An intelligent detection system and detection method for high and low voltage distribution cabinets

Abstract

The present invention relates to the technical field of distribution cabinet detection, and discloses an intelligent detection system and detection method for high and low voltage distribution cabinets. The system includes a data acquisition module, a feature extraction module, a network construction and update module, a fault location module, a positioning map generation module and a self-healing control module; a multi-physical field coupling data set is generated by synchronously collecting electromagnetic radiation spectrum and cabinet vibration waveform, and a time-frequency domain joint analysis is performed to extract fault feature vectors, a fault propagation network is constructed and dynamically updated, and the key fault source is identified by combining betweenness centrality and historical case matching, and a fault location map is generated; based on the map, a reverse electromagnetic pulse is applied and the contact pressure is adjusted to achieve precise self-healing control; the present invention solves the problems of low positioning accuracy and mechanical self-healing control means when high and low voltage distribution cabinets are detected, and significantly improves the fault detection efficiency and reliability.

Description

Technical Field

[0001] The present invention relates to the technical field of power distribution cabinet detection, and in particular to an intelligent detection system and a detection method for a high and low voltage power distribution cabinet. Background Art

[0002] Current fault detection in high and low voltage distribution cabinets relies heavily on static analysis of a single physical field (such as current or temperature), which fails to effectively capture the coupled characteristics of electromagnetic and mechanical vibrations, resulting in incomplete fault feature extraction. Traditional methods use fixed topology models to analyze fault propagation, lacking dynamic update mechanisms and adapting poorly to complex fault scenarios. Furthermore, self-healing control strategies often involve global power outages or mechanical lockouts, which carry a high risk of malfunction and fail to accurately locate the source of the fault and potential areas of spread.

[0003] In the existing technology, CN117553859A realizes data collection through multiple sensors, but does not solve the problems of multi-physical field data fusion and dynamic network modeling; CN105762775A proposes dynamic topology optimization, but does not combine the fault propagation energy characteristics with historical case matching, resulting in large positioning deviations. At the same time, the power grid self-healing system lacks a physical space mapping mechanism for distribution cabinets. Summary of the Invention

[0004] The present invention provides an intelligent detection system and detection method for high and low voltage distribution cabinets, the main purpose of which is to solve the problems of low positioning accuracy, insufficient fault propagation path tracing capability and mechanical self-healing control means in fault detection of high and low voltage distribution cabinets.

[0005] To achieve the above object, the present invention provides an intelligent detection system for high and low voltage distribution cabinets, characterized in that the system includes:

[0006] The data acquisition module is used to synchronously collect the electromagnetic radiation spectrum and cabinet surface vibration waveform of the high and low voltage distribution cabinets to generate a multi-physics field coupling data set;

[0007] A feature extraction module is used to perform a time-frequency domain joint analysis on the multi-physics field coupling data set to obtain a fault feature vector of the high and low voltage distribution cabinet;

[0008] A network construction and update module, configured to construct and dynamically update a fault propagation network based on the fault feature vector;

[0009] A fault location module is used to match the betweenness centrality and energy distribution of nodes in the fault propagation network with the fault patterns in the historical fault case library, identify the key fault sources of the high and low voltage distribution cabinets, and mark potential fault diffusion areas;

[0010] A positioning map generation module is used to map the key fault source and the potential fault diffusion area to the physical space coordinates of the high and low voltage distribution cabinet to obtain a fault location map of the high and low voltage distribution cabinet;

[0011] A self-healing control module is used to output self-healing control instructions for the high and low voltage distribution cabinets based on the fault location map, wherein the self-healing control instructions include applying a reverse electromagnetic pulse to the key fault source and adjusting the contact pressure of the potential fault diffusion area.

[0012] In order to solve the above problems, the present invention also provides an intelligent detection method for high and low voltage distribution cabinets, the method comprising:

[0013] S1. Synchronously collect the electromagnetic radiation spectrum and cabinet surface vibration waveform of the high and low voltage distribution cabinets to generate a multi-physics field coupling data set;

[0014] S2. Performing a time-frequency domain joint analysis on the multi-physics field coupling data set to obtain a fault feature vector of the high and low voltage distribution cabinet;

[0015] S3. Constructing and dynamically updating a fault propagation network based on the fault feature vector;

[0016] S4. Based on the betweenness centrality and energy distribution of nodes in the fault propagation network, match the fault patterns in the historical fault case library to identify key fault sources and mark potential fault diffusion areas;

[0017] S5. Mapping the key fault source and the potential fault diffusion area to the physical space coordinates of the high and low voltage distribution cabinet to obtain a fault location map of the high and low voltage distribution cabinet;

[0018] S6. Outputting self-healing control instructions for the high and low voltage distribution cabinet based on the fault location map, wherein the self-healing control instructions include applying a reverse electromagnetic pulse to the key fault source and adjusting the contact pressure of the potential fault diffusion area.

[0019] The present invention forms a multi-physical field coupling data set through the synchronous collection and alignment of electromagnetic and vibration data, and extracts multi-dimensional fault features through joint analysis in the time-frequency domain, which solves the problem of incomplete extraction of single physical field features and significantly improves the accuracy of fault detection; the fault features are mapped to network nodes and edges, and the topological structure is dynamically updated based on the energy transfer probability model to reflect the changes in the fault propagation path in real time, overcoming the defect that traditional static models cannot track dynamic propagation and improving the analysis capability of complex fault scenarios; key fault sources are screened through betweenness centrality, and the root cause of the fault is quickly locked by combining historical case matching and energy distribution analysis, and reverse electromagnetic pulses are applied based on the fault location map to suppress electromagnetic interference, and the contact pressure is dynamically adjusted to block the spread of mechanical faults, thus achieving targeted intervention; the network nodes are mapped to the physical coordinates of the distribution cabinet to generate a fault location map, providing intuitive fault distribution information for operation and maintenance personnel, shortening the fault handling response time and improving on-site operation efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 This is a system architecture diagram of an intelligent detection system for high and low voltage distribution cabinets provided by one embodiment of the present invention.

[0021] Figure 2 A schematic flow chart of an intelligent detection method for high and low voltage distribution cabinets provided in one embodiment of the present invention.

[0022] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0023] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0024] The terms used in the embodiments of the present invention are for the purpose of describing specific embodiments only and are not intended to limit the present invention. The singular forms "a," "an," "the," and "the" used in the embodiments of the present invention and the appended claims are also intended to include plural forms, and unless the context clearly indicates otherwise, "a plurality" generally includes at least two.

[0025] As used herein, the words "if" and "if" may be interpreted as "at the time of" or "when" or "in response to determining" or "in response to detecting," depending on the context. Similarly, the phrases "if it is determined" or "if (stated condition or event) is detected" may be interpreted as "when it is determined" or "in response to the determination" or "when detecting (stated condition or event)" or "in response to detecting (stated condition or event)," depending on the context.

[0026] In addition, the step sequence in the following method embodiments is only an example and not a strict limitation.

[0027] In practice, the server-side device deployed by the intelligent detection system for high and low voltage distribution cabinets may be composed of one or more devices. The intelligent detection system for high and low voltage distribution cabinets can be implemented as: a business instance, a virtual machine, and a hardware device. For example, the intelligent detection system for high and low voltage distribution cabinets can be implemented as a business instance deployed on one or more devices in a cloud node. Simply put, the intelligent detection system for high and low voltage distribution cabinets can be understood as a software deployed on a cloud node, used to provide the intelligent detection system for high and low voltage distribution cabinets to each user end. Alternatively, the intelligent detection system for high and low voltage distribution cabinets can also be implemented as a virtual machine deployed on one or more devices in a cloud node. The virtual machine is installed with application software for managing each user end. Alternatively, the intelligent detection system for high and low voltage distribution cabinets can also be implemented as a server-side composed of many hardware devices of the same or different types, with one or more hardware devices being provided to provide the intelligent detection system for high and low voltage distribution cabinets to each user end.

[0028] In terms of implementation, the intelligent detection system for high and low voltage distribution cabinets and the user end are mutually compatible. Specifically, if the intelligent detection system for high and low voltage distribution cabinets is an application installed on a cloud service platform, the user end serves as a client that establishes a communication connection with the application. Alternatively, if the intelligent detection system for high and low voltage distribution cabinets is implemented as a website, the user end is implemented as a webpage. Alternatively, if the intelligent detection system for high and low voltage distribution cabinets is implemented as a cloud service platform, the user end is implemented as a mini-program within an instant messaging application.

[0029] like Figure 1 1 is a system architecture diagram of an intelligent detection system for high and low voltage distribution cabinets provided by an embodiment of the present invention.

[0030] The intelligent detection system 100 for high and low voltage distribution cabinets of the present invention can be set in a cloud server. In terms of implementation, it can be used as one or more service devices, or it can be installed as an application on the cloud (such as a server of a mobile service operator, a server cluster, etc.), or it can be developed as a website. According to the functions implemented, the intelligent detection system 100 for high and low voltage distribution cabinets can include a data acquisition module 101, a feature extraction module 102, a network construction and update module 103, a fault location module 104, a location map generation module 105 and a self-healing control module 106. The module of the present invention can also be referred to as a unit, which refers to a series of computer program segments that can be executed by an electronic device processor and can complete fixed functions, which are stored in the memory of the electronic device.

[0031] In an embodiment of the present invention, in the intelligent detection system of the high and low voltage distribution cabinet, each of the above modules can be implemented independently and called with other modules. The call here can be understood as that a certain module can connect to multiple modules of another type and provide corresponding services to the multiple modules connected to it. For example, the sharing evaluation module can call the same information acquisition module to obtain the information collected by the information acquisition module. Based on the above characteristics, in the intelligent detection system of the high and low voltage distribution cabinet provided by the embodiment of the present invention, the scope of application of the intelligent detection system architecture of the high and low voltage distribution cabinet can be adjusted by adding modules and directly calling them without modifying the program code, so as to achieve cluster-type horizontal expansion, so as to achieve the purpose of quickly and flexibly expanding the intelligent detection system of the high and low voltage distribution cabinet. In actual applications, the above modules can be set in the same device or different devices, or they can be set in virtual devices, such as service instances in cloud servers.

