Power system operation and maintenance method and system of intelligent power distribution cabinet for weak current control

Abstract

The invention relates to the technical field of power operation and maintenance, in particular to a power system operation and maintenance method and system of an intelligent power distribution cabinet for weak current control. The method comprises the following steps: acquiring power operation data and sensing data of a power distribution cabinet, and analyzing a topological structure of a power system to obtain a dynamic topological structure of the power system; performing potential load anomaly analysis on the sensing data of the power distribution cabinet to obtain load anomaly node data of the power system; performing load anomaly influence structure division based on the load anomaly node data of the power system to obtain load anomaly influence nodes; performing abnormal node classification on the load abnormal influence nodes according to the sensing data of the power distribution cabinet to obtain three-phase power load offset nodes and three-phase power improper wiring nodes; and performing power distribution reconstruction on the three-phase power load offset node and the three-phase power improper wiring node to obtain abnormal node power distribution data. According to the invention, the weak current control efficiency and the power distribution energy efficiency can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power operation and maintenance, and in particular to a power system operation and maintenance method and system for an intelligent distribution cabinet for weak current control. Background Art

[0002] Smart distribution cabinets not only distribute and manage power loads but also perform real-time monitoring, fault warnings, and energy optimization to ensure stable system operation. However, with the rapid development of intelligent and digital technologies, existing low-voltage control distribution cabinets still face numerous challenges in power system operation and maintenance. The operation and maintenance of low-voltage control distribution cabinets primarily relies on traditional manual inspections. This involves personnel regularly checking the operating status of electrical equipment, measuring key parameters such as current, voltage, and temperature, and conducting fault diagnosis and maintenance based on experience. Manual inspections are periodic and, due to limited inspection intervals, may not detect sudden faults in a timely manner. For example, certain electrical faults, such as line short circuits, cable aging, and load overloads, may occur during inspection intervals, causing system anomalies or even safety incidents. Traditional distribution cabinet operation and maintenance methods typically focus solely on equipment operating status while neglecting energy optimization management. Due to the lack of intelligent analysis of the power consumption and operating mode of each load, the system struggles to dynamically adjust power distribution strategies, resulting in energy waste and impacting overall energy efficiency. Summary of the Invention

[0003] Based on this, it is necessary for the present invention to provide a power system operation and maintenance method and system for an intelligent distribution cabinet for weak current control to solve at least one of the above technical problems.

[0004] To achieve the above objectives, a power system operation and maintenance method for an intelligent power distribution cabinet for weak current control is provided, comprising the following steps:

[0005] Step S1: Obtain power operation data and sensor data of the distribution cabinet, and perform power system topology analysis based on the power operation data of the distribution cabinet to obtain a dynamic topology of the power system;

[0006] Step S2: performing power system energy efficiency pattern recognition on the distribution cabinet sensor data through the dynamic topology structure of the power system to obtain power system energy efficiency pattern data; performing potential load anomaly analysis on the power system energy efficiency pattern data to obtain power system load abnormality node data;

[0007] Step S3: Analyze the load anomaly dominant factors based on the power system load anomaly node data, and divide the power system dynamic topology into load anomaly impact structures according to the load anomaly dominant factors to obtain load anomaly impact nodes;

[0008] Step S4: Evaluate the three-phase current balance of the distribution cabinet for nodes affected by abnormal loads based on the distribution cabinet sensor data, and classify abnormal nodes based on the current balance of the distribution cabinet to obtain three-phase load offset nodes and three-phase improper connection nodes;

[0009] Step S5: Reconstruct the power distribution of the three-phase load offset nodes and the three-phase improperly connected nodes to obtain the abnormal node power distribution data, and transmit it to the distribution cabinet management platform to execute the power distribution task.

[0010] The present invention provides an intelligent distribution cabinet anomaly detection and power reconstruction method, which can effectively solve the problems of traditional weak-current control distribution cabinet operation and maintenance relying on manual inspection, delayed response, insufficient anomaly perception, and weak energy efficiency management capabilities compared to the existing technology, and has significant technical advantages and engineering application value. By obtaining the power operation data and sensor data of the distribution cabinet, and combining it with the dynamic topology structure of the power system for in-depth modeling, the electrical connection relationship at the device level can be restored while ensuring the comprehensiveness of the data, providing a structural basis for subsequent energy efficiency identification and anomaly analysis. The introduced Z-score standardization and time offset correction threshold (±1 second) can eliminate the errors caused by asynchronous sampling of the equipment while ensuring data accuracy, and improve the stability and robustness of anomaly analysis. Energy efficiency pattern recognition based on topological structure can not only realize node-level energy consumption behavior modeling, but also dig out high-risk load areas associated with topological positions, and realize accurate locking of potential load abnormal nodes, effectively making up for the defects of coarse granularity and low sensitivity of anomaly detection in traditional strategies. By extracting the dominant factors of load anomalies and partitioning the topological impact structure, a multi-dimensional analysis of the causes of load anomalies is achieved. This not only identifies "where" the problem occurs but also further clarifies "why" it occurs, facilitating targeted system optimization and scheduling. During the three-phase current balance assessment phase, setting balance interval thresholds (e.g., 0.85 and 0.65) allows for hierarchical and classified management of abnormal nodes, improving the accuracy of anomaly identification. The introduction of the improper wiring identifier variable fi∈{0,1} constructs a joint structural and behavioral identification framework, effectively improving the interpretability and reliability of anomaly classification. Finally, through power distribution reconstruction, differentiated distribution strategies are adjusted for the classified abnormal nodes. Combined with the task execution capabilities of the power management platform, closed-loop anomaly handling is implemented, ensuring the distribution system's dynamic recovery and energy efficiency optimization capabilities under load or wiring anomalies. The path tracing depth is set to [3,5] layers to balance the tracing computation overhead with the coverage of anomaly information dissemination. A factor combination extraction strategy with a cumulative contribution ratio greater than 0.85 ensures sufficient retention of key factors and simplicity after dimensionality reduction.

[0011] Optionally, step S1 specifically includes:

[0012] Step S11: acquiring power operation data and sensor data of the distribution cabinet, and performing data preprocessing on the power operation data and sensor data of the distribution cabinet to obtain the power operation data to be analyzed and the sensor data of the distribution cabinet to be analyzed;

[0013] Step S12: extracting the voltage distribution characteristics and current flow characteristics of the distribution cabinet based on the power operation data to be analyzed;

[0014] Step S13: performing a power system backbone topology analysis based on the voltage distribution characteristics of the power distribution cabinet to obtain a power system backbone topology structure;

[0015] Step S14: using the current flow characteristics of the distribution cabinet to complete the power system global topology of the power system backbone topology, thereby obtaining the power system topology;

[0016] Step S15: performing power flow calculation on the power system topology, and constructing a dynamic power system topology according to the power flow calculation results.

[0017] The present invention improves data consistency and integrity by uniformly pre-processing the power operation data and sensor data of the distribution cabinet, effectively reduces the impact of data noise and missing values, and ensures the reliability of power flow analysis and topology reconstruction. By utilizing the voltage distribution characteristics of the distribution cabinet, the node voltage level and change trend are accurately identified to help identify the backbone network of the power system, especially to assist in identifying weak-current risk nodes with large voltage drops. The topological structure is completed by the current flow direction characteristics, which makes up for the shortcomings of traditional topology identification in the directionality of branch connection, improves the accuracy of local branch identification, and automatically corrects incorrect wiring by setting the current vector angle change threshold (such as 15°) to enhance the fault tolerance of the system. In the power flow calculation, parameters such as the upper and lower limits of the node load (such as ±20%) are reasonably set to adapt to different load fluctuations and improve the convergence and real-time performance of the power flow solution. The power flow results are mapped to the topological structure to form a dynamic topological model, which can not only track the operating status in real time, but also provide an accurate reference for the weak-current control and adjustment strategy.

[0018] Optionally, step S13 is specifically as follows:

[0019] Step S131: classify the voltage levels according to the voltage distribution characteristics of the power distribution cabinet, set the voltage threshold interval to [400V, 220V] to partition the bus nodes, and construct a bus hierarchical structure to obtain bus hierarchical data;

[0020] Step S132: Calculate the potential difference between different bus levels based on the bus level data, select lines with a potential difference constraint less than or equal to [5V, 20V] as trunk and branch lines, perform topology consistency verification, and generate initial trunk and branch line data;

[0021] Step S133: constructing a preliminary trunk topology graph with the busbar nodes in the busbar level data as vertices and the initial trunk-branch data as edges, and performing connectivity analysis to remove isolated nodes to generate a preliminary trunk topology structure;

[0022] Step S134: Analyze the power flow direction based on the voltage distribution characteristics and current flow characteristics of the distribution cabinet, and adjust the branch weights of the preliminary trunk topology according to the power flow direction to obtain an optimized trunk topology;

[0023] Step S135: performing topological mapping on the optimized backbone topology structure to obtain the power system backbone topology structure.

[0024] The present invention sets [400V, 220V] as the voltage threshold interval in the voltage hierarchy classification, which helps to distinguish the main bus from the terminal branch and improve the accuracy of bus hierarchical structure recognition. Setting the potential difference constraint to [5V, 20V] in the main branch screening can effectively eliminate non-physical connection relationships, improve the reliability of topology boundary judgment, and adapt to the situation where the voltage drop between buses in the weak current area is small but there is still a logical connection. Topology consistency verification and isolated node elimination can avoid redundant connections from interfering with the overall structure. Power flow direction analysis combined with voltage and current flow characteristics can realize dynamic correction of branch directions and improve the physical consistency of the topology diagram. The final trunk topology structure mapping result provides a clear network skeleton for weak current control, supporting node regulation and abnormality diagnosis.

[0025] Optionally, step S14 is specifically as follows:

[0026] Step S141: performing weighted directed graph modeling on the current flow characteristics of the power distribution cabinet, setting the current direction threshold to [0.3A, 3.0A], thereby constructing a current flow association graph;

[0027] Step S142: performing node alignment and edge mapping analysis on the current flow association graph and the power system backbone topology structure, identifying missing connection parts in the power system backbone topology structure, and obtaining the backbone structure node edge data to be completed;

[0028] Step S143: Perform graph structure reasoning on the edge data of the nodes to be completed in the backbone structure, perform missing connection completion, and obtain global topology prediction data;

[0029] Step S144: performing edge weight fusion and node uniform coding on the global topology prediction data and the power system backbone topology structure to obtain a preliminary power system topology structure;

[0030] Step S145: performing consistency check and closed-loop verification on the preliminary power system topology structure to obtain the power system topology structure.

