Fault diagnosis method and system for electric power centralized communication control device

Abstract

The invention discloses a fault diagnosis method and system for an electric power centralized communication control device, which are used for accurately identifying cascade faults so as to improve the safety and stability of an electric power system. The method comprises the following steps: acquiring operation data of the power centralized communication control device, and extracting a frequency feature set from the operation data; calculating association strength among variables in the frequency feature set by using a conditional probability table, and screening out a key feature vector set to generate an interactive topology among different devices in the power system; calculating propagation paths and propagation speeds of the key feature vectors in the interactive topology according to a path reasoning algorithm, and screening out a candidate fault propagation path set based on the propagation paths and the propagation speeds; calculating fault propagation probabilities of the candidate fault propagation paths to determine a cascade fault propagation feature set; and constructing a high-dimensional space classification boundary according to the cascade fault propagation feature set so as to perform fault mode classification and obtain a fault diagnosis result.

Description

Technical Field

[0001] The present application relates to the technical field of power systems, and in particular to a fault diagnosis method and system for a power centralized communication control device. Background Art

[0002] Centralized power communication control devices are key equipment for information transmission, centralized control, and coordinated operation across power systems. They are widely used in substations, distribution networks, and power system dispatching. Within power systems, these devices perform important functions such as real-time monitoring, data transmission, and command coordination. Their operational status is crucial to the stability of the entire power grid.

[0003] Existing fault diagnosis methods for centralized power communication control devices typically rely on monitoring parameters of individual devices in the power system to identify faults. However, as power systems continue to expand in size and become increasingly complex, they fail to fully consider the complex interactions between devices and the fault propagation characteristics when faced with complex fault scenarios. This makes it difficult to detect cascading faults in a timely manner, thereby reducing the safety and stability of the power system. Summary of the Invention

[0004] The present application provides a fault diagnosis method and system for a centralized power communication control device, which can accurately identify cascading faults to improve the safety and stability of the power system.

[0005] A first aspect of the present application provides a fault diagnosis method for a centralized power communication control device, comprising: Acquiring operating data of a centralized power communication control device and extracting a frequency feature set from the operating data; Calculating the correlation strength between the variables in the frequency feature set using a conditional probability table, and screening out a set of key feature vectors to generate an interaction topology between different devices in the power system; Calculating the propagation path and propagation speed of each key feature vector in the interaction topology according to a path inference algorithm, and screening a set of candidate fault propagation paths based on the propagation path and propagation speed; Calculating the fault propagation probability of the candidate fault propagation path to determine the cascade fault propagation feature set; A high-dimensional space classification boundary is constructed according to the cascade fault propagation feature set to perform fault mode classification and obtain a fault diagnosis result.

[0006] Optionally, constructing a high-dimensional space classification boundary based on the cascade fault propagation feature set to perform fault mode classification and obtain a fault diagnosis result includes: Extracting a cascading fault propagation feature vector from the cascading fault propagation feature set; Using the cascading fault propagation feature vector as an input to a support vector machine to construct a high-dimensional space classification boundary; Adjusting the interval of the high-dimensional space classification boundary by a kernel function; Calculating a distance metric of the classification boundary of the high-dimensional space after adjusting the interval; Determining whether the distance metric is less than a preset distance threshold; If yes, the failure mode is determined to be a cascading failure.

[0007] Optionally, calculating the propagation path and propagation speed of each key feature vector in the interaction topology according to a path inference algorithm, and screening out a set of candidate fault propagation paths based on the propagation path and propagation speed, includes: Calculate the propagation path of each key feature vector in the interaction topology according to the path inference algorithm; Calculate the propagation speed of each propagation path based on the factors affecting the propagation speed and the corresponding assigned weights; Determine a propagation path with a propagation speed greater than a preset speed threshold as a candidate sub-path; Calculating the path coverage of the candidate subpath; The candidate sub-paths whose path coverage range is greater than a preset coverage threshold are determined as candidate fault propagation paths to obtain a set of candidate fault propagation paths.

[0008] Optionally, calculating the fault propagation probability of the candidate fault propagation path to determine the cascading fault propagation feature set includes: Construct an initial state transfer matrix based on candidate fault propagation paths; Adopting an evidence updating mechanism to adjust the conditional probability distribution in the initial state transfer matrix; Calculating the convergence index of the adjusted conditional probability distribution to obtain a convergence constraint result; Filtering a target path set from the candidate fault propagation path set based on the convergence constraint result; Constructing a path feature matrix based on the target path set; The fault propagation probability of each target path is calculated according to the path characteristic matrix, and a cascade fault propagation feature set is determined based on the fault propagation probability.

[0009] Optionally, the method of calculating the correlation strength between the variables in the frequency feature set using a conditional probability table and screening out a set of key feature vectors to generate an interaction topology between different devices in the power system includes: Calculate the conditional probability of state associations between all variables in the frequency feature set and construct a complete conditional probability table; Determining the association strength between the variables in the frequency feature set according to the conditional probability table; The variable combination with the correlation strength greater than the preset correlation threshold is determined as the key feature vector; Taking each device in the power system as a node and the key eigenvectors as edges, the interaction topology between different devices in the power system is generated.