[0032] The following describes the various components and specific workflows of the intelligent detection system for high and low voltage distribution cabinets in conjunction with specific embodiments:

[0033] The data acquisition module 101 is used to synchronously acquire the electromagnetic radiation spectrum and cabinet surface vibration waveform of the high and low voltage distribution cabinets to generate a multi-physics field coupling data set.

[0034] The data acquisition module of the present invention realizes the synchronous acquisition and fusion of multi-physical field data by collaboratively deploying electromagnetic sensors and vibration sensors and establishing a precise time synchronization mechanism.

[0035] When the data acquisition module performs synchronous acquisition of the electromagnetic radiation spectrum and cabinet surface vibration waveform of the high and low voltage distribution cabinets to generate a multi-physics field coupling data set, it is specifically used to:

[0036] Arrange electromagnetic sensors in the electromagnetic radiation hotspots of high and low voltage distribution cabinets, install vibration sensors at vibration-sensitive points on the cabinet surfaces of the high and low voltage distribution cabinets, and perform time alignment on the electromagnetic sensors and the vibration sensors;

[0037] Using the electromagnetic sensor to collect the electromagnetic radiation spectrum of the high and low voltage distribution cabinet;

[0038] Using the vibration sensor to collect the cabinet surface vibration waveform of the high and low voltage distribution cabinet;

[0039] The electromagnetic radiation spectrum and the cabinet surface vibration waveform are aligned according to a time window to generate a multi-physics field coupling data set of the high and low voltage distribution cabinet.

[0040] Specifically, an electromagnetic sensor is a device used to detect electromagnetic radiation signals generated during the operation of high- and low-voltage distribution cabinets. In this technical solution, a high-frequency current probe is used as the electromagnetic sensor, which can effectively detect changes in high-frequency electromagnetic radiation caused by faults such as partial discharge and arcing.

[0041] Specifically, vibration sensors are primarily used to measure mechanical vibration waveforms on the surfaces of high- and low-voltage distribution cabinets. This solution uses MEMS accelerometers as vibration sensors. These sensors offer high sensitivity and a compact size, accurately capturing vibration signals generated by loose mechanical components and resonance.

[0042] Specifically, time alignment involves calibrating the time bases of electromagnetic and vibration sensor data, ensuring consistency in the temporal dimension of the data collected by the two types of sensors for subsequent multi-physics coupling analysis. This solution uses the IEEE 1588 Precision Time Protocol (PTP) for time alignment, which maintains extremely tight synchronization accuracy.

[0043] Specifically, a multi-physics coupled dataset is a time-domain integration of electromagnetic radiation spectrum data and vibration waveform data from high- and low-voltage distribution cabinets. This dataset, encompassing both electromagnetic and vibration fields, comprehensively reflects the physical state changes of the distribution cabinet during operation.

[0044] Specifically, based on the physical characteristics of the high- and low-voltage distribution cabinets, electromagnetic sensors are placed in hotspots of electromagnetic radiation within the cabinets. For example, the circuit breaker contacts are prone to faults like partial discharge and are also areas of concentrated electromagnetic radiation. Installing high-frequency current probes here can prioritize capturing abnormal electromagnetic signals generated by faults. Meanwhile, vibration sensors are installed at vibration-sensitive points on the cabinet surface, such as the disconnector operating mechanism. These areas produce significant vibration changes when mechanical components operate or fail. MEMS accelerometers can promptly sense and record these vibration signals.

[0045] Specifically, the IEEE 1588 Precision Time Protocol (PTP) was used to synchronize the time of the deployed electromagnetic and vibration sensors. This protocol enables precise exchange and calibration of time information between the sensors, minimizing the time error between the collected data and ensuring high temporal consistency in subsequent electromagnetic and vibration data.

[0046] Specifically, high-frequency current probes installed in electromagnetic radiation hotspots begin operating, continuously monitoring electromagnetic radiation signals generated during the operation of high- and low-voltage distribution cabinets. These probes sense and convert electromagnetic signals within specific frequency bands, converting them into electrical signals that can be recognized and processed by the system. These signals are conditioned and digitized to generate spectrum data reflecting the distribution cabinet's electromagnetic radiation status.

[0047] Specifically, MEMS accelerometers located at vibration-sensitive points on the cabinet's surface sense the cabinet's vibration state in real time and convert mechanical vibrations into electrical signals. Similar to electromagnetic signal acquisition, the vibration signals are processed by signal conditioning circuits, including amplification and filtering, before being converted into digital signals by an analog-to-digital conversion module. This ultimately generates waveform data representing the cabinet's surface vibration characteristics, including amplitude, frequency, and phase information.

[0048] Specifically, after completing the acquisition of the electromagnetic radiation spectrum and the cabinet surface vibration waveform, the system performs time alignment processing on the two types of data according to a pre-set time window. The setting of the time window needs to comprehensively consider data processing efficiency and the ability to capture fault characteristics. For example, it can be set to 100 milliseconds. The system integrates the electromagnetic radiation spectrum data and the vibration waveform data within the same time window to form a complete data recording unit. Over time, multiple such data recording units are superimposed in sequence, eventually generating a multi-physics field coupling data set covering different moments and fully reflecting the changes in the electromagnetic and vibration physical fields during the operation of the distribution cabinet.

[0049] In general, only by properly placing electromagnetic sensors and vibration sensors in key locations and achieving precise time synchronization can the collected electromagnetic radiation spectrum and cabinet surface vibration waveform data be ensured to have practical analytical value. Electromagnetic radiation spectrum acquisition and cabinet surface vibration waveform acquisition are parallel data acquisition processes. Both are performed synchronously based on the already arranged sensors, acquiring raw data from the electromagnetic and vibration physical field dimensions respectively. Finally, the multi-physics field coupling dataset generation step integrates the data acquired in the first two acquisition steps and aligns them according to the time window to form a complete dataset containing multi-physics field information, providing unified data input for the subsequent fault analysis process. Each step is closely linked to form a complete data acquisition system.

[0050] The feature extraction module 102 is used to perform a time-frequency domain joint analysis on the multi-physics field coupling data set to obtain a fault feature vector of the high and low voltage distribution cabinet.

[0051] During the operation of high- and low-voltage distribution cabinets, internal faults often exhibit the complex characteristics of multi-physics coupling between electromagnetic and mechanical vibration. Traditional fault detection methods based on single physical quantities struggle to effectively identify such coupled faults, resulting in a high rate of false positives. This feature extraction module performs a joint time-frequency domain analysis of multi-physics coupled datasets to deeply mine fault characteristics from electromagnetic and vibration signals. This module then generates fault feature vectors that accurately characterize the fault mode, providing critical data support for subsequent fault diagnosis and location.

[0052] When the feature extraction module performs a time-frequency domain joint analysis on the multi-physics field coupling data set to obtain the fault feature vector of the high and low voltage distribution cabinet, it is specifically used to:

[0053] Performing frequency domain analysis on the electromagnetic radiation spectrum in the multi-physics field coupling data set to extract abnormal frequency band features of energy mutation;

[0054] Performing time-domain phase analysis on the cabinet surface vibration waveform in the multi-physics field coupling data set to extract time-series-related phase mutation features;

[0055] The abnormal frequency band feature is fused with the phase mutation feature to obtain a fault feature vector representing the fault mode.

[0056] Specifically, frequency domain analysis is a signal processing technique that converts time-domain signals into the frequency domain for analysis. Frequency domain analysis reveals the energy distribution of a signal across different frequency components and is commonly used to analyze a signal's frequency characteristics and harmonic content. In this solution, frequency domain analysis is primarily used to analyze the electromagnetic radiation spectrum to identify unusual frequency components and energy distribution variations.

[0057] Specifically, time-domain phase analysis analyzes the phase information in time-domain signals. Phase reflects the relative position of a signal on the time axis. By analyzing the time-domain phase, we can capture the transient characteristics of a signal's temporal variation. In this technology, time-domain phase analysis is primarily applied to cabinet surface vibration waveforms to extract information about sudden phase changes associated with faults.

[0058] Specifically, abnormal frequency band characteristics refer to the characteristic information of specific frequency ranges where energy undergoes significant abrupt changes, as identified through frequency-domain analysis of the electromagnetic radiation spectrum. These abnormal frequency bands are often closely associated with electromagnetic faults (such as partial discharge and arcing) within the power distribution cabinet and are crucial for diagnosing electromagnetic faults.

[0059] Specifically, phase mutations are detected by performing time-domain phase analysis on the cabinet surface vibration waveform, detecting sudden phase jumps in the time series. Phase mutations are typically caused by mechanical component failures (such as looseness, wear, and resonance) and are a key characteristic of mechanical failures.

[0060] Furthermore, the process integrates features extracted from different physical field signals. Through feature fusion, the fault information contained in both electromagnetic and vibration signals is organically combined to form a more comprehensive and representative fault feature vector, thereby more accurately describing the fault mode. This fault feature vector is a multidimensional vector composed of abnormal frequency band characteristics and phase mutation characteristics after feature fusion. This vector comprehensively characterizes the possible fault modes of high and low voltage distribution cabinets and serves as the core data foundation for subsequent fault diagnosis, location, and network modeling.

[0061] Specifically, a Fast Fourier Transform (FFT) is used to convert the electromagnetic radiation spectrum in a multiphysics coupled dataset from the time domain to the frequency domain. FFT efficiently decomposes the time domain signal into a superposition of different frequency components, thereby obtaining the amplitude and phase information of the electromagnetic radiation signal at each frequency, forming a complete frequency domain spectrum.

[0062] Furthermore, energy analysis is performed on the converted frequency domain spectrum to calculate the energy distribution within each frequency band. By setting an appropriate energy threshold, frequency bands with significant energy changes are screened out. For example, when the energy within a frequency band shows a significant increase or decrease compared to the energy under normal operating conditions, the frequency band is marked as an abnormal band. Key parameters of the abnormal frequency band, such as center frequency, bandwidth, and energy mutation amplitude, are extracted as abnormal frequency band features. These features can intuitively reflect abnormal changes in electromagnetic radiation signals and provide important clues for diagnosing electromagnetic faults.