[0031] The present invention achieves the accurate expansion of the backbone topology to the global topology by introducing weighted directed graph modeling and graph structure reasoning technology. Setting the current directionality threshold to [0.3A, 3.0A] can effectively filter out background current and invalid current fluctuations, retaining only lines with physical flow direction significance, which is conducive to the identification of branches with weak signals but topological significance in weak current areas. Node alignment and edge mapping analysis can accurately identify topological void areas caused by sensor loss or modeling errors, thereby improving topological integrity. Graph structure reasoning combined with edge prediction mechanism can fill in missing connections based on known structures and enhance topological recovery capabilities. Edge weight fusion and unified coding processing enhance the consistent expression of topological data from different sources, facilitating the rapid positioning of nodes and scheduling paths in subsequent weak current regulation. Finally, consistency verification and closed-loop verification are used to ensure the logical closed-loop nature and operational rationality of the topological structure.

[0032] Optionally, step S2 is specifically:

[0033] Step S21: performing spatial mapping and time window slicing on the distribution cabinet sensor data to be analyzed, and performing time window sensor-topology node binding based on the node-branch relationship in the dynamic topology structure of the power system to obtain a spatiotemporal feature map of the power system nodes;

[0034] Step S22: extracting weighted aggregated energy efficiency indicators based on the spatiotemporal characteristic graph of power system nodes to obtain node energy efficiency indicator data;

[0035] Step S23: constructing an energy efficiency attribute graph using the node energy efficiency index data, and extracting node energy efficiency behavior characteristics based on the energy efficiency attribute graph;

[0036] Step S24: clustering energy efficiency behaviors according to the node energy efficiency behavior characteristics to obtain power system energy efficiency model data;

[0037] Step S25: perform abnormal factor detection and pattern deviation analysis on the power system energy efficiency pattern data, extract potential load abnormality candidate nodes, and perform confidence scoring on the potential load abnormality candidate nodes to obtain power system load abnormality node data.

[0038] By constructing a spatiotemporal feature map of power system nodes, the present invention effectively achieves dynamic binding between distribution cabinet sensor data and topological structures, giving sensor data structural semantics and enhancing spatiotemporal perception capabilities. The node-branch relationship is introduced in the spatial mapping and time window processing links, so that energy efficiency analysis has a weak current structure basis and improves analysis accuracy. Aggregate energy efficiency index extraction can unify multiple source indicators such as voltage, current, and active power into one, enhance data expression consistency, and facilitate feature modeling. Behavioral feature extraction based on energy efficiency attribute graphs can significantly improve the ability to characterize node operating status, and is particularly suitable for identifying nodes with small but frequent load fluctuations in weak current areas. The energy efficiency clustering process strengthens the analysis of behavioral differences between nodes and improves sensitivity to structural deviations. Abnormal factor detection integrates deviation analysis and confidence scoring mechanisms to screen out occasional fluctuation interference and effectively lock in the source node of load anomaly. The overall process takes into account both weak current operating characteristics and energy efficiency pattern evolution, and is suitable for abnormal monitoring and refined operation and maintenance in high-density sensing scenarios.

[0039] Optionally, step S25 is specifically as follows:

[0040] Step S251: constructing an energy efficiency behavior residual matrix of each node based on the characteristic vector of each cluster center in the power system energy efficiency model data, and calculating the Euclidean distance between the node energy efficiency behavior characteristics and the energy efficiency behavior residual matrix of the corresponding node to obtain node energy efficiency behavior residual data;

[0041] Step S252: performing slope fitting and mutation detection based on the node energy efficiency behavior residual data, calculating the abnormal factor score, and obtaining abnormal factor score data; setting the abnormal score dynamic threshold range according to the abnormal factor score data;

[0042] Step S253: Evaluate the energy efficiency behavior coordination degree of adjacent nodes in the dynamic topology structure of the power system to obtain the energy efficiency behavior coordination degree of the adjacent nodes;

[0043] Step S254: combining the anomaly score dynamic threshold interval and the adjacent node energy efficiency behavior coordination degree, constructing a weighted anomaly distribution map, identifying and extracting high anomaly factor score nodes, and obtaining potential load anomaly candidate node data;

[0044] Step S255: perform node confidence evaluation on the potential load anomaly candidate node data, and set the scoring confidence threshold to [0.6, 0.9] to screen high-confidence nodes to obtain power system load anomaly node data.

[0045] By calculating the Euclidean distance between the node energy efficiency behavior characteristics and the behavior residual matrix, the present invention can accurately measure the difference between the node energy efficiency performance and the expected behavior, effectively identify nodes with energy efficiency deviations, and is particularly suitable for small fluctuations caused by load changes in the power system, and enhances sensitivity to abnormal behavior. Slope fitting and mutation detection technology provides an accurate basis for the scoring of abnormal factors, ensuring a rapid response to sudden changes. At the same time, a dynamic threshold interval is set so that the system can adapt to different loads and environmental changes, thereby improving the robustness of the system. By evaluating the degree of coordination of energy efficiency behavior, the potential behavioral correlation between adjacent nodes can be further captured, and the accuracy of anomaly detection can be optimized. Combined with the construction of a weighted anomaly distribution map, the degree of coordination between nodes and the scoring of anomaly factors are used to accurately extract nodes with high anomaly factors, avoiding the problem of false positives. By setting a scoring confidence threshold, it is ensured that the screened load abnormality nodes have a high reliability.

[0046] Optionally, the analysis load abnormality dominant factor described in step S3 is specifically:

[0047] The path tracing depth is set to [3,5] layers to dynamically trace the upstream and downstream paths of the power system load abnormality node data, construct a load abnormality path graph, and extract the structural position of each abnormal node in the dynamic topology of the power system and the correlation degree of adjacent nodes to obtain the load abnormality topology path feature data;

[0048] Based on the load anomaly topology path feature data and combined with the node energy efficiency index data, a topology structure-energy efficiency composite feature vector matrix is constructed;

[0049] Perform structural impact factor embedding training on the topology-energy efficiency composite feature vector matrix to generate load abnormal structure impact feature data;

[0050] Obtain the historical load fluctuation sequence and multi-dimensional sensor data of the distribution cabinet, and combine it with the abnormal load node data of the power system to perform characteristic factor correlation analysis to obtain the weight data of the abnormal load physical factor;

[0051] The load anomaly structural impact feature data and load anomaly physical factor weight data are integrated to perform multidimensional factor compression and reduction, retaining factor combinations with a cumulative contribution rate threshold greater than or equal to 0.85 to obtain the load anomaly impact factor spectrum;

[0052] Based on the load anomaly influencing factor spectrum, significance ranking and dominant factor identification are performed. The factor contribution rate threshold is set to 0.15 to extract the dominant factors of abnormal nodes and generate the load anomaly dominant factors.

[0053] By setting the path tracing depth to [3,5] layers, this method dynamically traces upstream and downstream paths of abnormal load nodes in the power system. This allows accurate capture of the upstream and downstream relationships of abnormal load nodes within the system and their impact on the global power system. Constructing a load anomaly path diagram effectively reveals the propagation path of abnormal nodes and their influence. Extracting the structural position of nodes and the degree of correlation between adjacent nodes helps identify the key role of abnormal nodes in the power system. Combining topological structure with energy efficiency indicators to construct a composite feature vector matrix enhances the system's multidimensional analysis capabilities for node behavior and provides a more comprehensive basis for anomaly identification. Embedding training of structural influencing factors accurately extracts the key influencing features of load anomalies, improving the accuracy of anomaly prediction. Correlation analysis of characteristic factors between load fluctuation sequences and sensor data provides the system with a more detailed physical-level factor analysis, further enhancing its ability to identify anomaly sources. Multidimensional factor compression and reduction and the setting of factor contribution rate thresholds help remove redundant information, highlight key factors, and improve the system's response speed and accuracy to abnormal patterns.

[0054] Optionally, the three-phase current balance evaluation of the power distribution cabinet described in step S4 is specifically as follows:

[0055] Extract the three-phase current time series data of the load abnormality-affecting node from the distribution cabinet sensor data, perform period segmentation, missing compensation, and distortion point cleaning on the three-phase current time series data, and obtain the calibrated three-phase current standardized time series data;

[0056] A three-phase current time-domain waveform tensor model is constructed based on the calibrated three-phase current standardized time series data. The three-phase current inter-phase offset rate, phase sequence inconsistency rate, and time-domain imbalance fluctuation rate are extracted using the three-phase current time-domain waveform tensor model to obtain the three-phase current imbalance characteristic data.

[0057] Based on the three-phase current imbalance characteristic data and the dynamic topology of the power system, a node-level three-phase current imbalance distribution map is constructed. The current offset tension of each abnormal node is calculated using the node-level three-phase current imbalance distribution map, and the current tension propagation weight matrix is obtained.

[0058] The current tension propagation weight matrix and the three-phase current imbalance characteristic data are fused to calculate the three-phase current balance index and obtain the abnormal node current balance data.

[0059] The present invention ensures the accuracy and continuity of the data by extracting the three-phase current time series data of the nodes affected by the load anomaly, and performs period segmentation, missing compensation and distortion point cleaning on it, thereby avoiding interference with subsequent analysis due to data missing or noise. The calibrated three-phase current standardized time series data can effectively eliminate the measurement differences between different nodes, so that the analysis results have higher reliability. The imbalance feature data extracted based on the three-phase current time domain waveform tensor model can accurately reflect the current phase offset, phase sequence inconsistency and time domain imbalance fluctuation, providing an accurate basis for subsequent abnormality analysis. By constructing a node-level three-phase current imbalance distribution map and calculating the current offset tension, the degree of current imbalance of each node can be clearly identified, which is helpful for diagnosing potential current imbalance problems in the power system, especially in the weak current control area, and can provide early warning of the risks brought by current imbalance.

[0060] Optionally, the abnormal node classification in step S4 is specifically:

[0061] The abnormal node current balance data is divided into balance intervals to obtain graded current balance label data;

[0062] Based on the dynamic topology structure of the power system, the topological adjacent information and branch connection relationship of the nodes affected by load anomalies are extracted, and the node-phase connection matrix is constructed.

[0063] Combined with the node interphase connection matrix and the graded current balance label data, the three-phase structure distribution of each load abnormality-affected node is analyzed to obtain the three-phase phase sequence consistency characteristic data;

[0064] Perform topological structure projection and constraint rule matching on the three-phase phase sequence consistency feature data, perform wiring pattern recognition, and obtain improper wiring identification data;

[0065] The hierarchical current balance label data and improper wiring identification data are integrated for joint logical judgment, and the load abnormality-affected nodes are divided into three-phase load offset nodes and three-phase improper wiring nodes.

