is retained, and the data in the remaining columns is set to 0 to obtain the reset feature matrix; The nonnegative matrix factorization algorithm is used to decompose and reset the feature matrix to obtain the coefficient matrix. The sum of the data in each column of the coefficient matrix is used as the fault contribution of the key node corresponding to each column.
7. The simulation method of the combined weak current grid simulation device as described in claim 1, characterized in that, The assignment of weights to the relevant nodes includes: Any anomalous node and its related nodes in the graph model are marked. The marked graph model is then input into a link analysis algorithm, which outputs the weight vectors of the related nodes. The entropy of the weight vectors of the related nodes is used as the weight of the related nodes. The normalized value of the correlation factor between any anomalous node and its related nodes is used as the initial weight of the related nodes.
8. The simulation method of the combined weak current grid simulation device as described in claim 1, characterized in that, The expression for the faulty connection weight is: In the formula, This represents the fault connection weight of the j-th related node of the v-th anomalous node; This represents the correlation factor between the v-th anomalous node and its j-th related node; , ϵ represents the fault contribution degree and weight of the j-th related node of the v-th abnormal node, respectively; ϵ represents a preset positive number.
9. The simulation method of the combined weak current grid simulation device as described in claim 1, characterized in that, The method for obtaining the fault area is as follows: In the weighted graph model, the node weights of each related node of any abnormal node are its fault connection weights, and the node weights of the remaining key nodes are 0; a graph traversal algorithm is used to obtain each fault propagation path in the weighted graph model with any abnormal node as the root node. Calculate the sum of the node weights of all critical nodes along each fault propagation path; For the fault propagation path with the largest sum, count the number of edges connected to each key node on the fault propagation path, calculate the product of the node weight of each key node on the fault propagation path and the number of edges, and form the fault region of any abnormal node by combining all the key nodes and edges traversed from any abnormal node to the key node with the largest product.
Citation Information
Patent Citations
Power distribution network fault diagnosis and positioning method based on super-resolution and graph neural network
CN114779015A
Automotive suspension strength testing method based on artificial intelligence
CN118332473A