[0063] In detail, the original vibration signal is converted into an analytical signal through Hilbert transform, and then the instantaneous phase information of the signal is obtained. The instantaneous phase can accurately describe the phase state of the vibration signal at each moment.

[0064] Specifically, the calculated instantaneous phase is analyzed to detect phase mutation points. By setting an appropriate phase change threshold, a phase mutation is determined when the instantaneous phase change exceeds the threshold. For example, when the instantaneous phase derivative exceeds a certain value (e.g., 0.5 rad / s), a phase mutation is considered to have occurred. Information such as the time of the phase mutation and the phase values ​​before and after the mutation are recorded as phase mutation signatures. These signatures can effectively reflect fault conditions in mechanical components, such as abnormal vibration phase changes caused by loose components.

[0065] Specifically, the extracted abnormal frequency band features and phase mutation features are combined. According to a predefined feature vector format, the various parameters of the abnormal frequency band (such as center frequency, energy mutation amplitude, etc.) and the relevant information of the phase mutation (such as mutation time, phase change, etc.) are arranged in sequence to form a multidimensional vector.

[0066] Furthermore, in order to eliminate the differences in dimensions and numerical ranges between different feature parameters, the combined feature vector is Z-score normalized. Through normalization, each element in the feature vector is within a reasonable numerical range.

[0067] In general, by performing targeted frequency domain analysis and time domain phase analysis on the electromagnetic radiation spectrum and cabinet surface vibration waveform, respectively, we can deeply explore the fault characteristics contained in these two physical field signals. Compared with traditional single signal analysis methods, joint time-frequency domain analysis can more comprehensively capture the changing characteristics of the signal, thereby significantly improving the ability to identify fault characteristics. For example, for faults caused by the coupling of partial discharge and mechanical resonance, frequency domain analysis can not only detect abnormal frequency bands in the electromagnetic radiation, but also detect sudden phase changes in the vibration waveform through time domain phase analysis, effectively avoiding the omission of fault characteristics.

[0068] In general, feature fusion technology organically integrates fault features from different physical fields to form a fault feature vector that contains richer and more comprehensive fault information. This multi-dimensional fault feature description can more accurately reflect the actual fault situation, effectively reduce the fault misdiagnosis rate, and improve the accuracy and reliability of fault diagnosis.

[0069] In general, frequency-domain analysis of the electromagnetic radiation spectrum and time-domain phase analysis of the cabinet surface vibration waveform are two parallel and independent feature extraction processes. They analyze and process the multi-physics coupled dataset from the electromagnetic and vibration dimensions, respectively, extracting fault characteristics within their respective domains. These two steps are the core fundamental operations of the feature extraction module, providing the necessary raw feature data for subsequent feature fusion. The feature fusion step integrates the abnormal frequency band features and phase mutation features extracted from these two steps, combining and normalizing them to generate the final fault feature vector. As the output of the entire feature extraction module, the fault feature vector not only integrates the fault characteristics of the electromagnetic and vibration signals but also provides key input data for subsequent network construction and update modules, fault location modules, and other modules. It serves as a critical bridge connecting data acquisition with subsequent processing steps such as fault diagnosis and location, and plays a key role in connecting the entire high- and low-voltage distribution cabinet intelligent detection system.

[0070] The network construction and updating module 103 is configured to construct and dynamically update a fault propagation network based on the fault feature vector.

[0071] In the field of high and low voltage distribution cabinet fault detection, traditional technologies use static topology models to analyze faults. These models cannot dynamically reflect the energy transfer paths during fault propagation, resulting in large positioning errors and difficulty adapting to complex fault scenarios. This network construction and update module constructs and dynamically updates the fault propagation network based on fault feature vectors. Through innovative node mapping, edge generation, and topology optimization mechanisms, it achieves accurate modeling and real-time tracking of the fault propagation process.

[0072] When the network construction and update module constructs and dynamically updates the fault propagation network based on the fault feature vector, it is specifically used to:

[0073] Mapping the abnormal frequency band features in the fault feature vector into nodes of the fault propagation network, wherein the weight of each node is determined by the electromagnetic radiation energy density corresponding to the abnormal frequency band features;

[0074] generating directed connection edges between the nodes based on the time correlation of the phase mutation features in the fault feature vector, wherein the direction of the directed connection edges is determined by the temporal relationship of the phase mutation features;

[0075] The topology of the fault propagation network is dynamically updated through an energy transfer probability model.

[0076] Specifically, a fault propagation network is a network model that uses a graph structure to represent the fault propagation paths and states within high- and low-voltage distribution cabinets. Nodes in the network represent fault characteristics, while edges represent fault propagation relationships. The dynamic changes in nodes and edges reflect the development of faults within the distribution cabinet.

[0077] Specifically, in the fault propagation network, nodes are mapped from the abnormal frequency band characteristics in the fault feature vector. Each node corresponds to a specific abnormal electromagnetic radiation frequency band. The node weight is determined by the electromagnetic radiation energy density corresponding to the frequency band, which is used to quantify the importance of the frequency band in the fault signature.

[0078] Specifically, directed edges are directed line segments connecting nodes in the fault propagation network. Their directions are determined by the temporal order of the phase mutation features in the fault feature vector and represent the direction of fault propagation. The edge weights reflect the probability of the fault propagating through this path and are dynamically adjusted using an energy transfer probability model.

[0079] Specifically, the energy transfer probability model is a mathematical model based on probability theory that describes the probability of fault energy being transferred between nodes in a fault propagation network. By analyzing inter-node energy relationships and historical fault data, the model calculates and updates the inter-node energy transfer probability, providing a basis for dynamic updates to the network topology.

[0080] Specifically, topology optimization is the process of adjusting and optimizing the node connectivity and node attributes of a fault propagation network. By merging redundant nodes and adding key edges, the network structure is simplified, computational efficiency is improved, and the network more accurately reflects fault propagation patterns.

[0081] Furthermore, node mapping refers to using the abnormal frequency band characteristics in the fault feature vector as the basis for constructing network nodes. For example, if the fault feature vector contains information about energy mutations in the 30-50kHz frequency band, a corresponding node is created in the fault propagation network. The weight of each node is determined by the electromagnetic radiation energy density corresponding to the abnormal frequency band. The higher the energy density, the greater the node weight. If a frequency band has a high energy density, the weight of its mapping node can be set to a high value (such as 0.8), indicating that this frequency band occupies an important position in the fault characteristics and may be a key indicator of the fault.

[0082] Furthermore, directed edge generation refers to determining the connection relationship between nodes based on the temporal correlation of the phase mutation characteristics in the fault feature vector. When a phase mutation is detected in the vibration signals associated with the abnormal frequency band characteristics corresponding to two nodes, and the phase mutations have a sequential order, a directed edge is generated from the node corresponding to the first phase mutation to the node corresponding to the later phase mutation. For example, if the vibration signal associated with node A undergoes a phase mutation at time t1, and the vibration signal associated with node B undergoes a phase mutation at time t2 (t2>t1), a directed edge is generated from node A to node B, clarifying the possible propagation path of the fault.

[0083] The network construction and update module dynamically updates the topology of the fault propagation network using an energy transfer probability model, specifically for:

[0084] When it is detected that the difference in electromagnetic radiation energy density between nodes in the fault propagation network exceeds a preset difference threshold, an edge weight decay operation is triggered, and the edge weights corresponding to the low-probability propagation paths in the directed connection edges are decayed based on the energy transfer probability model;

[0085] When it is detected that the graph structure similarity between the newly added nodes in the fault propagation network and the historical topology structure is lower than a preset similarity threshold, a topology optimization instruction is triggered, the topology structure is dynamically optimized according to the optimization instruction, and the model parameters of the energy transfer probability model are updated;

[0086] The updated edge weights, optimized topology, and updated energy transfer probability model are synchronized to the fault propagation network.

[0087] Specifically, edge weight decay refers to real-time monitoring of electromagnetic radiation energy density differences between nodes in a fault propagation network. When the energy density difference between nodes exceeds a preset difference threshold, it indicates that the probability of fault propagation on some paths is low, triggering edge weight decay. Based on the energy transfer probability model, the edge weights corresponding to low-probability propagation paths in directed connected edges are decayed. For example, if the energy density difference between node X and node Y exceeds the threshold, the weight of the edge connecting them is reduced from the initial value of 0.6 to 0.2, weakening the role of this low-probability propagation path in the network and making the network more focused on high-probability fault propagation paths.

[0088] In detail, the triggering conditions for edge weight decay are as follows:

[0089] ;

[0090] in, is the judgment value of the edge weight decay trigger condition, yes The electromagnetic radiation energy density of the node, yes The electromagnetic radiation energy density of the node, is the larger value of the two node energy densities, is the preset difference threshold (the value can be 0.3-0.5).

[0091] Specifically, the edge weight decay triggering condition determines whether the connection edges between nodes in the fault propagation network need to be weighted. and The difference in electromagnetic radiation energy density exceeds the preset threshold hour, Trigger decay; otherwise Not triggered.

[0092] Furthermore, the value of the preset difference threshold is determined based on the statistical difference in energy density of 95% of high-probability propagation paths in the historical fault case library, which is used to distinguish between "high-probability propagation paths" and "low-probability propagation paths"; the electromagnetic radiation energy density can be obtained by actual measurement of electromagnetic sensors; when the energy density difference is large, attenuation is triggered.

[0093] In detail, Used to normalize the differences to avoid judgment bias caused by differences in the absolute value of energy density (for example, small differences between high-energy-density nodes may be more practical than large differences between low-energy-density nodes); trigger condition judgment value (0 or 1) is used to control whether to perform edge weight decay operations.

[0094] Furthermore, edge weight decay involves local updates of the fault propagation network, while topology optimization and parameter updates involve global updates of the fault propagation network.