[0066] The present invention ensures the quantitative analysis and accurate classification of current balance by dividing the current balance data of abnormal nodes into balance intervals, which helps to identify potential current imbalance problems in the power system, especially in the weak current control area. By extracting hierarchical label data, fine management of different current balance levels can be achieved. Based on the dynamic topology of the power system, the topological adjacent information and branch connection relationship of the nodes are extracted, which can accurately depict the electrical connection relationship of each node in the power system, providing a basis for further analysis. By combining the node phase connection matrix and the current balance label data, the three-phase structure distribution of the load abnormality node can be fully analyzed, which helps to identify the three-phase phase sequence consistency problem and improper wiring in the power system, avoiding power system instability caused by wiring errors. By combining the logic judgment method, the load abnormality affecting nodes are divided into three-phase load offset nodes and three-phase improper wiring nodes, which helps to accurately locate the source of the problem, respond quickly and take effective measures.

[0067] Optionally, this specification also provides a power system operation and maintenance system for an intelligent power distribution cabinet for weak current control, which is used to execute the power system operation and maintenance method for the intelligent power distribution cabinet for weak current control as described above. The power system operation and maintenance system for the intelligent power distribution cabinet for weak current control includes:

[0068] The topology analysis module is used to obtain the power operation data and sensor data of the distribution cabinet, and perform power system topology analysis based on the power operation data of the distribution cabinet to obtain the dynamic topology of the power system;

[0069] The potential load anomaly analysis module is used to identify the power system energy efficiency pattern of the distribution cabinet sensor data through the dynamic topology of the power system to obtain the power system energy efficiency pattern data; perform potential load anomaly analysis on the power system energy efficiency pattern data to obtain the power system load anomaly node data;

[0070] The abnormal impact structure division module is used to analyze the load abnormality dominant factors based on the load abnormality node data of the power system, and divide the load abnormality impact structure of the power system dynamic topology structure according to the load abnormality dominant factors to obtain the load abnormality impact nodes;

[0071] The three-phase current balance evaluation module is used to evaluate the three-phase current balance of the distribution cabinet at the nodes affected by abnormal loads based on the distribution cabinet sensor data, and classify abnormal nodes based on the current balance of the distribution cabinet to obtain the three-phase load offset nodes and the three-phase improper wiring nodes;

[0072] The abnormal node power redistribution module is used to reconstruct the power distribution of three-phase power load offset nodes and three-phase power improperly connected nodes, obtain the abnormal node power distribution data, and transmit it to the distribution cabinet management platform to execute the power distribution task.

[0073] The power system operation and maintenance system of the intelligent distribution cabinet for weak current control of the present invention can realize any power system operation and maintenance method of the intelligent distribution cabinet for weak current control of the present invention, and is used to combine the operation and signal transmission medium between each module to complete the power system operation and maintenance method of the intelligent distribution cabinet for weak current control. The internal modules of the system cooperate with each other, thereby improving the weak current control efficiency and overall energy efficiency of the distribution cabinet. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Other features, objects and advantages of the present invention will become more apparent upon reading the detailed description of non-limiting embodiments thereof made with reference to the following drawings:

[0075] Figure 1 A schematic flow chart of the steps of the power system operation and maintenance method of the intelligent power distribution cabinet for weak current control of the present invention;

[0076] Figure 2 Detailed step flow diagram of step S1 in the present invention;

[0077] 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

[0078] The following is a clear and complete description of the technical method of the present invention in conjunction with the accompanying drawings. It is obvious that the embodiments described are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts are within the scope of protection of the present invention.

[0079] In addition, the accompanying drawings are merely schematic illustrations of the present invention and are not necessarily drawn to scale. Identical reference numerals in the figures denote identical or similar parts, and thus repetitive descriptions thereof will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities that do not necessarily correspond to physically or logically separate entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor and / or microcontroller approaches.

[0080] To achieve this, please refer to Figures 1 to 2 The present invention provides a power system operation and maintenance method for an intelligent power distribution cabinet for weak current control, the method comprising the following steps:

[0081] Step S1: Obtain power operation data and sensor data of the distribution cabinet, and perform power system topology analysis based on the power operation data of the distribution cabinet to obtain a dynamic topology of the power system;

[0082] In this embodiment, a three-phase electric energy metering module (such as a DTSD1352 meter supporting the Modbus RTU protocol) and multiple sensor nodes (including current, voltage, temperature, humidity, etc.) installed inside the distribution cabinet are used to collect real-time data on the electrical operating status of the distribution cabinet and device sensor data every 5 seconds. The data is aggregated and processed by an edge computing unit (such as a Raspberry Pi CM4 equipped with an EMQX lightweight MQTT server), and outlier removal, time alignment, and multi-source synchronization are completed through a data preprocessing module. Subsequently, a preliminary connection diagram is constructed based on the current data, and the branch edges are screened using a set current direction threshold q = 1.2A. Based on the connection connectivity conditions, a graph neural network (GraphSAGE) is used in combination with the voltage distribution level to complete the topological node hierarchical reasoning. The topology structure is optimized through three rounds of iterations, and the final output is the dynamic topology structure diagram data of the power system including busbars, branches, and load terminals (encoded using an adjacency matrix method) as the structural basis input for subsequent steps.

[0083] Step S2: performing power system energy efficiency pattern recognition on the distribution cabinet sensor data through the dynamic topology structure of the power system to obtain power system energy efficiency pattern data; performing potential load anomaly analysis on the power system energy efficiency pattern data to obtain power system load abnormality node data;

[0084] In this embodiment, the constructed dynamic topology of the power system is used as a structural constraint. Relying on the time window sliding sequence in the distribution cabinet sensor data (the window width is set to 15 minutes and the step length is 5 minutes), the energy efficiency indicators of each node (unit power consumption, current fluctuation rate, phase sequence stability, etc.) are structured and spatiotemporally aligned. Subsequently, an energy efficiency feature vector is constructed for each topological node, and a pattern recognition algorithm based on spectral clustering is used to divide the nodes into typical energy efficiency pattern clusters (such as "high load and low fluctuation", "periodic oscillation type", etc.). For each type of pattern, a center vector is constructed and the residual score between each node and the center of the cluster to which it belongs is calculated. If the node residual score exceeds 0.75 (normalized score), it is marked as a potential load anomaly candidate node; then, combined with the load fluctuation curve slope change threshold s ≥ 20% in the past 24 hours, a secondary confirmation is performed, and finally, the load anomaly nodes that meet the dual conditions of structural deviation and behavioral fluctuation are extracted, providing target objects for the next stage of abnormal factor analysis.

[0085] Step S3: Analyze the load anomaly dominant factors based on the power system load anomaly node data, and divide the power system dynamic topology into load anomaly impact structures according to the load anomaly dominant factors to obtain load anomaly impact nodes;

[0086] In this embodiment, the path of the obtained load abnormality node is traced back in the dynamic topology, the maximum traceback depth is set to 3 hops, and all its upstream and downstream associated nodes and branches are extracted by the Dijkstra algorithm, and a topological path graph is generated. Subsequently, the path graph nodes and their energy efficiency index vectors are spliced into a topology-energy efficiency composite feature vector matrix, and the matrix is trained using GCN (Graph Convolutional Network), with the number of training rounds set to 200 rounds and the learning rate set to 0.001, to extract the abnormal propagation structure influencing factor vector. At the same time, the historical load data and multidimensional sensor data (such as temperature rise and current distortion rate) of the distribution cabinet in the past 48 hours are retrieved, and the correlation between these physical indicators and the abnormal annotation is calculated by the maximum information coefficient (MIC). The significance threshold is set to 0.05, and the highly correlated physical factors are screened out. The weights are normalized and assigned according to their influence to generate the load abnormality physical factor weight vector. Finally, the structural influencing factors are fused with the physical factors, PCA compression is performed and more than 85% of the cumulative variance is retained. The abnormal factor spectrum is output, and the contribution rate threshold θ = 0.15 is set to extract the dominant abnormal factor of each abnormal node.

[0087] Step S4: Evaluate the three-phase current balance of the distribution cabinet for nodes affected by abnormal loads based on the distribution cabinet sensor data, and classify abnormal nodes based on the current balance of the distribution cabinet to obtain three-phase load offset nodes and three-phase improper connection nodes;

[0088] In this embodiment, the three-phase current data of the target node affected by the load anomaly within the last 30 minutes is extracted from the distribution cabinet sensor data. The sampling period is 2 seconds and the sliding window length is 60 seconds. The period segmentation and missing compensation (using linear interpolation) are performed. The distortion points are then removed through wavelet noise reduction processing to construct a three-phase time series waveform tensor (dimension 60×3). In the feature extraction stage, the offset rate between the three-phase average currents, the phase sequence inconsistency rate (based on time delay cross-correlation analysis), and the imbalance fluctuation rate represented by the coefficient of variation are calculated respectively, and the three are combined into a current imbalance feature vector. Based on the dynamic topology of the power system, the adjacent propagation strength of each node is calculated. The node is simulated for influence diffusion using the current tension propagation weight matrix (normalized tension index λ = 0.6). Based on the propagation results, the current balance index CBI is calculated (formula CBI = 1-||propagation tensor|| / θ, θ is 1.0). Finally, the balance level is classified according to the CBI value, where CBI < 0.7 indicates a severely unbalanced node. The system combines phase-to-phase connection methods with standard phase sequence rules to perform wiring consistency matching analysis. Any phase sequence inconsistencies or cross-connections are flagged as wiring anomalies. Finally, the abnormal node type is classified based on the following rule set: imbalance only → load offset, wiring anomaly → wiring anomaly, and both → wiring anomaly as the primary cause.

[0089] Step S5: Reconstruct the power distribution of the three-phase load offset nodes and the three-phase improperly connected nodes to obtain the abnormal node power distribution data, and transmit it to the distribution cabinet management platform to execute the power distribution task.

[0090] In this embodiment, the identified "three-phase load offset nodes" and "three-phase improper wiring nodes" are stored separately, and a binding relationship is established with their topological positions and branch load data. For offset nodes, a current redistribution model based on linear programming is used, and the current adjustment target of the branch where it is located is set to the three-phase balance target (error ≤ ± 5%), and the adjacent backup branch access plan is used as a regulating variable, and the maximum phase-to-phase offset correction amplitude is set to 10A. For improperly wired nodes, the construction wiring drawings and phase sequence identification records are retrieved, and a wiring replacement recommendation list (such as swapping AC) is automatically generated and pushed to the platform operation interface. The platform sends the task data to the on-site intelligent circuit breaker control unit based on the SCADA system interface (such as IEC 61850 or DL / T645), triggering the current limiting switching or phase sequence reconnection operation process. The entire process adopts a control command execution feedback mechanism, setting the response timeout to 3 seconds to ensure the closed-loop execution of the command and the implementation of the allocation strategy, and finally generates the allocation execution status data through the event log for subsequent maintenance and verification.