[0010] Optionally, obtaining operation data of the electric power centralized communication control device and extracting a frequency feature set from the operation data includes: Obtaining operating data of the power centralized communication control device; Slice the operating data to obtain a slicing data set; Extracting a dynamic data set from the fragmented data set; Extracting frequency components and periodic features from the dynamic data set using a small Bode transform; A frequency feature set is generated based on the frequency components and the periodic features.

[0011] A second aspect of the present application provides a fault diagnosis system for a centralized power communication control device, comprising: an acquisition unit, configured to acquire operating data of the electric power centralized communication control device and extract a frequency feature set from the operating data; a first calculation unit, configured to calculate the correlation strength between the variables in the frequency feature set using a conditional probability table, and screen out a set of key feature vectors to generate an interaction topology between different devices in the power system; a second calculation unit, configured to calculate a propagation path and a propagation speed of each key feature vector in the interaction topology according to a path inference algorithm, and screen out a set of candidate fault propagation paths based on the propagation path and the propagation speed; a determination unit, configured to calculate a fault propagation probability of a candidate fault propagation path to determine a cascade fault propagation feature set; The classification unit is used to construct a high-dimensional space classification boundary according to the cascade fault propagation feature set to perform fault mode classification and obtain a fault diagnosis result.

[0012] Optionally, the classification unit is specifically used to: Extracting a cascading fault propagation feature vector from the cascading fault propagation feature set; Using the cascading fault propagation feature vector as an input to a support vector machine to construct a high-dimensional space classification boundary; Adjusting the interval of the high-dimensional space classification boundary by a kernel function; Calculating a distance metric of the classification boundary of the high-dimensional space after adjusting the interval; Determining whether the distance metric is less than a preset distance threshold; If yes, the failure mode is determined to be a cascading failure.

[0013] Optionally, the second computing unit is specifically configured to: Calculate the propagation path of each key feature vector in the interaction topology according to the path inference algorithm; Calculate the propagation speed of each propagation path based on the factors affecting the propagation speed and the corresponding assigned weights; Determine a propagation path with a propagation speed greater than a preset speed threshold as a candidate sub-path; Calculating the path coverage of the candidate subpath; The candidate sub-paths whose path coverage range is greater than a preset coverage threshold are determined as candidate fault propagation paths to obtain a set of candidate fault propagation paths.

[0014] Optionally, the determining unit is specifically configured to: Construct an initial state transfer matrix based on candidate fault propagation paths; Adopting an evidence updating mechanism to adjust the conditional probability distribution in the initial state transfer matrix; Calculating the convergence index of the adjusted conditional probability distribution to obtain a convergence constraint result; Filtering a target path set from the candidate fault propagation path set based on the convergence constraint result; Constructing a path feature matrix based on the target path set; The fault propagation probability of each target path is calculated according to the path characteristic matrix, and a cascade fault propagation feature set is determined based on the fault propagation probability.

[0015] It can be seen from the above technical solutions that this application has the following effects: The method obtains operating data from a centralized power communication control device and extracts a frequency feature set from the operating data. A conditional probability table is used to calculate the correlation strength between the variables in the frequency feature set, screening a set of key feature vectors to generate an interaction topology between different devices in the power system. A path inference algorithm is used to calculate the propagation path and propagation velocity of each key feature vector in the interaction topology, and based on the propagation path and propagation velocity, a set of candidate fault propagation paths is screened. The fault propagation probability of each candidate fault propagation path is calculated to determine a set of cascading fault propagation features. A high-dimensional spatial classification boundary is constructed based on the set of cascading fault propagation features to classify fault modes and obtain fault diagnosis results. In this way, the correlation strength between the corresponding variables of each device in the power system is used to screen out key feature vectors with correlation. The propagation path and propagation velocity of the key feature vectors are then obtained to determine a set of cascading fault path features where cascading faults may occur. Finally, the specific fault mode is determined based on the high-dimensional spatial boundary. This method comprehensively considers the complex interactions between devices in the power system and the fault propagation characteristics to accurately identify cascading faults, thereby improving the safety and stability of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A schematic diagram of an embodiment of a fault diagnosis method for a centralized power communication control device in this application; Figure 2-1 、 Figure 2-2 as well as Figure 2-3 This is a schematic diagram of another embodiment of a fault diagnosis method for a centralized power communication control device in the present application; Figure 3 This is a schematic diagram of an embodiment of a fault diagnosis system for a centralized power communication control device in the present application. DETAILED DESCRIPTION

[0017] The present application provides a fault diagnosis method and system for a centralized power communication control device, which are used to accurately identify cascading faults to improve the safety and stability of the power system.