[0095] In detail, update result synchronization refers to synchronizing the edge weight values ​​after edge weight attenuation, the network structure after topology optimization, and the updated energy transfer probability model parameters to the fault propagation network, so that the network can reflect the latest status of fault propagation in real time and provide an accurate analysis model for subsequent fault location.

[0096] In general, by converting fault feature vectors into network nodes and edges and dynamically updating them using the energy transfer probability model, dynamic modeling of the fault propagation process is achieved, effectively solving the problem of difficulty in tracing the fault propagation path and providing a reliable basis for quickly locating the fault source.

[0097] In summary, edge weight decay and topology optimization mechanisms make the fault propagation network adaptive. When energy distribution changes or new fault characteristics occur within the network, the network automatically adjusts its structure and parameters through local updates and global optimization. This removes redundant information, strengthens key propagation paths, avoids network structural redundancy, and improves model computational efficiency, ensuring the network always operates in an optimal state and adapting to the analysis needs of different fault scenarios.

[0098] In summary, the network, combining electromagnetic radiation energy density with the characteristics of sudden vibration phase changes, leverages the advantages of multi-physics field coupling data. By reflecting electromagnetic energy characteristics through node weights and reflecting the temporal correlation of vibration phases through directed edges, this method achieves a deep fusion of multi-physics field fault characteristics, effectively capturing the complex relationships between fault propagation. Compared to single-physics analysis methods, this method significantly improves the analysis capability of multi-physics field coupled faults and reduces the risk of misdiagnosis.

[0099] When the network construction and update module dynamically optimizes the topology structure according to the optimization instruction and updates the model parameters of the energy transfer probability model, it is specifically used to:

[0100] Adjusting the node connection relationship of the fault propagation network according to the optimization instruction to generate an optimized topology structure includes:

[0101] When the difference in electromagnetic radiation energy density between two nodes in the fault propagation network is less than a preset merging threshold and the directions of the connecting edges are consistent, the two nodes are merged into one node;

[0102] Adding key connection edges between unconnected nodes based on high-probability propagation paths in the historical fault case library;

[0103] The energy transfer probability between nodes in the optimized topological structure is calculated according to the Markov chain transition probability, and the model parameters of the energy transfer probability model are iteratively updated through a gradient descent algorithm.

[0104] Specifically, Markov chain transition probability, based on Markov chain theory, describes the probability of a node state transitioning from one moment to the next in a fault propagation network. Using this probability to construct an energy transfer probability model effectively analyzes the dynamics of fault propagation and provides quantitative support for network updates.

[0105] Furthermore, the energy transfer probability between nodes in the optimized topology is calculated according to the Markov chain transition probability model based on fault energy transfer, where the initial value of the transition probability is determined by the high-probability propagation path in the historical fault case library.

[0106] In detail, the calculation formula of the transition probability is as follows:

[0107] ;

[0108] in, is the probability that a fault propagates from node a to node b via an edge, is the weight of the directed connection from node a to node b, It is the sum of the weights of all outgoing edges of node a (i.e., edges pointing from a to other nodes).

[0109] Furthermore, the calculation formula of the transfer probability calculates the energy transfer probability from node a to node b in the fault propagation network, which is used to describe the probability distribution of fault energy transferred from the current node to the adjacent node. Reflects the fault propagation "intensity" of the path from node a to node b, which is determined by the difference in electromagnetic radiation energy density between nodes. Used for normalization to ensure that the sum of transition probabilities is 1.

[0110] In detail, the greater the difference in electromagnetic radiation energy density, the lower the transfer probability.

[0111] Furthermore, the calculation formula for the weight of the directed connection edge is as follows:

[0112] ;

[0113] in, is the weight of the directed connection from node a to node b, is the electromagnetic radiation energy density of node a, is the electromagnetic radiation energy density of node b, is the energy attenuation coefficient.

[0114] Furthermore, assuming that the preset merging threshold , when the difference in electromagnetic radiation energy density between nodes a and b in the fault propagation network is less than , and the directions of the connecting edges are consistent, then nodes a and b are merged into one node.

[0115] In detail, the energy attenuation coefficient is obtained by fitting and optimizing the energy transfer path in the historical fault case library to ensure that the model error is less than 5%.

[0116] Furthermore, the energy density difference The larger the value, the smaller the exponential function value, and the edge weight The lower it is, at the same time, due to the characteristics of the exponential function, It is always within the interval (0, 1] to ensure the validity of the transition probability calculation.

[0117] Furthermore, when nodes a and b meet the following two conditions, they are merged into one node: the difference in electromagnetic radiation energy density (Preset merge threshold ); The directions of the connection edges are consistent, ensuring that the merged nodes are consistent in the direction of fault propagation.

[0118] Specifically, the gradient descent algorithm is a commonly used optimization algorithm that calculates the gradient direction of the objective function and iteratively adjusts the model parameters to minimize the objective function. In this solution, it is used to update the parameters of the energy transfer probability model to optimize the model's ability to describe fault propagation.

[0119] Furthermore, the loss function of the gradient descent algorithm is the mean square error function MSE, and the calculation formula of the loss function is as follows:

[0120] ;

[0121] in, is the loss function value, is the sample size, is a node and nodes The predicted energy transfer frequency between It is a node recorded in the historical fault case library and nodes The actual energy transfer frequency between

[0122] Furthermore, the loss function value represents the overall measure of the model prediction error, and its value range is [0, +∞). The smaller the value, the higher the prediction accuracy. The number of samples is the number of node pairs with energy transfer records in the historical fault case library. The predicted energy transfer frequency is calculated by the energy transfer probability model. The actual energy transfer frequency reflects the fault frequency between nodes in the historical data. and nodes The actual propagation probability between them is used to supervise model training.

[0123] For example: extract the actual propagation frequency of all node pairs from the historical case library, calculate the predicted frequency based on the current model parameters, calculate the MSE as the loss, backpropagate the gradient to update parameters such as β, and iterate until the loss converges or the error is <5%.

[0124] Furthermore, the actual energy transfer frequency is determined by the node To Node The number of fault propagation times divided by the number of nodes The total number of fault propagation times is obtained.

[0125] Specifically, topology optimization refers to calculating the graph structure similarity between the newly added node and the historical topology structure when a new node is detected in the fault propagation network. If the similarity is lower than the preset similarity threshold, the topology optimization instruction is triggered. For example, the graph edit distance can be used to calculate the graph structure similarity, and the similarity threshold is set to 0.7. At this time, the network topology structure is adjusted according to the instruction: when the difference in electromagnetic radiation energy density between two nodes is less than the preset merging threshold and the direction of the connecting edge is consistent, the two nodes are merged into one node to reduce redundant nodes; at the same time, referring to the high-probability propagation paths recorded in the historical fault case library, key connection edges are added between nodes that were originally unconnected but conform to the high-probability propagation logic, and the network topology structure is improved to make it more in line with the actual fault propagation law.

[0126] Specifically, parameter updating involves using Markov chain transition probabilities to calculate the energy transfer probability between nodes in the optimized topology. The parameters of the energy transfer probability model are then iteratively updated using a gradient descent algorithm. In each iteration, the model parameters are adjusted based on the difference between the calculated energy transfer probability and the actual fault data, enabling the model to more accurately describe the fault propagation process and improve the network's adaptability to dynamic fault changes.

[0127] In general, through node mapping and directed edge generation, fault feature vectors are converted into network structures. Based on the constructed network, the network is adjusted through edge weight decay, topology optimization, and parameter updates. Edge weight decay optimizes local low-probability propagation paths within the network, while topology optimization and parameter updates improve the network structure and model parameters from a global perspective. The two work together, with local updates quickly responding to changes in energy distribution and global updates adapting to major adjustments to the network structure, jointly ensuring the accuracy and efficiency of the network. Finally, the optimized network state is fed back to the entire system, providing the fault location module with an up-to-date and accurate network model. This creates a tightly coordinated closed loop between the modules and enables efficient detection and analysis of high and low voltage distribution cabinet faults.

[0128] The fault location module 104 is used to match the fault patterns in the historical fault case library based on the betweenness centrality and energy distribution of the nodes in the fault propagation network, identify the key fault sources of the high and low voltage distribution cabinets, and mark the potential fault diffusion areas.

[0129] During the operation and maintenance of high- and low-voltage distribution cabinets, traditional fault location technology cannot dynamically track fault propagation paths, resulting in significant positioning errors and making it difficult to quickly and accurately identify the fault source and potential propagation area. This fault location module integrates the topological and physical characteristics of the fault propagation network and intelligently matches it with a historical fault case database. This allows for precise identification of key fault sources and effective labeling of potential fault propagation areas, fundamentally resolving the positioning challenges inherent in existing technologies.

[0130] The fault location module is specifically used to match the fault patterns in the historical fault case library based on the betweenness centrality and energy distribution of the nodes in the fault propagation network, identify the key fault sources of the high and low voltage distribution cabinets, and mark the potential fault diffusion areas:

[0131] The Brandes algorithm is used to calculate the betweenness centrality of each node in the fault propagation network, including:

[0132] Traversing all nodes in the fault propagation network as source nodes, calculating the shortest paths from the source nodes to other nodes based on breadth-first search, counting the number of occurrences of each node in all shortest paths and normalizing the counts to betweenness values;

[0133] extracting the energy distribution of the electromagnetic radiation energy density from the node weights of the fault propagation network;

[0134] Constructing a multidimensional feature vector of the betweenness centrality and the energy distribution, and calculating the matching degree between the multidimensional feature vector and the failure mode in the historical failure case library based on the cosine similarity algorithm, wherein the historical failure case library stores the mapping relationship between the failure mode and the node attribute through an association matrix;

[0135] According to the sequence table of betweenness centrality and the matching degree, selecting nodes with betweenness centrality higher than a preset betweenness centrality threshold and the highest matching degree score as key fault sources;

[0136] The adjacent nodes of the critical fault source are traversed, and the adjacent nodes with a high probability of propagation paths to the critical fault source are identified according to the energy transfer probability model, and marked as potential fault diffusion areas.