[0091] Optionally, step S1 specifically includes:

[0092] Step S11: acquiring power operation data and sensor data of the distribution cabinet, and performing data preprocessing on the power operation data and sensor data of the distribution cabinet to obtain the power operation data to be analyzed and the sensor data of the distribution cabinet to be analyzed;

[0093] In this embodiment, the power distribution cabinet power operation data and distribution cabinet sensor data are obtained respectively through a three-phase power metering module (such as the DTSD1352 meter that supports the Modbus RTU protocol) and multiple types of sensor nodes (including current, voltage, temperature, humidity, etc.) installed inside the distribution cabinet. The power operation data is uploaded to the edge processing terminal (model EdgeBox-RK3399) via the RS-485 bus. The sampling frequency is set to 1Hz and the data window length is 60 seconds. The environmental sensors (temperature and humidity, door magnet, smoke detector) have a sampling period of 5 seconds. After all data enters the data receiving module, timestamp alignment is first performed, and the maximum time offset correction threshold is set to ±1 second. Then, the Z-score normalization method is used to normalize the voltage and current values, and data points exceeding the set abnormal range (such as voltage deviation >±15%) are eliminated. The cubic spline interpolation method is used to fill in the missing segments, and the minimum valid sampling ratio threshold is set to 90%. Sample data below this ratio is not used. The final output format is unified into a structured JSON format, with fields including timestamp, phase, voltage value, current value, sampling quality and other information, which serves as data input for subsequent analysis steps.

[0094] Step S12: extracting the voltage distribution characteristics and current flow characteristics of the distribution cabinet based on the power operation data to be analyzed;

[0095] In this embodiment, for the processed power operation data to be analyzed, three-phase voltage time series and three-phase current time series are constructed respectively. In order to extract the voltage distribution characteristics of the distribution cabinet, a sliding window period of 5 minutes is used to calculate the average voltage value and standard deviation of each phase in each window as part of the node voltage feature vector; at the same time, the maximum offset ΔV between the phase voltages is counted, and if ΔV>20V, it is marked as a voltage unbalanced node. For the current flow direction characteristics, the flow direction is judged by the positive and negative directions of the current phase angle, and the directional threshold q=1.5A is set, that is, when the current difference between a node and the adjacent node is greater than the threshold and the direction is consistent, a directed edge is constructed. The connection relationship between all nodes that meet the conditions is formed into an initial flow direction matrix and encoded in a weighted directed graph (the edge weight is the current amplitude) for subsequent topological reasoning and analysis.

[0096] Step S13: performing a power system backbone topology analysis based on the voltage distribution characteristics of the power distribution cabinet to obtain a power system backbone topology structure;

[0097] In this embodiment, based on the obtained voltage distribution characteristics, bus level determination and trunk path extraction are implemented at this stage. First, the voltage level threshold interval [400V, 220V] is set, and the nodes with voltage values within this interval are used as trunk candidate nodes. Subsequently, a bus-load connectivity graph is constructed, and the depth-first search (DFS) algorithm is used to identify the connectivity of all nodes in the graph, retaining only the largest connected subgraph part that meets the trunk conditions. The subgraph is topologically sorted, the relative potential difference between the nodes is calculated, and the node group with a potential difference of less than 15V is selected as the trunk cluster. Finally, the bus node set in the trunk structure and the corresponding branch connection relationship are output to form a preliminary power system trunk topology structure to assist the next step of flow reasoning.

[0098] Step S14: using the current flow characteristics of the distribution cabinet to complete the power system global topology of the power system backbone topology, thereby obtaining the power system topology;

[0099] In this embodiment, based on the backbone topology, the current flow weighted graph established in the process is used to perform global topology completion analysis. First, all backbone nodes and edges are extracted, and node alignment and edge mapping operations are performed with the current flow graph. If it is found that there is an edge with obvious current transmission but missing connection in the backbone structure (current amplitude > 1.5A, voltage difference < 5V), it is marked as an edge to be completed. A graph structure reasoning model (such as a GAT-based edge prediction network) is used to score the existence of these candidate edges, and the prediction confidence threshold is set to 0.7. Edges that exceed the threshold are included in the topology structure. The newly added edges are uniformly encoded, and the connectivity and edge weights of the nodes are recalculated. The updated global topology is jointly expressed using the adjacency matrix + edge weight matrix. In order to ensure the rationality of the structure, a topology closed-loop verification must be performed to ensure that all newly added branches do not introduce redundant loops or isolated nodes, and ultimately form a structurally complete and connected power system topology.

[0100] Step S15: performing power flow calculation on the power system topology, and constructing a dynamic power system topology according to the power flow calculation results.

[0101] In this embodiment, after the construction of the power system topology is completed, the Newton-Raphson method is used to calculate the power flow of the entire topology. The specific process includes setting the initial value of the bus voltage to 400V, the phase angle to 0°, and the power flow calculation accuracy to an error threshold of ε=10-3. The power injection, branch current, and voltage drop of each node are solved by the power flow equation. In the calculation, the load model adopts a constant impedance model, and the line parameters are obtained by the actual measured values of the sensor (such as branch resistance and reactance). The actual power flow calculation is implemented in the MATPOWER (7.1 version) environment. After the calculation is completed, the results are mapped back to the original topology to generate a topology attribute matrix containing real-time active power, reactive power, voltage phase angle, current amplitude and other information. Combined with the power flow results, the load status of each node is classified (light load / heavy load / overload), and the voltage power flow diagram and current transmission diagram are constructed at the same time. As the final expression of the dynamic topology of the power system, it provides basic support for subsequent energy efficiency analysis and anomaly detection.

[0102] Optionally, step S13 is specifically as follows:

[0103] Step S131: classify the voltage levels according to the voltage distribution characteristics of the power distribution cabinet, set the voltage threshold interval to [400V, 220V] to partition the bus nodes, and construct a bus hierarchical structure to obtain bus hierarchical data;

[0104] In this embodiment, the three-phase voltage mean extracted from the power operation data of the distribution cabinet is used as the basis for the voltage distribution characteristics of the node, and a fixed voltage threshold interval [400V, 220V] is used to classify the busbar hierarchy. Among them, nodes with a voltage greater than or equal to 380V are classified as high-voltage busbars, and nodes with a voltage between 220V and 380V are classified as low-voltage busbars. The K-means++ clustering algorithm is used to cluster the node voltages into two categories, with the initial center values set to 420V and 230V, and the iterative convergence threshold set to 0.001. According to the clustering results, the level to which each node belongs is marked, and a busbar hierarchy mapping table is formed. Further, a connection weight matrix is established for adjacent nodes in the same level to calculate the internal connectivity between busbar levels. At the same time, the node number, corresponding voltage level and hierarchical relationship are recorded to form a structured busbar hierarchical data file for the next step of branch screening and topology construction.

[0105] Step S132: Calculate the potential difference between different bus levels based on the bus level data, select lines with a potential difference constraint less than or equal to [5V, 20V] as trunk and branch lines, perform topology consistency verification, and generate initial trunk and branch line data;

[0106] In this embodiment, the potential difference between each pair of upper and lower bus nodes is calculated based on the generated bus level data. The absolute voltage difference model ΔV = |V_upper-V_lower| is used, and the screening threshold interval is set to [5V, 20V], that is, only the connection lines with potential differences within this range are retained as trunk branch candidates. During the calculation process, for each pair of reachable inter-layer node paths, if the voltage difference meets the set constraints and the line impedance between them is less than 0.5Ω, it is determined to be a valid connection. All candidate branch information (starting point number, end point number, potential difference, resistance, current directionality) is uniformly stored in the candidate branch data set. Subsequently, a graph traversal algorithm is used to verify the consistency of the branch set, eliminate invalid paths that form loops or repeated connections, and retain the branch combination with the strongest structural connectivity. Finally, the initial trunk branch data table is output for subsequent construction of the trunk graph.

[0107] Step S133: constructing a preliminary trunk topology graph with the busbar nodes in the busbar level data as vertices and the initial trunk-branch data as edges, and performing connectivity analysis to remove isolated nodes to generate a preliminary trunk topology structure;

[0108] In this embodiment, the filtered trunk branches are used as the edge information of the graph, and the nodes in the bus level data are used as the vertices of the graph to construct a preliminary trunk topology graph. The adjacency matrix is used to represent the graph structure, and the connectivity analysis of the graph is performed. The breadth-first search (BFS) algorithm is used to start from the high-voltage bus node and mark all reachable nodes. For nodes that are not traversed in the connectivity graph, they are defined as isolated nodes and are removed. The removal criterion is that the sum of the node out-degree and in-degree is less than 1. The trunk part retained in the graph is normalized and renumbered, and the node ID is uniformly converted to a structured number starting with M (such as M001, M002...) to ensure the data readability and system accessibility of the topological structure. Finally, a preliminary trunk topology structure diagram and a structured table are generated, and the fields include topological meta-information such as node number, adjacent nodes, connection voltage difference, and current amplitude.

[0109] Step S134: Analyze the power flow direction based on the voltage distribution characteristics and current flow characteristics of the distribution cabinet, and adjust the branch weights of the preliminary trunk topology according to the power flow direction to obtain an optimized trunk topology;

[0110] In this embodiment, based on the preliminary trunk topology, the power flow direction analysis is performed on each trunk branch in combination with the voltage distribution characteristics and current flow characteristics. The power direction determination formula P = VIcosθ is adopted, the direction is determined by the positive or negative power flow, and the minimum effective power threshold is set to 50W. If the current direction of two adjacent nodes is consistent, the voltage attenuation is reasonable (not exceeding 10%), and the power flow is positive, then the branch is identified as a normal power flow branch. A power direction identifier is established for all branches, and the directional weight of the branch is calculated based on the power flow direction and amplitude. The formula is W = αP + βI, where α = 0.7 and β = 0.3. The updated weight value is written into the topology edge attribute for branch priority sorting. Finally, the optimized trunk topology structure diagram is output. On the basis of keeping the node connection logic unchanged, the dynamic adjustment of the branch direction weight is realized, which is convenient for subsequent optimization of path selection in power dispatching or simulation.

[0111] Step S135: performing topological mapping on the optimized backbone topology structure to obtain the power system backbone topology structure.