[0018] The present application describes a method for diagnosing a fault of a centralized electric power communication control device, which is implemented on a system, a server, or a terminal with logic analysis capabilities.

[0019] See also Figure 1 As shown, an embodiment of the fault diagnosis method of the electric power centralized communication control device in the present application includes: 101. Obtain operating data of a centralized power communication control device, and extract a frequency feature set from the operating data; In this embodiment, the centralized power communication control device collects real-time operating parameters from each node in the power system, such as voltage, current, and power parameters of distributed power sources, smart meters, transformers, switchgear, and protective devices during operation. The centralized power communication control device transmits the collected operating parameters from each node to the system via a communication network, thereby acquiring the operating data of the centralized power communication control device. The operating data is converted from the time domain to the frequency domain using time-frequency analysis methods such as Fourier transform or Little Porter transform to obtain a set of frequency features.

[0020] 102. Use the conditional probability table to calculate the correlation strength between the variables in the frequency feature set, and select the key feature vector set to generate the interaction topology between different devices in the power system; In this embodiment, conditional probability refers to the probability of event A occurring given the occurrence of event B. Each variable in the frequency feature set is divided into several meaningful intervals based on its value range and actual application requirements. For example, the output frequency variable of a transformer can be divided into normal operating frequency ranges, slight fluctuation ranges, and abnormal fluctuation ranges. Similar interval divisions are also performed for other frequency-related variables, such as the frequency change rate of a power grid node. These interval divisions should comprehensively consider the operating characteristics of the power system and the actual monitoring accuracy. For each pair of variables, the number of samples in which one variable is within a specific interval and the other variable is within a different interval is counted. Based on the counted number of samples, the conditional probability is calculated: that is, the probability that one variable takes a specific value (within a specific interval) and the other variable takes a different value (within a different interval). Organizing these conditional probabilities into a table creates a conditional probability table, which clearly demonstrates the probabilistic relationships between different variables.

[0021] Conditional probability tables are used to assess the degree of dependency between variables. A higher degree of dependency indicates a stronger correlation between the variables. For example, if a change in the value of one variable causes a significant shift in the conditional probability distribution of another variable, this indicates a strong dependency between the two variables. Conversely, if the conditional probability distribution does not change significantly, this indicates a weak dependency. Then, based on the actual operating requirements and experience of the power system and the analysis results of the correlation strength, a reasonable screening criterion is established. Based on this screening criterion, variable pairs with strong correlations are selected from all pairs and used as key eigenvectors. These key eigenvectors represent the frequency characteristics of different devices in the power system that are closely connected. Finally, based on the correlation strength between the key eigenvectors, an interaction topology is constructed between the different devices in the power system to clearly demonstrate the mutual influence between the devices.

[0022] 103. Calculate the propagation path and propagation speed of each key feature vector in the interactive topology according to the path inference algorithm, and screen out a set of candidate fault propagation paths based on the propagation path and propagation speed; In this embodiment, the interactive topology structure describes the connection relationship between each node, and the key eigenvector represents the starting point or key factor that may cause the fault to propagate. In the process of using the path reasoning algorithm to explore the propagation path, starting from the node where the key eigenvector is located, based on the connection relationship between the nodes and certain rules, such as finding the shortest path or a path that meets specific conditions, the propagation trajectory of the signal or influence from the key eigenvector node to other nodes is determined to simulate the most likely propagation method that the fault will follow in the actual power system. After obtaining the propagation path, other factors that affect the propagation speed, such as time, delay, etc., are combined to calculate the propagation speed of the propagation path. It can be understood that the propagation speed reflects how fast the key eigenvector spreads in the interactive topology. According to the calculated propagation path and propagation speed, some screening conditions can be set to determine the set of candidate fault propagation paths. 104. Calculate the fault propagation probability of the candidate fault propagation path to determine a cascade fault propagation feature set; In this embodiment, the fault propagation probability of the candidate fault propagation path is calculated by a full probability formula or a Markov chain, and then the critical path is identified by the fault propagation probability, and the cascading fault propagation characteristics are extracted from the critical path. The cascading fault propagation characteristics may include characteristics such as the order of fault occurrence and the time interval. Among them, cascading faults are usually not single, isolated events, but multiple faults occur in sequence. The order in which the faults occur is an important propagation feature, which can reveal the propagation path and logical relationship of the faults in the system. For example, in the power system, a transmission line may first overload and trip, causing the related bus voltage to fluctuate, which in turn causes other lines to overcurrent, and ultimately causes multiple components to fail one after another. The time interval between adjacent faults is also an important component of the cascading fault propagation characteristics. Different time intervals may reflect the speed and mechanism of fault propagation.

[0023] 105. A high-dimensional space classification boundary is constructed based on the cascade fault propagation feature set to classify the fault mode and obtain the fault diagnosis result.