[0137] Specifically, betweenness centrality is a metric used to measure the importance of nodes in a network. In a fault propagation network, a higher betweenness centrality for a node indicates a greater number of shortest paths passing through it, implying that the node plays a stronger hub role in the fault propagation process and is more likely to be a key fault source or a critical node for fault propagation.

[0138] In detail, the Brandes algorithm is an efficient algorithm for calculating the betweenness centrality of network nodes, with a time complexity of O(nm) (where n is the number of nodes and m is the number of edges). The algorithm traverses all nodes in the network as source nodes, calculates the shortest paths from the source nodes to other nodes, and counts the number of times each node appears in these shortest paths, thereby determining the betweenness centrality of the nodes.

[0139] Specifically, energy distribution refers to the distribution of electromagnetic radiation energy density at each node in a fault propagation network. Node weights are determined by electromagnetic radiation energy density. Energy distribution reflects the concentration of energy in different fault regions and is an important indicator for determining fault severity and propagation trends.

[0140] Specifically, a multidimensional feature vector is a combination of multiple dimensions of features, such as the betweenness centrality and energy distribution of nodes in the fault propagation network. This vector comprehensively describes the topological attributes and physical characteristics of the nodes and is used to match and analyze failure patterns in a historical fault case library.

[0141] Specifically, the cosine similarity algorithm is used to calculate the similarity between two vectors. It measures their directional similarity by calculating the cosine of the angle between the two vectors. In this solution, it is used to calculate the matching degree between the current multidimensional feature vector and the fault pattern vectors in the historical fault case library. The closer the cosine value is to 1, the higher the matching degree.

[0142] Specifically, the historical fault case database stores a large amount of historical fault information for high and low voltage distribution cabinets. It records the mapping between different fault modes and node attributes (such as betweenness centrality, energy distribution, and fault eigenvectors) in the form of an association matrix, providing a reference for current fault location.

[0143] Specifically, a critical fault source refers to the core location or component that causes a fault in a high- and low-voltage distribution cabinet. By analyzing the fault propagation network and combining betweenness centrality with the matching degree with historical fault patterns, the node most likely to be the origin of the fault is selected as the critical fault source.

[0144] Specifically, a potential fault spread region identifies areas where a fault may further spread, based on the connectivity and energy transfer probabilities between the critical fault source and its adjacent nodes in the fault propagation network. Identifying potential fault spread regions helps operations personnel take proactive measures to prevent the fault from escalating.

[0145] Furthermore, the Brandes algorithm is used to process the fault propagation network. Calculating node betweenness centrality involves the following steps: First, all nodes in the network are traversed sequentially as source nodes. For each source node, the shortest path from that source node to all other nodes in the network is calculated using a breadth-first search algorithm. During the calculation process, the nodes traversed by each shortest path are recorded. After the traversal is complete, the number of times each node appears in all shortest paths is counted. To facilitate comparison of node importance in networks of different sizes, these occurrences are normalized to obtain a betweenness value for each node. For example, if node A appears 50 times in all shortest path calculations, its betweenness value after normalization might be 0.7 (the specific value depends on the network size and calculation method), indicating that node A has a high betweenness centrality in the fault propagation network and may play a key role in the fault propagation process.

[0146] Furthermore, the electromagnetic radiation energy density information corresponding to each node in the fault propagation network is directly extracted from the node weights, forming the energy distribution of the entire network. Because the node weights are determined during the network construction process based on the electromagnetic radiation energy density corresponding to the abnormal frequency band characteristics, energy-related data for each node can be directly obtained. By analyzing this data, the distribution of fault energy at different locations within the distribution cabinet can be understood. For example, higher energy density at certain nodes indicates that these areas may have more serious faults or fault development trends.

[0147] Specifically, the calculated node betweenness centrality and the extracted energy distribution information are combined to construct a multidimensional feature vector. For example, for a node, its multidimensional feature vector might be [betweenness centrality = 0.85, energy density = 0.7], which comprehensively describes the node's properties in terms of network topology and physical characteristics.

[0148] Furthermore, the cosine similarity algorithm is used to calculate the matching degree between the constructed multidimensional feature vector and the fault mode vectors stored in the historical fault case library. The historical fault case library stores the node attribute feature vectors corresponding to different fault modes through the association matrix. For example, the arc fault mode may correspond to the feature vector of [betweenness centrality = 0.82, energy density = 0.68]. The cosine similarity between the multidimensional feature vector of the current node and each historical fault mode vector is calculated to obtain the matching degree score of the node with different fault modes. For example, if the calculated cosine similarity is 0.98, it means that the node has a high matching degree with the corresponding historical fault mode, and the fault condition represented by the node is relatively similar to a certain fault mode in the historical case.

[0149] Specifically, based on the calculated betweenness centrality sequence table and matching scores, a preset betweenness centrality threshold (e.g., 0.8) is set. Nodes with betweenness centralities above this threshold and the highest matching scores are selected and identified as key fault sources. For example, if node B has a betweenness centrality of 0.9 and a matching score of 95% with historical fault patterns, which is higher than other nodes and meets the screening criteria, node B can be identified as the key fault source for the current high and low voltage distribution cabinet. This means that the fault is most likely to originate from the location or component represented by this node.

[0150] Specifically, after identifying the critical fault source, traverse its adjacent nodes. Based on the energy transfer probability model, analyze the energy transfer probability between the critical fault source and each adjacent node. Identify adjacent nodes with high-probability propagation paths to the critical fault source, and mark these nodes and the areas they are in as potential fault spread areas. For example, if the energy transfer probability between node C and the critical fault source, node B, is high (e.g., greater than 0.7), then node C and its surrounding area are marked as potential fault spread areas, alerting operations personnel that this area may be affected by the fault and requires special attention and protective measures.

[0151] Specifically, by calculating node betweenness centrality, the team can identify nodes that play a key role in fault propagation from the perspective of network topology. By combining energy distribution with a historical fault case database, the team can further verify the correlation between nodes and fault modes based on physical characteristics and historical experience. This multi-dimensional analysis method significantly improves fault identification accuracy to 98% compared to traditional methods that rely solely on physical quantities or simple topological structures. This effectively addresses the large fault location deviations found in existing technologies and enables rapid and accurate identification of the root cause of a fault.

[0152] Specifically, potential fault propagation areas are mapped based on an energy transfer probability model, fully considering the propagation patterns and likelihood of faults within the distribution cabinet. By proactively identifying high-probability affected areas, operations and maintenance personnel can take targeted preventive measures, such as enhanced monitoring and preemptive maintenance arrangements, to prevent further expansion of the fault, reduce the scope of the outage and economic losses, and improve the reliability and stability of power system operations.

[0153] Specifically, the introduction of a historical fault case library provides a rich reference for fault location. By matching historical fault patterns, it can draw on past troubleshooting experience to quickly determine the type and possible cause of the current fault, avoiding misjudgments or missed detections due to lack of experience with new fault scenarios. This enhances the reliability and stability of fault location, making it particularly suitable for complex and changing distribution cabinet fault scenarios.

[0154] In general, calculating node betweenness centrality and extracting energy distribution are the basic steps for fault location. The fault propagation network is analyzed from the two dimensions of network topology and physical characteristics, providing raw data for the subsequent construction of multidimensional feature vectors. The construction of multidimensional feature vectors and the matching degree calculation link integrate the features of the above two dimensions, establish a connection with the historical fault case library, and quantify the degree of match between the current fault characteristics and historical patterns through the cosine similarity algorithm. The key fault source identification step is based on the betweenness centrality and matching degree calculated previously. By setting a threshold, the most likely fault source is selected, which is the core link of fault location. Finally, the potential fault diffusion area is marked with the key fault source as the starting point, and the energy transfer probability model is used to analyze the possibility of fault propagation and determine the potentially affected area.

[0155] The positioning map generation module 105 is used to map the key fault source and the potential fault diffusion area to the physical space coordinates of the high and low voltage distribution cabinet to obtain the fault positioning map of the high and low voltage distribution cabinet.

[0156] In the fault detection and processing process of high and low voltage distribution cabinets, although the key fault sources and potential fault diffusion areas have been identified through the fault location module, the abstract network node information is difficult to directly apply to on-site operation and maintenance. Traditional methods lack intuitive means of displaying the spatial distribution of faults, making it difficult for operation and maintenance personnel to quickly grasp the fault location, affecting the efficiency of fault handling. This positioning map generation module generates a visual fault location map by mapping the fault information with the physical space coordinates of the distribution cabinet, providing operation and maintenance personnel with intuitive and accurate fault location information, effectively solving the pain points in the background technology.

[0157] When the positioning map generation module maps the key fault source and the potential fault diffusion area to the physical space coordinates of the high and low voltage distribution cabinet to obtain the fault location map of the high and low voltage distribution cabinet, it is specifically used to:

[0158] Extracting the node identifier of the key fault source and the node set of the potential fault diffusion area from the fault propagation network;

[0159] Calling a predefined sensor layout coordinate mapping table of the high and low voltage power distribution cabinet, and mapping the node identifier and the node set to the physical space coordinates of the high and low voltage power distribution cabinet based on the sensor layout coordinate mapping table;

[0160] A fault location map of the high and low voltage distribution cabinet is generated based on the physical space coordinates, and the fault location map marks the spatial positions of the key fault source and the potential fault diffusion area.

[0161] Specifically, a node identifier is a unique identifier for each node in the fault propagation network, used to distinguish nodes with different fault characteristics. During the network construction process, each node mapped by an abnormal frequency band feature is assigned a specific identifier, such as "node1" or "node2," to facilitate accurate identification and manipulation in subsequent processing.

[0162] Specifically, a node set is a collection of multiple nodes, which is used in this scheme to represent all nodes involved in the potential fault propagation area. These nodes are identified as nodes with high-probability propagation paths to the critical fault source through the energy transfer probability model, and together they constitute the area potentially affected by the fault.

[0163] Specifically, the sensor layout coordinate mapping table is a predefined and stored table that records the correspondence between the installation locations of each sensor within the high- and low-voltage distribution cabinet and its physical coordinates (including X, Y, and Z coordinates). It also establishes a mapping relationship between sensor detection data and fault propagation network nodes. This mapping table is the key data foundation for converting network nodes to physical coordinates.