[0112] In this embodiment, on the basis of obtaining the optimized backbone topology structure, topology mapping is implemented to obtain the standardized power system backbone topology structure. First, each node in the optimized structure is mapped to a physical node according to the actual physical position and device number, and a one-to-one correspondence table between the logical node and the device address is established. GIS coordinates are used to constrain the position of the nodes, and the coordinate error tolerance threshold is ±2 meters. Weighted shortest path correction is used for nodes with position offsets. Afterwards, a topology closed-loop consistency check is performed, including three verification criteria: branch direction rationality, current closure check, and load power supply continuity. If there are breakpoints, load hanging, or loop abnormalities in the topology structure, it will automatically roll back to the previous step structure version and issue an alarm. The final output power system backbone topology structure includes a node topology table, a standardized connectivity diagram, a GIS spatial location diagram, and a topology attribute configuration file. The format is unified in XML and JSON dual formats, and supports importing mainstream power system modeling and simulation platforms such as DIgSILENT or CYME for system-level deployment and testing.

[0113] Optionally, step S14 is specifically as follows:

[0114] Step S141: performing weighted directed graph modeling on the current flow characteristics of the power distribution cabinet, setting the current direction threshold to [0.3A, 3.0A], thereby constructing a current flow association graph;

[0115] In this embodiment, the current directionality characteristics of each line in the distribution cabinet sensor data are extracted, and a weighted directed graph model is constructed using the current flow direction identifier (forward / reverse) and amplitude information. To ensure the validity of the modeling data, the current directionality threshold range is set to [0.3A, 3.0A], that is, only the lines within this interval are included in the modeling. For all node pairs that meet the conditions, the node number is used as the graph vertex, the current direction as the directed edge direction, and the current amplitude as the edge weight, to construct a current flow association graph G(I) = (V, E). The Python NetworkX library is used for graph modeling and weight configuration, and the edge weight is standardized to the interval [0, 1]. The graph structure does not contain isolated nodes, and all nodes must have at least one incoming or outgoing edge to ensure connectivity. The final output current flow association graph file includes fields such as node ID, current direction, edge weight, etc., and supports importing the graph reasoning engine for structure completion analysis.

[0116] Step S142: performing node alignment and edge mapping analysis on the current flow association graph and the power system backbone topology structure, identifying missing connection parts in the power system backbone topology structure, and obtaining the backbone structure node edge data to be completed;

[0117] In this embodiment, the constructed current flow association diagram and the generated power system backbone topology are read, and node alignment and edge mapping analysis are performed. Double comparison is performed by node numbering and physical location (GIS coordinates), allowing a maximum numbering error of ±2 digits and a maximum coordinate deviation of ±1.5 meters to achieve accurate alignment of the nodes in the diagram. After alignment, an edge overlap analysis strategy is used to detect the connection relationship in the backbone topology that is different from the current diagram, especially to identify the node connections that are missing in the backbone topology but exist in the current flow diagram. The basis for judging the difference of each edge includes whether there is a connection between the nodes, whether the connection direction is consistent, and whether the edge weight difference exceeds 20%. The node edge data table to be completed in the backbone structure is finally output, and the fields include: starting node, end node, current direction, consistency score, etc., which are used for subsequent graph reasoning and inference completion.

[0118] Step S143: Perform graph structure reasoning on the edge data of the nodes to be completed in the backbone structure, perform missing connection completion, and obtain global topology prediction data;

[0119] In this embodiment, according to the node edge data to be completed in the identified backbone structure, a graph neural network (GNN) is used to perform graph structure reasoning. The GraphSAGE model is used for training, and the input data includes node features (such as voltage, current value, node degree) and edge features (such as current directionality, current difference). The training set is the known backbone structure part, and the edges to be completed are used as test sets for edge existence prediction. The sampling batch is set to 128 edges, the learning rate is 0.001, the number of training rounds is 500 rounds, and the Adam optimizer is used. The edge prediction output result is the edge existence probability, and the completion threshold is set to 0.7, that is, the edges with a predicted probability greater than 0.7 will be included in the structure as completed edges. The prediction result output is global topology prediction data, including the starting point, end point, edge probability and prediction source of the newly added connection edge, in JSON and CSV dual formats for structural fusion.

[0120] Step S144: performing edge weight fusion and node uniform coding on the global topology prediction data and the power system backbone topology structure to obtain a preliminary power system topology structure;

[0121] In this embodiment, the global topology prediction data is read and fused with the original power system backbone topology. First, the edges in the two types of topologies are merged, and the nodes are recoded using a unified node numbering rule (prefix "M" + 6-digit number), and the edge weights of the newly added edges are fused with the weights of the original backbone edges by weighted average. The fusion weight is calculated as follows: W_new = αW_existing + (1-α)W_predicted, where α = 0.6. If the newly added edge conflicts with the original edge (inconsistent direction or repeated node ID), the predicted edge is used as the main factor for weight update and edge redirection. Finally, a preliminary power system topology diagram is formed, including a complete node connection relationship, a weight matrix and a unified coding table. The output results support DOT and GEXF formats for the next consistency check.

[0122] Step S145: performing consistency check and closed-loop verification on the preliminary power system topology structure to obtain the power system topology structure.

[0123] In this embodiment, the generated power system topology is subjected to consistency verification, which mainly includes three types of verification: 1) node connectivity verification, ensuring that there are no isolated nodes and dangling branches in the topology; 2) edge direction consistency verification, judging whether the power flow direction matches the current direction, and allowing a maximum deviation angle of ±15°; 3) loop closure verification, detecting whether there are illegal passive closed loops. A custom topology verification engine is used to execute the above logic rules. If an abnormal connection is detected, it is marked as a red edge and an error report is output. After passing all the verifications, a closed-loop verification simulation is performed, and the voltage and current distribution are tested using a power flow calculation model (based on the Newton-Raphson method). The error standard is set to ±0.01V and ±0.05A. After passing all the verifications, the power system topology file is finally output, which contains data such as the topology structure, edge weights, and power flow verification results. The file is output in XML and graph database formats for system deployment.

[0124] Optionally, step S2 is specifically:

[0125] Step S21: performing spatial mapping and time window slicing on the distribution cabinet sensor data to be analyzed, and performing time window sensor-topology node binding based on the node-branch relationship in the dynamic topology structure of the power system to obtain a spatiotemporal feature map of the power system nodes;

[0126] In this embodiment, the sensor data of the distribution cabinet to be analyzed is spatially mapped, and a one-to-one mapping table between the sensor and the node is constructed based on the physical connection relationship between each node and the branch in the dynamic topology of the power system. The mapping relationship must meet the constraint conditions that the spatial distance of the sensor is less than 1.5 meters and the signal is unique. Subsequently, the sensor data is segmented into a sliding time window according to 15 minutes, and a fixed step window sharding strategy is adopted with a step size of 5 minutes to enhance the overlapping robustness of the time series samples. The data in each time window is bound to its corresponding topological node, and the binding method adopts a key-value pair structure storage (the key is the node ID and the value is the sensor vector in the corresponding time window), thereby constructing a spatiotemporal node data structure. Finally, it is integrated into the spatiotemporal feature map of the power system node, and the output is a multidimensional time series tensor with the dimension of the number of nodes × the number of time windows × the feature dimension (such as voltage, current, active power, and reactive power, a total of 4 channels).

[0127] Step S22: extracting weighted aggregated energy efficiency indicators based on the spatiotemporal characteristic graph of power system nodes to obtain node energy efficiency indicator data;

[0128] In this embodiment, based on the constructed node spatiotemporal feature map, a multi-index weighted aggregation energy efficiency calculation is performed. Specifically, the basic indicators such as voltage, current, active power, reactive power, etc. of each node in all time windows are extracted, and the time-weighted attenuation coefficient (set to γ = 0.85) is introduced to give higher weights to the closer time windows. The formula η = ∑(γ^t×x_t) is used, where x_t is the indicator value at time t, and the node-level weighted energy efficiency vector is calculated for each of the four types of indicators, and the vector dimension is 4. In order to further normalize the dimensional differences of different indicators, the Z-score normalization strategy is used to normalize all node indicators, and the mean and standard deviation are based on the distribution statistics of all nodes. The final output is a node energy efficiency indicator data table, with fields including node ID, voltage factor, active factor, reactive factor, current factor, etc., which serves as the input for the next step of behavioral analysis.

[0129] Step S23: constructing an energy efficiency attribute graph using the node energy efficiency index data, and extracting node energy efficiency behavior characteristics based on the energy efficiency attribute graph;

[0130] In this embodiment, based on the obtained node energy efficiency index data, an energy efficiency attribute graph is constructed using a graph structure modeling method. The graph vertices are node IDs, the graph edges come from the adjacent connection relationships in the dynamic topology structure, and the edge weights are defined according to the Euclidean energy efficiency index distance between the two nodes. The formula is w(i, j) = 1 / (1+d(i, j)), where d(i, j) is the Euclidean distance between nodes i and j in the energy efficiency index vector space. Through this graph structure, a node embedding method (such as Node2Vec, with a walk step length of 10, a walk number of 80, and an embedding dimension of 16) is further applied to generate an embedded representation of the energy efficiency behavior characteristics of each node. This feature vector reflects the node's hybrid representation capability in topological association and energy efficiency characteristics, and ultimately forms a node energy efficiency behavior feature set for clustering analysis and subsequent anomaly identification.

[0131] Step S24: clustering energy efficiency behaviors according to the node energy efficiency behavior characteristics to obtain power system energy efficiency model data;

[0132] In this embodiment, the generated node energy efficiency behavior feature vector is used for unsupervised cluster analysis. The K-means algorithm is selected for clustering, and the initial cluster number K=5 is set. The number of clusters is dynamically adjusted through silhouette coefficient analysis. The maximum silhouette coefficient (greater than 0.72) is obtained when the final number of clusters stabilizes at K=4. The cluster feature space is iteratively calculated based on the Euclidean distance, the maximum number of iterations is set to 300, and the convergence threshold ε is 0.001. After clustering is completed, each cluster represents a stable energy efficiency mode, which contains a group of topological nodes with similar behaviors. The output result is a power system energy efficiency mode data table, which contains information such as node ID, cluster ID, energy efficiency center vector, and time period, which is used for the next step of anomaly detection.

[0133] Step S25: perform abnormal factor detection and pattern deviation analysis on the power system energy efficiency pattern data, extract potential load abnormality candidate nodes, and perform confidence scoring on the potential load abnormality candidate nodes to obtain power system load abnormality node data.