[0024] To classify fault modes, the set of cascading fault propagation features must be mapped into a high-dimensional space. Each cascading fault propagation feature is considered a dimension in this high-dimensional space, so each fault sample can be represented by a point in this space. Then, using algorithms such as support vector machines and decision trees, we search for boundaries within this high-dimensional space that distinguish different fault modes. Once the classification boundaries are established in the high-dimensional space, new fault samples are projected into this space, and their respective fault modes are determined based on the region they fall into. It should be noted that the results of fault mode classification are the same as the results of fault diagnosis.

[0025] In this embodiment, the strength of the correlation between the corresponding variables of each device in the power system can be used to screen out relevant key feature vectors, and the propagation path and propagation speed of the key feature vectors can be obtained to determine the cascading failure path feature set where cascading failures may occur. Finally, the specific failure mode is determined based on the high-dimensional space boundary. This allows for the comprehensive consideration of the complex interactions between devices in the power system and the fault propagation characteristics to accurately identify cascading failures and improve the safety and stability of the power system.

[0026] See also Figure 2-1 、 Figure 2-2 as well as Figure 2-3 As shown, another embodiment of the fault diagnosis method of the electric power centralized communication control device in the present application includes: 201. Obtaining the operating data of the power centralized communication control device; 202. Slice the running data to obtain a slicing data set; 203. Extracting a dynamic data set from the sharded data set; 204. Extract frequency components and periodic features from dynamic data sets using small Bode transform; 205. Generate a frequency feature set based on the frequency components and the periodic features.

[0027] Optionally, in this embodiment, since the operating data may be a large continuous data stream, it can be segmented into several segments according to certain rules to facilitate subsequent processing. The segmentation method can be determined based on time intervals, data volume, or other logic, dividing the overall operating data into relatively independent and characteristic segmented data sets. For example, if the collected operating data consists of continuous voltage and current values ​​recorded on a second-by-second basis, the data can be segmented according to 10-minute intervals. For example, the operating data from 9:00 AM to 9:10 AM is considered a segment, containing the voltage and current values ​​every second during these 10 minutes, forming a single segment. Within the segmented data set, some data may be relatively stable, while others may be dynamic data that varies significantly over time or due to other factors. Dynamic data can be determined based on statistics such as the variance and coefficient of variation of the segmented data. Variance measures the degree of data dispersion; a larger variance indicates greater data fluctuation. The coefficient of variation is the ratio of the standard deviation to the mean, which eliminates the influence of data magnitude and more accurately reflects the relative degree of data fluctuation. By setting a certain threshold, data with a variance or coefficient of variation greater than the threshold is screened out from the sharded data set and determined as dynamic data.

[0028] After acquiring dynamic data, it is decomposed using the Bodelet transform. By calculating the energy distribution of the wavelet coefficients, the frequency components of the signal can be determined. If the signal exhibits periodicity, the coefficients after the wavelet transform will exhibit a regular distribution at specific scales and locations. By analyzing this distribution pattern, periodic features can be extracted. Each frequency component corresponds to a dictionary entry containing information such as the frequency value, energy amplitude, and position in the original data. The periodic features record the duration of the period and the pattern of data changes within the period. By structurally organizing the frequency components and periodic features, a frequency feature set can be obtained.

[0029] 206. Calculate the conditional probability of state association between all variables in the frequency feature set and construct a complete conditional probability table; 207. Determine the strength of association between variables in the frequency feature set based on the conditional probability table; 208. Determine the variable combination whose correlation strength is greater than a preset correlation threshold as a key feature vector; 209. Using each device in the power system as a node and the key feature vector as an edge, the interaction topology between different devices in the power system is generated.

[0030] Optionally, in this embodiment, the frequency feature set includes multiple variables. To calculate the conditional probability of the state association between the variables, one simply calculates the probability of another variable being in a specific state, given the known state of one variable. While the conditional probability table provides the probabilistic dependency between the variables, this relationship needs to be further quantified into a specific numerical value to represent the strength of the association for further comparison and analysis. This can be determined using metrics such as information gain and mutual information. It should be noted that information gain refers to the degree to which the uncertainty of one variable is reduced by knowing information about another variable. For discrete variables, information gain can be calculated by calculating the change in entropy. Entropy is a measure of the uncertainty of a random variable; a higher entropy value indicates a higher uncertainty. Mutual information is a measure of the degree of mutual dependence between two variables. It is equal to the information gain of one variable with respect to the other and ranges from 0 to positive infinity. A higher value indicates a higher strength of the association between the two variables. By calculating metrics such as information gain and mutual information, the quantitative strength of the association between the variables can be extracted from the conditional probability table.