[0164] In detail, physical space coordinates are used to describe the specific location information of each position inside the high and low voltage distribution cabinet in three-dimensional space. The spatial position of each point is accurately determined by the X-, Y-, and Z-axis coordinate values, such as "(x=1.2m, y=0.8m, z=0m)", so that the fault location can be clearly located in the actual cabinet.

[0165] Specifically, the fault location map is a visual chart that intuitively displays key fault sources and potential fault spread areas on the physical layout of high and low voltage distribution cabinets. Using visual elements such as color and shape to distinguish key fault sources and potential spread areas, it provides operators with clear and easy-to-understand fault location information.

[0166] Furthermore, extracting node information involves accurately extracting the node identifiers corresponding to the critical fault source and the node set of the potential fault propagation area from the fault propagation network output by the fault location module. For example, in a certain fault analysis, the fault location module identified node "node5" as the critical fault source and simultaneously identified nodes "node3," "node6," and "node8" as the potential fault propagation area. Therefore, "node5" was extracted as the key fault source node identifier, and "{node3, node6, node8}" was extracted as the node set of the potential fault propagation area. This step is a prerequisite for subsequent mapping and graph generation, ensuring accurate acquisition of fault-related node information.

[0167] Furthermore, calling the mapping table is the system calling a pre-stored sensor layout coordinate mapping table, which is configured during the installation and debugging phase of the distribution cabinet and records the association between each sensor's corresponding detection area and network node and the sensor's physical installation coordinates.

[0168] Furthermore, the coordinate conversion is based on the sensor layout coordinate mapping table, which maps the extracted key fault source node identifiers and the node set of the potential fault diffusion area to the physical space coordinates of the high and low voltage distribution cabinets one by one. In specific implementation, by querying the correspondence between the node identifier and the sensor in the mapping table, the installation coordinates of the sensor are obtained, which are the physical space coordinates of the fault location represented by the node in the cabinet. For example, if the mapping table shows that the node "node5" corresponds to the sensor installed at the coordinates "(x=1.5m, y=1.0m, z=0.5m)", then the coordinates are determined as the physical space coordinates of the key fault source "node5"; for the node set of the potential fault diffusion area, the coordinates corresponding to each node are also obtained through the mapping table.

[0169] Furthermore, data processing is to organize and format the physical space coordinate data of the mapped key fault sources and potential fault diffusion areas so that the visualization engine can process them.

[0170] Furthermore, visual presentation refers to the use of professional visualization technologies, such as the D3.js engine and ECharts, to generate fault location maps for high and low voltage distribution cabinets based on processed data. During the map generation process, different visualization elements are used to mark key fault sources and potential fault diffusion areas: key fault sources are usually marked in eye-catching red and with specific coordinate information; potential fault diffusion areas are marked with yellow or orange to clearly show the areas that may be affected by the fault. At the same time, auxiliary information such as the cabinet outline and sensor location can also be added to the map, allowing operation and maintenance personnel to more intuitively understand the relationship between the fault and the cabinet structure. For example, in the generated map, the key fault source is marked with a red dot at the coordinates "(x=1.5m, y=1.0m, z=0.5m)", and the potential fault diffusion area covers the relevant coordinate area with yellow shading, intuitively presenting the fault distribution.

[0171] Furthermore, the abstract fault propagation network node information is converted into an intuitive physical spatial location map, eliminating the difficulty for operations and maintenance personnel in understanding complex network data. Through the visual fault location map, operations and maintenance personnel can quickly locate the specific location of key fault sources and potential fault propagation areas in the distribution cabinet without complex data analysis and spatial visualization, significantly improving the efficiency of obtaining fault information.

[0172] Furthermore, the coordinate mapping step relies on a pre-defined sensor layout coordinate mapping table to convert node information into physical space coordinates. It is a key link in connecting abstract network data with actual physical locations, providing an accurate location data basis for map generation.

[0173] The self-healing control module 106 is used to output self-healing control instructions for the high and low voltage distribution cabinet based on the fault location map, wherein the self-healing control instructions include applying a reverse electromagnetic pulse to the key fault source and adjusting the contact pressure of the potential fault diffusion area.

[0174] During the operation of high- and low-voltage distribution cabinets, traditional self-healing control strategies, such as global power outages and mechanical lockouts, employ coarse-grained intervention methods. While these strategies can suppress fault progression to a certain extent, they inevitably trigger malfunctions in normal areas, causing unnecessary power outages and severely impacting the reliability and stability of power supply. This self-healing control module, based on a fault location map, achieves efficient fault suppression and isolation through a precise control strategy that applies reverse electromagnetic pulses to critical fault sources and adjusts contact pressure in potential fault propagation areas. This effectively addresses the lack of precision in self-healing control in existing technologies.

[0175] The execution of the self-healing control instruction in the self-healing control module includes:

[0176] According to the coordinates of the key fault source in the fault location map, applying a reverse electromagnetic pulse in the opposite direction of the propagation direction of the electromagnetic wave with the abnormal frequency band characteristics to the corresponding area of ​​the high and low voltage distribution cabinet;

[0177] According to the coordinates of the potential fault diffusion area in the fault location map, the contact pressure of the contacts in the potential fault diffusion area is adjusted to a target pressure value, and the target pressure value is dynamically calculated according to the propagation risk level of the energy transfer probability model.

[0178] Specifically, a reverse electromagnetic pulse (REP) is an electromagnetic pulse signal that has the same frequency as the abnormal electromagnetic wave generated by the key fault source in the high and low voltage distribution cabinets, but propagates in the opposite direction. By emitting a reverse electromagnetic pulse, the principle of electromagnetic interference is used to offset the abnormal electromagnetic radiation generated by the fault source, thereby suppressing further development of the fault.

[0179] In detail, the reverse electromagnetic pulse parameters are as follows:

[0180] The frequency range of the reverse electromagnetic pulse is ±10% of the center frequency of the detected abnormal frequency band, and the pulse intensity is 120% of the electromagnetic radiation energy of the fault source;

[0181] In detail, the phase difference is the core parameter for achieving destructive interference, ensuring that the amplitudes of the two electromagnetic waves cancel each other out when superimposed in space; the frequency of the reverse electromagnetic pulse is completely consistent with the frequency of the fault source, ensuring that the reverse pulse and the abnormal electromagnetic wave are at the same frequency, satisfying the interference condition; the phase of the reverse electromagnetic pulse (unit: rad) is obtained by superimposing the phase difference on the phase of the fault source and taking the modulus, so that the reverse pulse and the abnormal electromagnetic wave have opposite phases.

[0182] Furthermore, only electromagnetic waves with the same frequency can produce stable interference phenomena, which is a necessary prerequisite for interference; the phase difference makes the electric field intensity vector direction of the reverse pulse and the abnormal electromagnetic wave opposite, and after superposition, destructive interference is generated, thereby suppressing the electromagnetic radiation of the fault source.

[0183] Specifically, reverse electromagnetic pulse parameters are generated with the same frequency but opposite phase as the fault source, and the abnormal electromagnetic radiation generated by the fault source is offset by the principle of electromagnetic interference, thereby suppressing the development of the fault; this complies with Maxwell's equations and the principle of electromagnetic wave superposition.

[0184] Specifically, contact pressure refers to the contact pressure between electrical contacts within high- and low-voltage distribution cabinets. This pressure directly affects the electrical conductivity and mechanical stability of the contacts. Insufficient contact pressure can easily lead to faults such as increased contact resistance, heating, and even arcing. Properly adjusting contact pressure can effectively improve electrical connection performance and reduce the probability of faults.

[0185] Specifically, the target pressure value is calculated based on an energy transfer probability model and defines the contact pressure value to which the contacts should be adjusted within the potential fault propagation zone. This value is dynamically determined based on the fault propagation risk level, aiming to suppress fault propagation within the diffusion zone by optimizing contact pressure.

[0186] Furthermore, the contact pressure adjustment formula is as follows:

[0187] ;

[0188] in, is the target pressure value, As the basic pressure, is the risk level coefficient, The transmission risk level is 1-3.

[0189] In detail, the contact pressure adjustment formula dynamically calculates the target contact pressure of the contact according to the fault propagation risk level of the potential fault propagation area. , by adjusting the contact pressure to improve electrical connection performance and inhibit the spread of faults.

[0190] Furthermore, the base pressure , according to the minimum contact pressure requirements of GB / T14048.1-2020 Low-voltage switchgear and control equipment, to ensure the basic conductive performance of the contacts; the risk level coefficient can be determined by experiment and used to adjust the incremental amplitude of the pressure with the risk level. For example, in high-risk scenarios, the contact resistance needs to be reduced to below 5mΩ, corresponding to .

[0191] In detail, the propagation risk level is evaluated by the energy transfer probability model and takes values ​​of 1 (low risk), 2 (medium risk), and 3 (high risk). A larger value indicates a higher possibility of fault propagation, which means a larger pressure increment on the contact.

[0192] In detail, the energy transfer probability model calculates the fault propagation risk level of different areas by analyzing the energy relationship between nodes in the fault propagation network and historical fault data, providing a quantitative basis for the self-healing control strategy.

[0193] Furthermore, the three-dimensional physical coordinates (X, Y, and Z coordinates) of the critical fault source are extracted from the fault location map output by the fault location map generation module. For example, if the coordinates of the critical fault source are (X=1.2m, Y=0.8m, Z=0.5m), this coordinate information will serve as the spatial positioning basis for the subsequent electromagnetic pulse application.

[0194] Furthermore, equipment deployment and parameter configuration involves deploying phased array antennas (e.g., a 4×4 array) near the coordinates of the critical fault source within the high and low voltage distribution cabinets. Phased array antennas achieve directional electromagnetic beam transmission by controlling the phase of each antenna element. Based on the abnormal frequency band characteristics obtained by the fault feature extraction module, the electromagnetic wave frequency to be offset is determined (e.g., if the abnormal frequency band is detected to be 58kHz), and the phased array antenna's transmit frequency is set to that frequency. Simultaneously, the antenna phase parameters are adjusted so that the electromagnetic pulses it transmits are in the opposite direction of the electromagnetic waves generated by the fault source (a phase difference of 180°).