[0134] In this embodiment, the energy efficiency pattern data is detected for outlier factors using the LOF (Local Outlier Factor) method. The number of neighbors is set to k = 20, and the deviation of the local density of each node is calculated. Nodes with LOF values higher than 2.0 are initially considered potential outliers. Subsequently, the node offset is calculated for each cluster center. If the offset exceeds twice the standard deviation within the cluster, it is also marked as a pattern deviation node. The union of the above two types of results is taken as a potential load anomaly candidate node. A confidence scoring mechanism is further introduced. The scoring items include the degree of local density deviation, offset distance, and the frequency of anomalies in the time window to which it belongs. A weighted comprehensive method is used for scoring, with weights of 0.5, 0.3, and 0.2 respectively. The scoring threshold is set to 0.65, and nodes with scores higher than the threshold are marked as final power system load anomaly nodes. The output result is an abnormal node data table with fields including node ID, anomaly score, scoring details, and tag type.

[0135] Optionally, step S25 is specifically as follows:

[0136] Step S251: constructing an energy efficiency behavior residual matrix of each node based on the characteristic vector of each cluster center in the power system energy efficiency model data, and calculating the Euclidean distance between the node energy efficiency behavior characteristics and the energy efficiency behavior residual matrix of the corresponding node to obtain node energy efficiency behavior residual data;

[0137] In this embodiment, the energy efficiency feature vectors (such as voltage, active power, reactive power, and current, totaling four dimensions) of each cluster center in the power system energy efficiency pattern data are extracted as representative behavior templates. For each node to be analyzed, the difference between its true energy efficiency behavior feature vector and the cluster center vector to which it belongs is calculated to construct the node energy efficiency behavior residual vector. Furthermore, a residual matrix is constructed in the residual space with nodes as samples and feature dimensions as axes, and the matrix shape is the number of nodes × 4. The Euclidean distance formula is used to evaluate the residual amplitude of the node vector and its residual vector, and the energy efficiency behavior residual data of each node is obtained after normalization. This data reflects the degree of deviation of the node outside the cluster behavior, and provides a quantitative basis for subsequent slope fitting and mutation detection.

[0138] Step S252: performing slope fitting and mutation detection based on the node energy efficiency behavior residual data, calculating the abnormal factor score, and obtaining abnormal factor score data; setting the abnormal score dynamic threshold range according to the abnormal factor score data;

[0139] In this embodiment, based on the obtained residual data of node energy efficiency behavior, the residual time series of each node is selected for linear slope fitting, and the least squares method is used to calculate the residual trend slope of each node. The slope sign and numerical value are used to identify the direction and intensity of energy efficiency deviation. For node residual sequences with significant mutations (for example, when the slope is greater than 0.25 or less than -0.25), the Pelt algorithm is introduced to detect mutation points, and the algorithm sets the penalty parameter β to 1.5. The abnormal factor score of each node is obtained by weighted synthesis of three factors: slope strength, mutation point frequency and residual mean, and the weights are set to 0.4, 0.4 and 0.2 respectively. Finally, the abnormal factor score data is output. Based on the score distribution, a dynamic threshold interval for abnormal scores is adaptively constructed, and the upper and lower limit intervals are calculated using the 95% quantile method.

[0140] Step S253: Evaluate the energy efficiency behavior coordination degree of adjacent nodes in the dynamic topology structure of the power system to obtain the energy efficiency behavior coordination degree of the adjacent nodes;

[0141] In this embodiment, combined with the dynamic topology of the power system, adjacent node pairs with physical connection relationships are screened out, and the statistical number is greater than 800 pairs. For each pair of nodes, the energy efficiency behavior feature vector is extracted, and the cosine similarity between the two is calculated as a quantitative indicator of the energy efficiency behavior coordination. If the coordination degree is greater than 0.85, it is considered to be highly coordinated, and if it is lower than 0.6, it is considered to be inconsistent in behavior. Considering the time evolution characteristics, it is also necessary to compare the behavioral coordination change trend in the past 3 hours. If the short-term coordination fluctuation is less than 0.05, it is marked as a stable coordination pair. The output result is the coordination matrix of adjacent node pairs, in which each item records the node pair ID, current coordination value, and historical coordination fluctuation value, which is used for subsequent abnormal distribution map construction.

[0142] Step S254: combining the anomaly score dynamic threshold interval and the adjacent node energy efficiency behavior coordination degree, constructing a weighted anomaly distribution map, identifying and extracting high anomaly factor score nodes, and obtaining potential load anomaly candidate node data;

[0143] In this embodiment, the anomaly factor score is integrated with the energy efficiency synergy to construct a weighted anomaly distribution graph. The graph vertices are topological nodes, the edges are synergy associations, the node attributes are anomaly scores, and the edge attributes are synergy weights. During the graph construction process, the node anomaly score weighting factor is set to 0.7, the edge synergy factor is set to 0.3, and the GAT (graph attention mechanism) neural network model is used for feature propagation modeling to learn the key paths for anomaly propagation in the graph. During the propagation process, node clusters with high scores are identified, and areas with dense anomaly distribution are extracted. Nodes with anomaly factor scores higher than the upper limit of the dynamic threshold are set as preliminary candidate nodes. After the anomaly factor score data is finally output, the dynamic threshold interval is constructed based on its overall distribution using the quantile method. Specifically, the distribution of all node anomaly score data is statistically analyzed, and the confidence level is set to 95%. The upper threshold of the score distribution is calculated as the 95% quantile (P95), and the lower threshold is the 5% quantile (P5), which constitute the dynamic threshold interval. For example, when the P95 value of all scores is 0.87, the upper limit of the dynamic threshold of the anomaly factor score can be set to 0.87. This threshold is then used as the scoring threshold for preliminary candidate nodes. The final output is the data of potential load anomaly candidate nodes, which includes fields such as node ID, anomaly score, collaborative propagation weight, and neighborhood cluster density.

[0144] Step S255: perform node confidence evaluation on the potential load anomaly candidate node data, and set the scoring confidence threshold to [0.6, 0.9] to screen high-confidence nodes to obtain power system load anomaly node data.

[0145] In this embodiment, confidence scoring is performed on the extracted potential load anomaly candidate nodes. The scoring dimensions include three aspects: (1) the weight of the abnormal factor score is set to 0.5; (2) the weight of the neighborhood collaborative average score is set to 0.3; (3) the weight of the energy efficiency cluster stability is set to 0.2. After the scoring results are normalized to the interval [0,1], all candidate nodes are sorted, and the score confidence threshold interval is set to [0.6,0.9]. Nodes above the confidence lower limit are selected as the final power system load anomaly nodes. The output data includes node ID, confidence score, score composition ratio and abnormal status label, which provides a basis for the next step of load management or alarm distribution.

[0146] Optionally, the analysis load abnormality dominant factor described in step S3 is specifically:

[0147] The path tracing depth is set to [3,5] layers to dynamically trace the upstream and downstream paths of the power system load abnormality node data, construct a load abnormality path graph, and extract the structural position of each abnormal node in the dynamic topology of the power system and the correlation degree of adjacent nodes to obtain the load abnormality topology path feature data;

[0148] In this embodiment, the path tracing depth is set to [3,5] layers. Based on the directed edge relationship of the dynamic topology of the power system, a traversal and tracing operation of upstream and downstream nodes is performed on each load abnormality node to identify the structural path nodes and branches that are directly or indirectly connected to the abnormal node within the tracing depth range. The distance weighted calculation of different paths is performed using the Dijkstra algorithm, and the path directionality is determined in combination with the branch current flow direction to construct a load abnormality path graph dominated by abnormal nodes. On this basis, the structural hierarchy, upstream and downstream distribution density, number of adjacent nodes, and connection edge weight distribution of the abnormal nodes in the topology are extracted to generate the load abnormality topology path feature data corresponding to each abnormal node, which is used to characterize its abnormal propagation potential and structural influence in the power grid.

[0149] Based on the load anomaly topology path feature data and combined with the node energy efficiency index data, a topology structure-energy efficiency composite feature vector matrix is constructed;

[0150] In this embodiment, the obtained topological path feature data is bound one by one to the energy efficiency index data corresponding to the power system node (including node power factor, unit power consumption, load utilization, etc.), and a structure-energy efficiency composite feature vector matrix is constructed by feature vector splicing. Each row of the matrix represents an abnormal node, and the columns it contains include 28 composite features, including topological path depth, path connectivity, node current variance, voltage stability index, energy efficiency utilization, etc. In order to enhance the robustness of subsequent training, the feature matrix is subjected to Z-score normalization processing so that its mean is 0 and its variance is 1, so as to enhance the sensitivity to various influencing factors.

[0151] Perform structural impact factor embedding training on the topology-energy efficiency composite feature vector matrix to generate load abnormal structure impact feature data;

[0152] In this embodiment, the constructed topology-energy efficiency composite feature vector matrix is input into a structural factor impact modeling network based on graph embedding. The network structure adopts a two-layer GCN (Graph Convolutional Network) with a residual connection mechanism, and uses node anomaly labels as weak supervision signals. Training is performed through a loss function that combines the node cluster embedding center with the node distribution similarity. The number of training iterations is set to 500, and the learning rate is 0.001. The interaction between the structural correlation and energy efficiency features between nodes is modeled during the graph convolution process, thereby extracting the load anomaly structural impact features of each node, and obtaining a load anomaly structural impact feature vector with a dimension of 16.

[0153] Obtain the historical load fluctuation sequence and multi-dimensional sensor data of the distribution cabinet, and combine it with the abnormal load node data of the power system to perform characteristic factor correlation analysis to obtain the weight data of the abnormal load physical factor;

[0154] In this embodiment, the distribution cabinet management platform is used to obtain the 180-day history of the distribution cabinet load power fluctuation sequence of the abnormal node and the multidimensional sensor data of five types of load fluctuations, including temperature, voltage, current, harmonics, and power factor, with a collection frequency of 10 minutes. This multidimensional time series data is matched with the abnormal node, and the Pearson correlation coefficient, mutual information entropy value, and Granger causality test results between each physical factor and the abnormality score are calculated to establish a factor correlation matrix. The factor correlation matrix is fused using a weighted average method to generate the average impact intensity score for each physical factor, and finally, the load abnormality physical factor weight data containing 12 types of factors is constructed.

[0155] The load anomaly structural impact feature data and load anomaly physical factor weight data are integrated to perform multidimensional factor compression and reduction, retaining factor combinations with a cumulative contribution rate threshold greater than or equal to 0.85 to obtain the load anomaly impact factor spectrum;

[0156] In this example, the obtained structural impact feature data and the obtained physical factor weight data are multi-dimensionally spliced, and the fused features are compressed using principal component analysis (PCA). The cumulative contribution rate retention threshold is set to 0.85, that is, the first several principal components are selected so that their cumulative variance contribution rate is not less than 85% to ensure the validity and representativeness of the abnormal factor characteristics. In the final load abnormality impact factor spectrum, each factor corresponds to a compressed principal component coefficient and contribution rate.