[0031] The preset correlation threshold can be set based on a variety of factors, such as system stability requirements and sensitivity to key features. When the correlation strength between variables exceeds the preset correlation threshold, it indicates a strong dependency between these variables. In this case, the combination of these variables can be identified as a key eigenvector. During the construction of the interaction topology, since the power system is a complex network composed of numerous interconnected devices, each device in the power system can be abstracted as a node, and the key eigenvectors can be used as edges to generate the interaction topology, visually demonstrating the inherent connections and interactions between these devices. For example, the key eigenvector {A, B} represents a strong correlation between the generator output frequency A and the transmission line frequency fluctuation amplitude B, while the key eigenvector {B, C} represents a strong correlation between the transmission line frequency fluctuation amplitude B and the load device frequency response C. In the generated interaction topology, the generator, transformer, transmission line, and load device can be used as nodes. An edge can be drawn between the generator and the transmission line to represent the key eigenvector {A, B}, and an edge can be drawn between the transmission line and the load device to represent the key eigenvector {B, C}.

[0032] 210. Calculate the propagation path of each key feature vector in the interaction topology according to the path inference algorithm; 211. Calculate the propagation speed of each propagation path based on the propagation speed influencing factors and the corresponding assigned weights; 212. Determine a propagation path with a propagation speed greater than a preset speed threshold as a candidate sub-path; 213. Calculate the path coverage of the candidate subpath; 214. Determine candidate sub-paths whose path coverage is greater than a preset coverage threshold as candidate fault propagation paths to obtain a set of candidate fault propagation paths.

[0033] Optionally, in this embodiment, a path inference algorithm such as depth-first search, breadth-first search, or Dijkstra's algorithm can be used to calculate the propagation path of each key eigenvector in the interaction topology. For example, if the interaction topology is a network with a clear hierarchical structure, breadth-first search can be preferred. This algorithm explores nodes sequentially in hierarchical order and can quickly find the shortest path from the starting point to other nodes, which is very helpful for preliminarily determining the scope of a fault's potential impact. If the interaction topology is more complex, with multiple paths of varying weights, Dijkstra's algorithm may be more optimal, as it can quickly find the shortest path from the source node to all other nodes in a weighted graph. After obtaining the propagation path, the propagation speed can be calculated based on factors influencing the propagation speed, such as the frequency of interactions between devices, data transmission delay, and processing time. Each edge in the interaction topology is assigned a weight that represents information about the propagation speed of the eigenvector along that edge. The propagation speed is calculated based on the weights assigned to the edges and the combination of edges along the path. For example, if the sum of the weights of all edges on a path is small, the propagation speed on that path is relatively fast. In this case, the propagation speed can be calculated using a specific formula, such as taking the sum of the inverses of the weights of all edges on the path as the calculated propagation speed. Then, by comparing the calculated propagation speed with a preset speed threshold, paths with faster propagation speeds are screened out to obtain candidate sub-paths. In an interactive topology graph, path coverage can be understood as the number of nodes or edges involved in a candidate sub-path, or the system functional modules or business areas covered. Specifically, a graph traversal algorithm can be used to calculate path coverage. Starting from the starting node, the path is explored as deeply as possible along the path until it is no longer possible to continue or the target node is reached. The path then traverses back to the previous node and continues exploring other paths, recording the nodes and edges traversed during the traversal process to determine the path coverage. Finally, the path coverage is compared with the preset coverage threshold to identify the candidate sub-path with the widest path coverage as the candidate fault propagation path.

[0034] 215. Construct an initial state transfer matrix based on the candidate fault propagation path; 216. Use evidence updating mechanism to adjust the conditional probability distribution in the initial state transfer matrix; 217. Calculate the convergence index of the adjusted conditional probability distribution and obtain the convergence constraint result; 218. Filtering a target path set from a set of candidate fault propagation paths based on the convergence constraint result; 219. Construct a path feature matrix based on the target path set; 220. Calculate the fault propagation probability of each target path according to the path characteristic matrix, and determine the cascade fault propagation feature set based on the fault propagation probability; Optionally, in this embodiment, after obtaining a candidate fault propagation path, an initial state transition matrix is ​​constructed based on the candidate fault propagation path. Initial probability values ​​can be assigned based on historical fault data, expert experience, or theoretical models. It is understood that the different operating states of each device are defined as rows and columns of the matrix, and the state transition probability represents the likelihood of transitioning from one state to another, with the sum of the probabilities in each row of the matrix equal to 1. The evidence update mechanism can employ Bayesian theory or other data-driven methods. As the power system continues to operate, the continuous generation of new evidence, such as real-time monitoring data and online test results, can reflect the current actual operating status and fault occurrence trends of each device. Leveraging this new evidence, the probabilities in the initial state transition matrix are updated using the Bayesian formula, making the probability distribution more consistent with actual conditions. For example, in a power system, real-time monitored data such as line load, equipment temperature, and voltage fluctuations can serve as evidence for updating probabilities. When the load on a line continuously exceeds the rated value, historical experience indicates that the probability of failure on that line increases, necessitating adjustments to the state transition probabilities associated with that line fault.