[0195] Furthermore, pulse emission and interference cancellation involves activating a phased array antenna to transmit a reverse electromagnetic pulse toward the critical fault source area. Utilizing the principle of electromagnetic interference, when the reverse electromagnetic pulse and the abnormal electromagnetic waves generated by the fault source intersect, they superimpose and cancel each other out, reducing the electromagnetic radiation intensity in the fault area. For example, in a specific partial discharge (PD) fault, the application of the reverse electromagnetic pulse reduced the electromagnetic interference intensity in the monitored area by 15dB, effectively suppressing the development of the PD fault.

[0196] Furthermore, regional positioning and data acquisition involves obtaining the coordinate range and node information of the potential fault propagation area from the fault location map. Simultaneously, the energy transfer probability model is used to calculate the fault propagation risk level for that area. For example, if the potential fault propagation area includes three nodes, the energy transfer probability model assesses its propagation risk level as "high."

[0197] Furthermore, target pressure calculation involves dynamically calculating the target pressure based on preset pressure regulation rules and the propagation risk level of the energy transfer probability model. This can be achieved using a pre-established pressure-risk level mapping table or through an algorithmic calculation. For example, when the risk level is "high," the target pressure is determined to be 50N based on the mapping table; if the risk level is "low," the target pressure is set to 30N.

[0198] Furthermore, pressure regulation involves installing a servo motor drive (such as a Panasonic MINASA6 series servo motor) for each contact within the potential fault propagation area. The servo motor drives the contact adjustment mechanism through high-precision position and torque control. The system converts the calculated target pressure value into a control signal for the servo motor, controlling the motor to precisely adjust the contact pressure to the target value. For example, when the target pressure value is 45N, the servo motor receives the control signal and, through a transmission device such as a screw-nut mechanism or a lever mechanism, precisely adjusts the contact pressure to 45N. This enhances the electrical connection performance of the contacts, blocks the mechanical vibration transmission path, and prevents further spread of the fault.

[0199] In summary, by using the precise location information provided by the fault location map, reverse electromagnetic pulses can directly target the critical fault source, using the principle of electromagnetic interference to specifically eliminate abnormal electromagnetic radiation. Contact pressure regulation is dynamically adjusted based on the actual risk level of the potential fault spread area, precisely targeting the potentially affected contact areas. Compared to traditional, more extensive intervention methods such as global power outages, this solution achieves precise fault location and control, avoids malfunctioning of equipment in normal areas, and significantly improves the accuracy of self-healing control.

[0200] In general, applying a reverse electromagnetic pulse can quickly reduce the electromagnetic interference intensity of the fault source, inhibiting the development of electromagnetic faults such as partial discharge. Adjusting contact pressure effectively improves electrical connection performance, reducing contact resistance and mechanical vibration, and preventing the fault from spreading further through the contact connection. The synergistic effect of these two factors can effectively suppress the fault within a short period of time, reducing the impact of the fault on the overall operation of the distribution cabinet, and improving the reliability and stability of the power system.

[0201] In general, both the reverse electromagnetic pulse application and contact pressure adjustment steps are based on fault location maps, obtaining the location information of key fault sources and potential fault propagation areas, ensuring that the control strategy can accurately act on the fault-related areas. During the specific implementation process, the reverse electromagnetic pulse application mainly suppresses the root cause of electromagnetic faults, while the contact pressure adjustment focuses on blocking the fault propagation path through mechanical connections and electrical performance. The two work together and synergistically to form a complete self-healing control strategy. At the same time, the energy transfer probability model provides data support for the fault propagation risk level during the contact pressure adjustment process, providing a basis for calculating the target pressure value.

[0202] Reference Figure 2 FIG. 1 is a flow chart of an intelligent detection method for a high and low voltage distribution cabinet provided by an embodiment of the present invention. In this embodiment, the intelligent detection method for a high and low voltage distribution cabinet includes:

[0203] S1. Synchronously collect the electromagnetic radiation spectrum and cabinet surface vibration waveform of high and low voltage distribution cabinets to generate a multi-physics field coupling data set;

[0204] S2. Performing a time-frequency domain joint analysis on the multi-physics field coupling data set to obtain a fault feature vector of the high and low voltage distribution cabinet;

[0205] S3. Constructing and dynamically updating a fault propagation network based on the fault feature vector;

[0206] S4. Based on the betweenness centrality and energy distribution of nodes in the fault propagation network, match the fault patterns in the historical fault case library to identify key fault sources and mark potential fault diffusion areas;

[0207] S5. Mapping the key fault source and the potential fault diffusion area to the physical space coordinates of the high and low voltage distribution cabinet to obtain a fault location map of the high and low voltage distribution cabinet;

[0208] S6. Outputting self-healing control instructions for the high and low voltage distribution cabinet based on the fault location map, wherein the self-healing control instructions include applying a reverse electromagnetic pulse to the key fault source and adjusting the contact pressure of the potential fault diffusion area.

[0209] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0210] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) refers to the theories, methods, techniques, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to achieve optimal results.

[0211] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. An intelligent detection system for high and low voltage distribution cabinets, characterized in that: The system comprises: The data acquisition module is used to synchronously collect the electromagnetic radiation spectrum and cabinet surface vibration waveform of the high and low voltage distribution cabinets to generate a multi-physics field coupling data set; A feature extraction module is used to perform a time-frequency domain joint analysis on the multi-physics field coupling data set to obtain a fault feature vector of the high and low voltage distribution cabinet; A network construction and update module is used to construct and dynamically update a fault propagation network based on the fault feature vector, wherein: The nodes of the fault propagation network are generated by mapping the abnormal electromagnetic frequency bands detected by electromagnetic sensors installed at specific physical coordinate positions of high and low voltage distribution cabinets. Each node corresponds to an abnormal electromagnetic radiation frequency band, and the node weight is determined by the electromagnetic radiation energy density corresponding to the abnormal electromagnetic radiation frequency band. The generation of directed connection edges between nodes in the fault propagation network refers to determining the connection relationship between nodes based on the time correlation of the phase mutation characteristics in the fault feature vector. When a phase mutation is detected in the vibration signals associated with the abnormal frequency band characteristics corresponding to two nodes, and the phase mutations have a sequential relationship, a directed connection edge is generated from the node corresponding to the first phase mutation to the node corresponding to the later phase mutation. If the vibration signal associated with node A undergoes a phase mutation at time t1, and the vibration signal associated with node B undergoes a phase mutation at time t2, a directed edge from node A to node B is generated to clarify the possible propagation path of the fault. A fault location module is used to match the betweenness centrality and energy distribution of nodes in the fault propagation network with the fault patterns in the historical fault case library, identify the key fault sources of the high and low voltage distribution cabinets, and mark potential fault diffusion areas; A positioning map generation module is used to map the key fault source and the potential fault diffusion area to the physical space coordinates of the high and low voltage distribution cabinet to obtain a fault location map of the high and low voltage distribution cabinet; A self-healing control module is configured to output a self-healing control instruction for the high and low voltage distribution cabinet based on the fault location map, wherein the self-healing control instruction includes applying a reverse electromagnetic pulse to the key fault source and adjusting the contact pressure of the potential fault diffusion area, wherein: Deploy phased array antenna equipment near the coordinates of key fault sources inside high and low voltage distribution cabinets. Phased array antennas can achieve directional transmission of electromagnetic beams by controlling the phase of each antenna element. Extracting the fault source frequency of the abnormal frequency band characteristic corresponding to the key fault source, and generating a reverse electromagnetic pulse based on the fault source frequency; The phased array antenna is activated to transmit reverse electromagnetic pulses to the critical fault source area. Utilizing the principle of electromagnetic interference, when the reverse electromagnetic pulses meet the abnormal electromagnetic waves generated by the fault source, the two superimpose and cancel each other out, reducing the electromagnetic radiation intensity in the fault area. The coordinate range and node information of the potential fault diffusion area are obtained from the fault location map. A servo motor drive device is installed for each contact in the potential fault diffusion area. The servo motor drives the contact adjustment mechanism through high-precision position control and torque control. Based on the propagation risk level corresponding to the potential fault diffusion area, the target pressure value of the contact is calculated, wherein the calculation formula of the target pressure value is: ； in, is the target pressure value, As the basic pressure, is the risk level coefficient, is the level of transmission risk; The servo motor is driven to adjust the contact pressure of the contact to the target pressure value.

2. The intelligent detection system for high and low voltage distribution cabinets according to claim 1, characterized in that: When the data acquisition module performs synchronous acquisition of the electromagnetic radiation spectrum and cabinet surface vibration waveform of the high and low voltage distribution cabinets to generate a multi-physics field coupling data set, it is specifically used to: Arrange electromagnetic sensors in the electromagnetic radiation hotspots of high and low voltage distribution cabinets, install vibration sensors at vibration-sensitive points on the cabinet surfaces of the high and low voltage distribution cabinets, and perform time alignment on the electromagnetic sensors and the vibration sensors; Using the electromagnetic sensor to collect the electromagnetic radiation spectrum of the high and low voltage distribution cabinet; Using the vibration sensor to collect the cabinet surface vibration waveform of the high and low voltage distribution cabinet; The electromagnetic radiation spectrum and the cabinet surface vibration waveform are aligned according to a time window to generate a multi-physics field coupling data set of the high and low voltage distribution cabinet.

3. The intelligent detection system for high and low voltage distribution cabinets according to claim 1, characterized in that: When the feature extraction module performs a time-frequency domain joint analysis on the multi-physics field coupling data set to obtain the fault feature vector of the high and low voltage distribution cabinet, it is specifically used to: Performing frequency domain analysis on the electromagnetic radiation spectrum in the multi-physics field coupling data set to extract abnormal frequency band features of energy mutation; Performing time-domain phase analysis on the cabinet surface vibration waveform in the multi-physics field coupling data set to extract time-series-related phase mutation features; The abnormal frequency band feature is fused with the phase mutation feature to obtain a fault feature vector representing the fault mode.