[0157] Based on the load anomaly influencing factor spectrum, significance ranking and dominant factor identification are performed. The factor contribution rate threshold is set to 0.15 to extract the dominant factors of abnormal nodes and generate the load anomaly dominant factors.

[0158] In this example, the resulting anomaly impact factor spectra are sorted in descending order by factor contribution rate, and key factors with a contribution rate greater than 0.15 are extracted as dominant factors. These factors include core variables such as topological path connectivity, load fluctuation sensitivity, energy efficiency utilization disturbance intensity, and node current mean square error (MSE). Ultimately, through logical judgment rules, these dominant factors are used as the basis for determining various types of load anomalies. The most important causal characteristics for each abnormal node are output, forming a dataset of dominant factors for power system load anomalies.

[0159] Optionally, the three-phase current balance evaluation of the power distribution cabinet described in step S4 is specifically as follows:

[0160] Extract the three-phase current time series data of the load abnormality-affecting node from the distribution cabinet sensor data, perform period segmentation, missing compensation, and distortion point cleaning on the three-phase current time series data, and obtain the calibrated three-phase current standardized time series data;

[0161] In this embodiment, the distribution nodes identified as affected in the load anomaly detection phase are extracted, and the historical data of the three-phase current sensors of the corresponding distribution cabinets are called, and a continuous three-phase current time series is divided according to a time granularity of 5 minutes. For the extracted data sequence, a variable-period Fourier window analysis is used to segment its periodic characteristics, and the length of each segment is set to a complete power frequency cycle (i.e., the number of sampling points in 20ms is 160 points, and the sampling frequency is 8kHz). For windows with missing data, a spline function based on multidimensional local interpolation is used to compensate for the missing data, and the interpolation range is limited to ±2 sampling points. Subsequently, the current waveform in each periodic segment is applied with a combined method of sliding median difference and gradient boundary detection to identify distortion points, clean high-frequency distortion values such as arc disturbances and inductive disturbances, and obtain periodically complete and distortion-free three-phase current standardized time series data.

[0162] A three-phase current time-domain waveform tensor model is constructed based on the calibrated three-phase current standardized time series data. The three-phase current inter-phase offset rate, phase sequence inconsistency rate, and time-domain imbalance fluctuation rate are extracted using the three-phase current time-domain waveform tensor model to obtain the three-phase current imbalance characteristic data.

[0163] In this embodiment, the obtained three-phase current standardized data is organized into a third-order tensor form to construct a three-phase current time domain waveform tensor model I∈R3×N×T, where 3 represents the three phases A, B, and C, N is the number of period segments, and T is the number of sampling points per period. The phase difference analysis of the tensor model is performed, and the relative mean square error (RMSE) is used to calculate the phase offset rate. The threshold is set to 0.15A, and the phase sequence change rate of each cycle waveform is calculated based on the principal component phase alignment method to identify the non-standard ABC phase sequence ratio. The time domain imbalance fluctuation rate is further evaluated by the current variance fluctuation coefficient within the cycle. The three-phase current imbalance feature data set obtained by the above feature extraction is recorded as F imb , including the phase offset rate matrix, phase sequence inconsistency rate vector and fluctuation rate vector.

[0164] Based on the three-phase current imbalance characteristic data and the dynamic topology of the power system, a node-level three-phase current imbalance distribution map is constructed. The current offset tension of each abnormal node is calculated using the node-level three-phase current imbalance distribution map, and the current tension propagation weight matrix is obtained.

[0165] In this embodiment, the corresponding three-phase imbalance characteristic data is injected into each node based on the real-time topology of the power system at the current moment, and a node-level three-phase current imbalance distribution diagram is constructed according to the branch connection direction. In the diagram, the degree of three-phase imbalance along the branch direction is represented by the gradient difference between nodes, and the current offset tension index T is defined using the graph structure tension analysis method. ij =∥F imb,i -F imb,j∥2, traverse all connected edges and construct the current tension propagation weight matrix W between nodes T =[T ij ], where only the tension paths with Tij>0.2 are retained to eliminate the interference of the propagation paths in the stable region.

[0166] The current tension propagation weight matrix and the three-phase current imbalance characteristic data are fused to calculate the three-phase current balance index and obtain the abnormal node current balance data.

[0167] In this embodiment, the current tension propagation weight matrix WT is combined with the node three-phase current imbalance characteristic data F imb Fusion uses a weighted-average structure-aware aggregation algorithm to calculate the three-phase current balance index for each node. The balance index ranges from 0 to 1, with values closer to 1 indicating closer balance. Finally, nodes with currents below a set threshold (perhaps 0.75) are marked as unbalanced. Current balance data for each abnormal node is output as a key evaluation metric for subsequent wiring structure modification and abnormality source tracing.

[0168] More importantly, the calculation of the three-phase current balance index is specifically as follows:

[0169] Normalize the current tension propagation weight matrix, set the propagation attenuation coefficient to 0.8, and generate the normalized current tension propagation matrix;

[0170] In this embodiment, the current tension propagation weight matrix is normalized, and the obtained current tension propagation weight matrix W is T =[T ij ] to perform row-by-row normalization, using the standardization formula The normalized propagation attenuation coefficient α is set to 0.8, indicating that the influence of each level of propagation will decrease by 80%, which is used to simulate the nonlinear attenuation effect of current imbalance along the branch path in the power topology. The matrix W generated after normalization is norm This represents the normalized influence weight of each node on its neighboring nodes in the propagation of three-phase current imbalance, providing a structural foundation for the subsequent propagation of node feature information. To improve processing stability, a minimum retention weight threshold of 0.01 is set. Path influences below this value are truncated, ensuring that anomaly propagation occurs only between related nodes.

[0171] Perform node feature expansion on the three-phase current imbalance feature data to construct a node imbalance feature vector matrix;

[0172] In this embodiment, taking each distribution node as a unit, the three types of eigenvalues corresponding to the three-phase current phase offset rate, phase sequence inconsistency rate and time domain fluctuation rate are expanded into a three-dimensional eigenvector in Represents the average difference between the two-phase currents of node A and B, in amperes; θ i Indicates the maximum phase sequence displacement angle within a cycle, in degrees; σ i is the variance ratio fluctuation rate within the three-phase current cycle. The three-phase imbalance eigenvectors of all nodes are combined to form the node imbalance eigenvector matrix F = [f1,f2,...,f n ] T , where n is the number of candidate anomaly nodes involved in the analysis. To ensure the numerical stability of subsequent propagation modeling, all feature quantities are uniformly normalized using the Z-score, making each feature comparable across different dimensions.

[0173] Based on the normalized current tension propagation matrix and the node imbalance eigenvector matrix, the node imbalance characteristic propagation modeling is performed to obtain the initial imbalance propagation influence matrix;

[0174] In this embodiment, a weighted propagation model is used to model the impact of imbalance between nodes. The propagation model is in the form of F′=W norm F, where F′ is the node eigenvector matrix after propagation, representing the new imbalance perception characteristics formed at each node under the influence of the tension of adjacent nodes. This propagation process can be viewed as a node "sensing" the imbalance state of adjacent nodes in the topology and adaptively correcting its local current balance, reflecting the multi-point coupling nature of three-phase imbalance in actual power systems. During the propagation modeling process, the output eigenvectors of each node are considered to represent the overall imbalance perception influence it receives. Ultimately, they are combined to form the initial imbalance propagation influence matrix F′, which is used in subsequent current balance calculations.

[0175] The node current balance index is calculated based on the initial imbalance propagation influence matrix, the node current balance index of all nodes is truncated and normalized and interval mapped, the balance level division interval is set, and the abnormal node current balance data is generated.

[0176] In this embodiment, for each node eigenvector in the initial unbalanced propagation influence matrix F′, its mean square deviation value in the three-phase dimension is first calculated and recorded as the node initial imbalance index in is the mean of the feature vector after propagation of the node. i The index is globally normalized, mapped to the [0,1] interval, and then reversely processed to generate the final current balance index. Values closer to 1 indicate greater balance. The balance level range is set as [0.9, 1.0] for "high balance," [0.7, 0.9) for "medium balance," [0.5, 0.7) for "low balance," and [0.0, 0.5) for "severe imbalance." Each abnormal node is assigned a corresponding level identifier, and a final abnormal node current balance data table is output for subsequent fault handling and structural reconstruction.

[0177] Optionally, the abnormal node classification in step S4 is specifically:

[0178] The abnormal node current balance data is divided into balance intervals to obtain graded current balance label data;

[0179] In this embodiment, by analyzing the current balance data B i The nodes are divided into four levels according to the set balance level: [0.9, 1.0], [0.7, 0.9), [0.5, 0.7), [0.0, 0.5). The current balance value of each node is matched with its corresponding label according to the set interval standard, and a graded label dataset L is generated. i For example, if the current balance degree of node i is B i =0.92, the node is marked as “high balance” with a label of 1; if B i =0.68, it is marked as "medium balance" level, with a label of 2. This process ensures the applicability of current balance label data under different grid conditions, supporting accurate classification of load anomaly types and subsequent fault diagnosis and processing.

[0180] Based on the dynamic topology structure of the power system, the topological adjacent information and branch connection relationship of the nodes affected by load anomalies are extracted, and the node-phase connection matrix is constructed.

[0181] In this embodiment, the topology information of the power system is used to extract the adjacent nodes and branch connection relationships of the abnormal node. The branch connection distance threshold d is set thresh =3km as the association judgment condition. If the physical distance between two nodes is less than the threshold and they are in the same grid zone, it is considered that there is a direct electrical connection between the two nodes. To this end, the node connection matrix A = [a ij ], where a ij =1 indicates that there is a direct electrical connection between node i and node j, otherwise it is 0. The connection matrix provides important topological constraint information for subsequent structural analysis.

[0182] Combined with the node interphase connection matrix and the graded current balance label data, the three-phase structure distribution of each load abnormality-affected node is analyzed to obtain the three-phase phase sequence consistency characteristic data;

[0183] In this embodiment, according to the node phase connection matrix A and the node current balance degree label data L i , analyze the consistency of the three-phase current of each node on the phase-to-phase line. For each node, set the phase sequence consistency threshold θ thresh =15, if the phase angle deviation of the phase current waveform is greater than the threshold, it is considered that the node has a phase sequence inconsistency problem. Calculate the phase sequence consistency feature C of each node i =max(Δθ i ), where Δθ i is the average value of the phase sequence offset of the node. Finally, based on these characteristic values, the phase sequence consistency feature data of the node is generated for subsequent detection and repair of improper wiring. It is also possible to first count whether all three phases of the node are connected. If only one or two phases are connected, it is marked as "phase missing"; then check whether there are multiple branches connected to the same phase at the same time. If so, it is marked as "phase duplication"; finally, check whether there are cross connections that do not comply with the topology rules (such as phase A connected to phase B branch). Such nodes will be marked as "phase sequence confusion". Through the above structural analysis, three phase sequence consistency features are extracted: phase integrity, phase balance and access standardization, which are represented in vector form to provide structural feature support for subsequent wiring pattern recognition.