[0035] It should be noted that convergence metrics can include trace distance and KL divergence. Trace distance measures matrix differences by calculating the square root of the sum of the squared differences between two matrix elements, while KL divergence measures the difference between two probability distributions. The conditional probability distribution is updated through multiple iterations, and a convergence metric is calculated after each iteration. When the convergence metric is less than a pre-set threshold, the conditional probability distribution is considered converged. After the conditional probability distribution converges, the probabilities corresponding to candidate fault propagation paths are evaluated to screen a set of target paths. A path feature matrix is ​​then constructed based on parameters such as the length of the target path, the importance of nodes on the target path, the type of equipment on the target path, and the time delay of fault propagation. Finally, algorithms such as support vector machines, random forests, or numerical grey prediction models are used to establish a relationship between fault propagation probability and path characteristics to calculate the fault propagation probability of each target path. Finally, based on factors such as the fault propagation probability and path characteristics, the cascading fault propagation characteristics are analyzed. For example, high-probability paths share the common characteristic that the cascading fault originates from critical equipment in the power system and has a short fault propagation time. Therefore, a set of cascading fault propagation characteristics can be determined based on this information.

[0036] 221. Extracting a cascading fault propagation feature vector from the cascading fault propagation feature set; 222. The cascading fault propagation feature vector is used as the input of the support vector machine to construct the high-dimensional space classification boundary; 223. Adjust the interval of high-dimensional space classification boundaries through kernel function; 224. Calculate the distance metric of the classification boundary in the high-dimensional space after adjusting the interval; 225. Determine whether the distance metric is less than a preset distance threshold. If so, execute step 226. 226. Determine that the failure mode is a cascading failure.

[0037] Optionally, in this embodiment, algorithms such as the Little Porter transform or the Fourier transform can be used to extract cascading fault propagation feature vectors from the cascading fault propagation feature set. The cascading fault propagation feature vectors may include node state change rates, time intervals, and parameter change rates. Because cascading fault propagation feature vectors may have complex nonlinear relationships, it is difficult to find an effective classification boundary directly in the original low-dimensional space. However, mapping them to a high-dimensional space using a support vector machine can increase the linear separability of the data. Therefore, each cascading fault propagation feature vector can be used as input to a support vector machine, which constructs a classification boundary in the high-dimensional space based on this input data. This classification boundary is determined based on the inherent laws of the cascading fault propagation feature vectors and the characteristics of the data distribution.

[0038] After determining the classification boundaries in the high-dimensional space, the kernel parameters need to be optimized to adjust the spacing of the classification boundaries. If the spacing is too large, some data points that should have been correctly classified may be misclassified into other categories, as overly loose classification boundaries cannot accurately capture the subtle differences between different fault modes. If the spacing is too small, the classification model becomes overly complex and prone to overfitting. Specifically, kernel parameters can be optimized based on the characteristics of cascading fault propagation. By continuously adjusting the kernel parameters, the spacing of the classification boundaries can be adapted to the data distribution characteristics, thereby better distinguishing the characteristic data of different fault modes and further improving classification accuracy. By calculating the distance from each cascading fault propagation feature vector to the classification boundary, a distance metric can be calculated, specifically using methods such as Euclidean distance or Manhattan distance. This distance metric reflects the degree of fit between the current fault characteristics and the established cascading fault classification boundaries, providing an important basis for subsequent fault mode determination. A smaller distance metric value indicates a greater similarity between the fault condition represented by the cascading fault propagation feature vector and the cascading fault mode; a larger distance metric value indicates a greater difference between the two. When the distance metric is greater than a preset distance threshold, it indicates that the failure mode of the cascading failure propagation feature vector is a cascading failure mode.

[0039] See also Figure 3 As shown, an embodiment of the fault diagnosis system of the electric power centralized communication control device in the present application includes: An acquisition unit 301 is configured to acquire operating data of a centralized power communication control device and extract a frequency feature set from the operating data; The first calculation unit 302 is used to calculate the correlation strength between the variables in the frequency feature set using the conditional probability table, and to select a set of key feature vectors to generate an interaction topology between different devices in the power system; The second calculation unit 303 is configured to calculate the propagation path and propagation speed of each key feature vector in the interaction topology according to the path inference algorithm, and screen out a set of candidate fault propagation paths based on the propagation path and propagation speed; A determination unit 304 is configured to calculate the fault propagation probability of the candidate fault propagation path to determine a cascading fault propagation feature set; The classification unit 305 is used to construct a high-dimensional space classification boundary according to the cascade fault propagation feature set to perform fault mode classification and obtain a fault diagnosis result.

[0040] In this embodiment, the acquisition unit 301 acquires the operating data of the power centralized communication control device and extracts a frequency feature set from the operating data. The first calculation unit 302 uses a conditional probability table to calculate the correlation strength between the variables in the frequency feature set and screens out a set of key feature vectors to generate an interaction topology between different devices in the power system. The second calculation unit 303 calculates the propagation path and propagation speed of each key feature vector in the interaction topology based on the path inference algorithm and screens out a set of candidate fault propagation paths based on the propagation path and propagation speed. The determination unit 304 calculates the fault propagation probability of the candidate fault propagation path to determine a set of cascading fault propagation features. The classification unit 305 constructs a high-dimensional space classification boundary based on the set of cascading fault propagation features to classify the fault mode and obtain a fault diagnosis result. In this way, the correlation strength between the corresponding variables of each device in the power system can be used to screen out key feature vectors with correlation, and the propagation path and propagation speed of the key feature vectors can be obtained to determine a set of cascading fault path features where cascading faults may occur. Finally, the specific fault mode can be determined based on the high-dimensional space boundary. Therefore, the complex interaction relationship between devices in the power system and the fault propagation characteristics can be comprehensively considered to accurately identify cascading faults and improve the safety and stability of the power system.