4. The intelligent detection system for high and low voltage distribution cabinets according to claim 1, characterized in that: When the network construction and update module constructs and dynamically updates the fault propagation network based on the fault feature vector, it is specifically used to: The abnormal frequency band features in the fault feature vector are mapped to nodes of the fault propagation network, wherein the weight of each node is determined by the electromagnetic radiation energy density corresponding to the abnormal frequency band features. The nodes are generated as follows: Analyzing abnormal frequency band features in the fault feature vector, wherein the abnormal frequency band features include: center frequency and bandwidth; Creating a node corresponding to the abnormal frequency band feature in the fault propagation network according to the center frequency and the bandwidth; Assigning the node weight to the electromagnetic radiation energy density corresponding to the abnormal frequency band feature; generating directed connection edges between the nodes based on the time correlation of the phase mutation features in the fault feature vector, wherein the direction of the directed connection edges is determined by the temporal relationship of the phase mutation features; When it is detected that the difference in electromagnetic radiation energy density between nodes in the fault propagation network exceeds a preset difference threshold, an edge weight decay operation is triggered, and the edge weights corresponding to the low-probability propagation paths in the directed connection edges are decayed based on the energy transfer probability model, including: When the difference in electromagnetic radiation energy density between the first node and the second node exceeds a preset threshold, the judgment value of the edge weight decay trigger condition is 1, triggering the edge weight decay operation; When the difference in electromagnetic radiation energy density between the first node and the second node does not exceed a preset threshold, the judgment value of the edge weight decay triggering condition is 0, and the edge weight decay operation is not triggered. The triggering conditions of the edge weight decay operation are as follows: ； in, is the judgment value of the edge weight decay trigger condition, yes The electromagnetic radiation energy density of the node, yes The electromagnetic radiation energy density of the node, is the larger value of the two node energy densities, is the preset difference threshold; The directed connection edge weight of the edge between the first node and the second node is updated according to a directed connection edge weight formula, wherein the directed connection edge weight formula is: ； in, is the weight of the directed connection from node a to node b, is the electromagnetic radiation energy density of node a, is the electromagnetic radiation energy density of node b, is the energy attenuation coefficient; The energy transfer probability between nodes in the topological structure of the fault propagation network is calculated according to the directed connection edge weights, wherein the calculation formula of the energy transfer probability is as follows: ； in, is the probability that a fault propagates from node a to node b via an edge, is the weight of the directed connection from node a to node b, is the sum of the weights of all edges of node a; When it is detected that the graph structure similarity between the newly added node in the fault propagation network and the historical topology structure is lower than a preset similarity threshold, a topology optimization instruction is triggered, the historical topology structure is dynamically optimized according to the topology optimization instruction, and the model parameters of the energy transfer probability model are updated, wherein the graph structure similarity is calculated using a graph edit distance algorithm, including: Calculate the edit distance between the fault propagation network after adding the new node and the historical topology using a graph edit distance algorithm, and calculate the graph structure similarity based on the edit distance; The updated edge weights, optimized topology, and updated energy transfer probability model are synchronized to the fault propagation network.

5. The intelligent detection system for high and low voltage distribution cabinets according to claim 4, characterized in that: When the network construction and update module dynamically optimizes the historical topology structure according to the topology optimization instruction and updates the model parameters of the energy transfer probability model, it is specifically used to: Adjusting the node connection relationship of the fault propagation network according to the optimization instruction to generate an optimized topology structure, including: merging two nodes in the fault propagation network into one node when the difference in electromagnetic radiation energy density between them is less than a preset merging threshold and the directions of the connection edges are consistent; adding key connection edges between unconnected nodes based on high-probability propagation paths in the historical fault case library; The energy transfer probability between nodes in the optimized topological structure is calculated according to the Markov chain transition probability, and the model parameters of the energy transfer probability model are iteratively updated through a gradient descent algorithm.

6. The intelligent detection system for high and low voltage distribution cabinets according to claim 4, characterized in that: The fault location module is specifically used to match the fault patterns in the historical fault case library based on the betweenness centrality and energy distribution of the nodes in the fault propagation network, identify the key fault sources of the high and low voltage distribution cabinets, and mark the potential fault diffusion areas: The Brandes algorithm is used to calculate the betweenness centrality of each node in the fault propagation network, including: Traversing all nodes in the fault propagation network as source nodes, calculating the shortest paths from the source nodes to other nodes based on breadth-first search, counting the number of occurrences of each node in all shortest paths and normalizing the counts to betweenness values; extracting the energy distribution of the electromagnetic radiation energy density from the node weights of the fault propagation network; Constructing a multidimensional feature vector of the betweenness centrality and the energy distribution, and calculating the matching degree between the multidimensional feature vector and the failure mode in the historical failure case library based on the cosine similarity algorithm, wherein the historical failure case library stores the mapping relationship between the failure mode and the node attribute through an association matrix; According to the sequence table of betweenness centrality and the matching degree, selecting nodes with betweenness centrality higher than a preset betweenness centrality threshold and the highest matching degree score as key fault sources; The adjacent nodes of the critical fault source are traversed, and the adjacent nodes with a high probability of propagation paths to the critical fault source are identified according to the energy transfer probability model, and marked as potential fault diffusion areas.

7. The intelligent detection system for high and low voltage distribution cabinets according to claim 1, characterized in that: When the positioning map generation module maps the key fault source and the potential fault diffusion area to the physical space coordinates of the high and low voltage distribution cabinet to obtain the fault location map of the high and low voltage distribution cabinet, it is specifically used to: Extracting the node identifier of the key fault source and the node set of the potential fault diffusion area from the fault propagation network; Calling a predefined sensor layout coordinate mapping table of the high and low voltage power distribution cabinet, and mapping the node identifier and the node set to the physical space coordinates of the high and low voltage power distribution cabinet based on the sensor layout coordinate mapping table; A fault location map of the high and low voltage distribution cabinet is generated based on the physical space coordinates, and the fault location map marks the spatial positions of the key fault source and the potential fault diffusion area.

8. The intelligent detection system for high and low voltage distribution cabinets according to claim 4, characterized in that: The execution of the self-healing control instruction in the self-healing control module includes: According to the coordinates of the key fault source in the fault location map, applying a reverse electromagnetic pulse in the opposite direction of the propagation direction of the electromagnetic wave with the abnormal frequency band characteristics to the corresponding area of ​​the high and low voltage distribution cabinet; According to the coordinates of the potential fault diffusion area in the fault location map, the contact pressure of the contacts in the potential fault diffusion area is adjusted to a target pressure value, and the target pressure value is dynamically calculated according to the propagation risk level of the energy transfer probability model.

9. An intelligent detection method for high and low voltage distribution cabinets, characterized in that: The method comprises: S1. Synchronously collect the electromagnetic radiation spectrum and cabinet surface vibration waveform of high and low voltage distribution cabinets to generate a multi-physics field coupling data set; S2. Performing a time-frequency domain joint analysis on the multi-physics field coupling data set to obtain a fault feature vector of the high and low voltage distribution cabinet; S3. Constructing and dynamically updating a fault propagation network based on the fault feature vector, wherein: The nodes of the fault propagation network are generated by mapping the abnormal electromagnetic frequency bands detected by electromagnetic sensors installed at specific physical coordinate positions of high and low voltage distribution cabinets. Each node corresponds to an abnormal electromagnetic radiation frequency band, and the node weight is determined by the electromagnetic radiation energy density corresponding to the abnormal electromagnetic radiation frequency band. The generation of directed connection edges between nodes in the fault propagation network refers to determining the connection relationship between nodes based on the time correlation of the phase mutation characteristics in the fault feature vector. When a phase mutation is detected in the vibration signals associated with the abnormal frequency band characteristics corresponding to two nodes, and the phase mutations have a sequential relationship, a directed connection edge is generated from the node corresponding to the first phase mutation to the node corresponding to the later phase mutation. If the vibration signal associated with node A undergoes a phase mutation at time t1, and the vibration signal associated with node B undergoes a phase mutation at time t2, a directed edge from node A to node B is generated to clarify the possible propagation path of the fault. S4. Based on the betweenness centrality and energy distribution of nodes in the fault propagation network, match the fault patterns in the historical fault case library to identify key fault sources and mark potential fault diffusion areas; S5. Mapping the key fault source and the potential fault diffusion area to the physical space coordinates of the high and low voltage distribution cabinet to obtain a fault location map of the high and low voltage distribution cabinet; S6. Outputting a self-healing control instruction for the high and low voltage distribution cabinet based on the fault location map, wherein the self-healing control instruction includes applying a reverse electromagnetic pulse to the key fault source and adjusting the contact pressure of the potential fault diffusion area, wherein: Deploy phased array antenna equipment near the coordinates of key fault sources inside high and low voltage distribution cabinets. Phased array antennas can achieve directional transmission of electromagnetic beams by controlling the phase of each antenna element. Extracting the fault source frequency of the abnormal frequency band characteristic corresponding to the key fault source, and generating a reverse electromagnetic pulse based on the fault source frequency; The phased array antenna is activated to transmit reverse electromagnetic pulses to the critical fault source area. Utilizing the principle of electromagnetic interference, when the reverse electromagnetic pulses meet the abnormal electromagnetic waves generated by the fault source, the two superimpose and cancel each other out, reducing the electromagnetic radiation intensity in the fault area. The coordinate range and node information of the potential fault diffusion area are obtained from the fault location map. A servo motor drive device is installed for each contact in the potential fault diffusion area. The servo motor drives the contact adjustment mechanism through high-precision position control and torque control. The adjustment of the contact pressure in the potential fault propagation area is performed in the following manner: Locate the contacts according to the coordinates of the potential fault diffusion area in the fault location map; Based on the propagation risk level corresponding to the potential fault diffusion area, the target pressure value of the contact is calculated, wherein the calculation formula of the target pressure value is: ； in, is the target pressure value, As the basic pressure, is the risk level coefficient, is the level of transmission risk; The servo motor is driven to adjust the contact pressure of the contact to the target pressure value.

Images