[0041] Another embodiment of the fault diagnosis system of the electric power centralized communication control device in the present application includes: The acquisition unit 301 is specifically configured to acquire operating data of the power centralized communication control device; slice the operating data to obtain a slice data set; extract a dynamic data set from the slice data set; extract frequency components and periodic features from the dynamic data set using a small Bode transform; and generate a frequency feature set based on the frequency components and periodic features. The first calculation unit 302 is specifically configured to calculate the conditional probability of state associations between all variables in the frequency feature set and construct a complete conditional probability table; determine the strength of association between the variables in the frequency feature set based on the conditional probability table; determine the variable combination with a correlation strength greater than a preset correlation threshold as a key feature vector; and generate an interaction topology between different devices in the power system using each device in the power system as a node and the key feature vector as an edge; The second calculation unit 303 is specifically configured to calculate the propagation path of each key feature vector in the interaction topology according to the path inference algorithm; calculate the propagation speed of each propagation path based on the propagation speed influencing factors and the corresponding assigned weights; determine the propagation path with a propagation speed greater than a preset speed threshold as a candidate subpath; calculate the path coverage of the candidate subpath; determine the candidate subpath with a path coverage greater than a preset coverage threshold as a candidate fault propagation path, so as to obtain a set of candidate fault propagation paths; Determination unit 304 is specifically configured to construct an initial state transfer matrix based on the candidate fault propagation paths; adjust the conditional probability distribution in the initial state transfer matrix using an evidence update mechanism; calculate a convergence index of the adjusted conditional probability distribution to obtain a convergence constraint result; screen a target path set from the candidate fault propagation path set based on the convergence constraint result; construct a path feature matrix based on the target path set; calculate the fault propagation probability of each target path based on the path feature matrix, and determine a cascade fault propagation feature set based on the fault propagation probability; The classification unit 305 is specifically used to extract a cascading fault propagation feature vector from a cascading fault propagation feature set; use the cascading fault propagation feature vector as input to a support vector machine to construct a high-dimensional space classification boundary; adjust the interval of the high-dimensional space classification boundary through a kernel function; calculate the distance metric of the high-dimensional space classification boundary after the interval adjustment; determine whether the distance metric is greater than a preset distance threshold; if so, determine that the fault mode is a cascading fault.

[0042] In this embodiment, the functions of each unit are the same as those described above. Figure 2-1 、 Figure 2-2 as well as Figure 2-3 The functions of steps 201 to 226 in the illustrated embodiment are similar and will not be described in detail here.

[0043] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0044] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0045] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0046] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0047] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, read-only memory), random access memory (RAM, random access memory), disk or optical disk, and other media that can store program code.

Claims

1. A fault diagnosis method for a power centralized communication control device, characterized in that: include: Acquiring operating data of a centralized power communication control device and extracting a frequency feature set from the operating data; Calculating the correlation strength between the variables in the frequency feature set using a conditional probability table, and screening out a set of key feature vectors to generate an interaction topology between different devices in the power system; Calculating the propagation path and propagation speed of each key feature vector in the interaction topology according to a path inference algorithm, and screening a set of candidate fault propagation paths based on the propagation path and propagation speed; Calculating the fault propagation probability of the candidate fault propagation path to determine the cascade fault propagation feature set; A high-dimensional space classification boundary is constructed according to the cascade fault propagation feature set to perform fault mode classification and obtain a fault diagnosis result.

2. The fault diagnosis method for the centralized power communication control device according to claim 1, characterized in that: The constructing of a high-dimensional space classification boundary based on the cascade fault propagation feature set to perform fault mode classification and obtain a fault diagnosis result includes: Extracting a cascading fault propagation feature vector from the cascading fault propagation feature set; Using the cascading fault propagation feature vector as an input to a support vector machine to construct a high-dimensional space classification boundary; Adjusting the interval of the high-dimensional space classification boundary by a kernel function; Calculating a distance metric of the classification boundary of the high-dimensional space after adjusting the interval; Determining whether the distance metric is less than a preset distance threshold; If yes, the failure mode is determined to be a cascading failure.

3. The fault diagnosis method for the electric power centralized communication control device according to claim 1, characterized in that: Calculating the propagation path and propagation speed of each key feature vector in the interaction topology according to the path inference algorithm, and screening a set of candidate fault propagation paths based on the propagation path and propagation speed, includes: Calculate the propagation path of each key feature vector in the interaction topology according to the path inference algorithm; Calculate the propagation speed of each propagation path based on the factors affecting the propagation speed and the corresponding assigned weights; Determine a propagation path with a propagation speed greater than a preset speed threshold as a candidate sub-path; Calculating the path coverage of the candidate subpath; The candidate sub-paths whose path coverage range is greater than a preset coverage threshold are determined as candidate fault propagation paths to obtain a set of candidate fault propagation paths.

4. The fault diagnosis method for the electric power centralized communication control device according to claim 1, characterized in that: Calculating the fault propagation probability of the candidate fault propagation path to determine the cascading fault propagation feature set includes: Construct an initial state transfer matrix based on candidate fault propagation paths; Adopting an evidence updating mechanism to adjust the conditional probability distribution in the initial state transfer matrix; Calculating the convergence index of the adjusted conditional probability distribution to obtain a convergence constraint result; Filtering a target path set from the candidate fault propagation path set based on the convergence constraint result; Constructing a path feature matrix based on the target path set; The fault propagation probability of each target path is calculated according to the path characteristic matrix, and a cascade fault propagation feature set is determined based on the fault propagation probability.

5. The fault diagnosis method for the electric power centralized communication control device according to claim 1, characterized in that: The method of calculating the correlation strength between the variables in the frequency feature set using the conditional probability table and screening out a set of key feature vectors to generate an interaction topology between different devices in the power system includes: Calculate the conditional probability of state associations between all variables in the frequency feature set and construct a complete conditional probability table; Determining the association strength between the variables in the frequency feature set according to the conditional probability table; The variable combination with the correlation strength greater than the preset correlation threshold is determined as the key feature vector; Taking each device in the power system as a node and the key eigenvectors as edges, the interaction topology between different devices in the power system is generated.

6. The fault diagnosis method for a centralized power communication control device according to any one of claims 1 to 5, characterized in that: The obtaining of operating data of the centralized power communication control device and extracting a frequency feature set from the operating data includes: Obtaining operating data of the power centralized communication control device; Slice the operating data to obtain a slicing data set; Extracting a dynamic data set from the fragmented data set; Extracting frequency components and periodic features from the dynamic data set using a small Bode transform; A frequency feature set is generated based on the frequency components and the periodic features.

7. A fault diagnosis system for a centralized power communication control device, characterized in that: include: an acquisition unit, configured to acquire operating data of the electric power centralized communication control device and extract a frequency feature set from the operating data; a first calculation unit, configured to calculate the correlation strength between the variables in the frequency feature set using a conditional probability table, and screen out a set of key feature vectors to generate an interaction topology between different devices in the power system; a second calculation unit, configured to calculate a propagation path and a propagation speed of each key feature vector in the interaction topology according to a path inference algorithm, and screen out a set of candidate fault propagation paths based on the propagation path and the propagation speed; a determination unit, configured to calculate a fault propagation probability of a candidate fault propagation path to determine a cascade fault propagation feature set; The classification unit is used to construct a high-dimensional space classification boundary according to the cascade fault propagation feature set to perform fault mode classification and obtain a fault diagnosis result.

8. The fault diagnosis system for the centralized power communication control device according to claim 7, characterized in that: The classification unit is specifically used for: Extracting a cascading fault propagation feature vector from the cascading fault propagation feature set; Using the cascading fault propagation feature vector as an input to a support vector machine to construct a high-dimensional space classification boundary; Adjusting the interval of the high-dimensional space classification boundary by a kernel function; Calculating a distance metric of the classification boundary of the high-dimensional space after adjusting the interval; Determining whether the distance metric is less than a preset distance threshold; If yes, the failure mode is determined to be a cascading failure.

9. The fault diagnosis system for the centralized power communication control device according to claim 7, characterized in that: The second computing unit is specifically configured to: Calculate the propagation path of each key feature vector in the interaction topology according to the path inference algorithm; Calculate the propagation speed of each propagation path based on the factors affecting the propagation speed and the corresponding assigned weights; Determine a propagation path with a propagation speed greater than a preset speed threshold as a candidate sub-path; Calculating the path coverage of the candidate subpath; The candidate sub-paths whose path coverage range is greater than a preset coverage threshold are determined as candidate fault propagation paths to obtain a set of candidate fault propagation paths.

10. The fault diagnosis system for the centralized power communication control device according to claim 7, characterized in that: The determining unit is specifically configured to: Construct an initial state transfer matrix based on candidate fault propagation paths; Adopting an evidence updating mechanism to adjust the conditional probability distribution in the initial state transfer matrix; Calculating the convergence index of the adjusted conditional probability distribution to obtain a convergence constraint result; Filtering a target path set from the candidate fault propagation path set based on the convergence constraint result; Constructing a path feature matrix based on the target path set; The fault propagation probability of each target path is calculated according to the path characteristic matrix, and a cascade fault propagation feature set is determined based on the fault propagation probability.

Images