[0184] Perform topological structure projection and constraint rule matching on the three-phase phase sequence consistency feature data, perform wiring pattern recognition, and obtain improper wiring identification data;

[0185] In this embodiment, based on the topological structure projection, the three-phase connection mode of the power system is constructed and matched with the known connection rules. For example, assuming that the power system adopts the "star connection" mode, the phase sequence consistency feature C at each node i If the phase sequence pattern does not match the "star connection" pattern, the node is marked as a potential improper connection node. Set the threshold θ for matching the connection rule match =30, when the phase sequence consistency deviation exceeds this threshold, the node will be marked as improperly wired. Through the process of wiring rule matching, a set of improper wiring identification data R is obtained. i , each node is judged to have improper wiring based on its three-phase phase sequence consistency characteristics. Rules may also include but are not limited to: there should be no repeated phase access in a single node, the three-phase access to the same node should be evenly distributed, and there should be no cross-connection between phases. By matching the three-phase structure of each node with the topological path rules item by item, if a node is found to violate any rule, it will be marked as an "improperly wired node". The identification result is output in the form of a Boolean flag (1 for abnormal, 0 for normal), with an accompanying description of the violation type, which is used for subsequent classification decisions and power system maintenance prompts.

[0186] The hierarchical current balance label data and improper wiring identification data are integrated for joint logical judgment, and the load abnormality-affected nodes are divided into three-phase load offset nodes and three-phase improper wiring nodes.

[0187] In this embodiment, a logical combination discrimination model is established based on the two key abnormality indicators mentioned above: the current balance label and the wiring abnormality flag. The rules are as follows: If the wiring structure of a node meets the specifications (the improper wiring flag is 0), but the balance label is slightly or moderately unbalanced, it is judged as a "three-phase load offset node." If the node is identified as improperly wired or its balance label is severely unbalanced, it is marked as a "three-phase improper wiring node."

[0188] More importantly, the joint logic judgment is as follows:

[0189] The hierarchical current balance label data is encoded by label type, and the low balance interval threshold is set to 0.85 and the severe imbalance interval threshold is set to 0.65. Nodes with a current balance higher than 0.85 are classified as normal nodes, nodes with a current balance lower than 0.85 and higher than 0.65 are classified as slightly offset nodes, and nodes with a current balance lower than 0.65 are classified as severely offset nodes. This generates structured current balance status data.

[0190] In this embodiment, the label type encoding is performed for the current balance value of the abnormal load node of the power system. The balance classification standard is set as follows: the current balance value P greater than 0.85 is recorded as a "normal node", that is, the label encoding is 0; the node in the interval (0.65, 0.85] is regarded as a "slightly offset node", the label encoding is 1; the node less than or equal to 0.65 is marked as a "severely offset node", the label encoding is 2. This classification uses the pandas.cut function in Python to segment and encode the interval to ensure the consistency and automation of label generation. Finally, each abnormal node and its corresponding balance status label are combined into a structured table data format, and the fields include: node ID, current balance value, balance level label, which is convenient for subsequent fusion analysis with wiring status information.

[0191] According to the improper wiring identification data, the wiring state type of each load abnormality-affecting node is marked, and the identification variable fi∈{0,1} is set, where fi=1 indicates the presence of improper wiring behavior, to construct a wiring state vector;

[0192] In this embodiment, the set of improperly wired nodes identified in the previous step is used to set an identification variable, fi, to distinguish whether a node has wiring anomalies. The specific operation is as follows: an empty vector corresponding to the number of all abnormal nodes is created, with the initial value fi = 0; the node numbers in the improper wiring identification data are traversed, and if the node appears in the abnormal list, the corresponding fi value is assigned to 1, indicating that the node has a wiring pattern that does not meet standard specifications, such as phase sequence confusion or phase duplication. The wiring state vector is in the form of a one-dimensional array F = [f1, f2, ..., fn], where n is the number of nodes. It can be subsequently integrated with the current balance structured data to achieve logical judgment under multi-dimensional conditions.

[0193] The structured current balance state data and the wiring state vector are combined with logical rules, and the following joint judgment rules are set:

[0194] If the balance level of the node affected by the load anomaly is severely offset and the identification variable of the node is 0, the node is marked as a three-phase electric load offset node;

[0195] If the identification variable of the node affected by abnormal load is 1, the node is marked as a three-phase improperly connected node;

[0196] If the load abnormality affects the node and the balance level is severely offset and the identification variable of the node is 1, the node is marked as a three-phase improperly connected node.

[0197] In this embodiment, a joint logic judgment is performed based on the node's current balance state label (0, 1, 2) and the wiring state vector (0 / 1) to construct an abnormal type identification model. The specific logical rules are set as follows: if the current balance label of a node is 2 (severe offset) and the wiring state fi = 0, it is preliminarily judged as a "three-phase load offset node"; if the wiring state fi = 1 of a node, regardless of the current balance state, it is uniformly marked as a "three-phase improper wiring node"; if the node current balance label is 2 and fi = 1 are met at the same time, then according to the priority processing principle, it is still marked as a "three-phase improper wiring node". This joint logic can be implemented through conditional priority coding, using Python logical judgment statements or building a decision tree model to complete automatic judgment and classification label output. When performing logical judgment, the node label classification operation is performed through a preset logical rule chain. Taking a data table as input, the system verifies the "balance label" and "connection identification variable" for each node row by row, storing them in the variables level and fi, respectively. Then, nested judgment is performed based on the following rules: If level == 2 and fi == 0, the node type is output as "load offset"; if fi == 1, regardless of level, the output type is set to "improper connection"; if both level == 2 and fi == 1 are met, "improper connection" is still output to ensure that the fault identification results cover the most critical issues. Priority decision trees or multi-condition mapping matrices are used in the logical combination judgment process to improve efficiency. Results are output in a table format, including the node number, balance label, connection identification, and final judgment result.

[0198] The judgment logic mapping is performed on the load abnormality impact node to generate the three-phase load offset node and the three-phase improper wiring node.

[0199] In this embodiment, during the final judgment execution phase, each node classification result, derived through logical combination analysis, is converted into a standardized result set. The output is stored in two node lists: List A contains the numbers and status information of all nodes identified as "three-phase load offset nodes"; List B contains the numbers, anomaly types, and wiring feature descriptions of all nodes identified as "three-phase improper wiring nodes." This structured result can be used for subsequent equipment inspections, abnormal node re-inspections, and wiring adjustments within the power supply system. It can also be simultaneously input into the power operation and maintenance management platform for abnormality work order generation.

[0200] Optionally, this specification also provides a power system operation and maintenance system for an intelligent power distribution cabinet for weak current control, which is used to execute the power system operation and maintenance method for the intelligent power distribution cabinet for weak current control as described above. The power system operation and maintenance system for the intelligent power distribution cabinet for weak current control includes:

[0201] The topology analysis module is used to obtain the power operation data and sensor data of the distribution cabinet, and perform power system topology analysis based on the power operation data of the distribution cabinet to obtain the dynamic topology of the power system;

[0202] The potential load anomaly analysis module is used to identify the power system energy efficiency pattern of the distribution cabinet sensor data through the dynamic topology of the power system to obtain the power system energy efficiency pattern data; perform potential load anomaly analysis on the power system energy efficiency pattern data to obtain the power system load anomaly node data;

[0203] The abnormal impact structure division module is used to analyze the load abnormality dominant factors based on the load abnormality node data of the power system, and divide the load abnormality impact structure of the power system dynamic topology structure according to the load abnormality dominant factors to obtain the load abnormality impact nodes;

[0204] The three-phase current balance evaluation module is used to evaluate the three-phase current balance of the distribution cabinet at the nodes affected by abnormal loads based on the distribution cabinet sensor data, and classify abnormal nodes based on the current balance of the distribution cabinet to obtain the three-phase load offset nodes and the three-phase improper wiring nodes;

[0205] The abnormal node power redistribution module is used to reconstruct the power distribution of three-phase power load offset nodes and three-phase power improperly connected nodes, obtain the abnormal node power distribution data, and transmit it to the distribution cabinet management platform to execute the power distribution task.

[0206] The foregoing description is intended only to provide specific embodiments of the present invention, which will enable those skilled in the art to understand and implement the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown herein, but is to be construed in the widest possible manner consistent with the principles and novel features disclosed herein.

Claims

1. A power system operation and maintenance method for an intelligent distribution cabinet for weak current control, characterized in that: The following steps are involved: Step S1: Obtain power operation data and sensor data of the distribution cabinet, and perform power system topology analysis based on the power operation data of the distribution cabinet to obtain a dynamic topology of the power system; Step S2: performing power system energy efficiency pattern recognition on the distribution cabinet sensor data through the power system dynamic topology structure to obtain power system energy efficiency pattern data; Conduct potential load anomaly analysis on the power system energy efficiency model data to obtain power system load anomaly node data; Step S3: Analyze the load anomaly dominant factors based on the power system load anomaly node data, and divide the power system dynamic topology into load anomaly impact structures according to the load anomaly dominant factors to obtain load anomaly impact nodes; Step S4: Evaluate the three-phase current balance of the distribution cabinet for nodes affected by abnormal loads based on the distribution cabinet sensor data, and classify abnormal nodes based on the current balance of the distribution cabinet to obtain three-phase load offset nodes and three-phase improper connection nodes; Step S5: Reconstruct the power distribution of the three-phase load offset nodes and the three-phase improperly connected nodes to obtain the abnormal node power distribution data, and transmit it to the distribution cabinet management platform to execute the power distribution task.

2. The power system operation and maintenance method of the intelligent distribution cabinet for weak current control according to claim 1 is characterized in that: Step S1 is specifically as follows: Step S11: acquiring power operation data and sensor data of the distribution cabinet, and performing data preprocessing on the power operation data and sensor data of the distribution cabinet to obtain the power operation data to be analyzed and the sensor data of the distribution cabinet to be analyzed; Step S12: extracting the voltage distribution characteristics and current flow characteristics of the distribution cabinet based on the power operation data to be analyzed; Step S13: performing a power system backbone topology analysis based on the voltage distribution characteristics of the power distribution cabinet to obtain a power system backbone topology structure; Step S14: using the current flow characteristics of the distribution cabinet to complete the power system global topology of the power system backbone topology, thereby obtaining the power system topology; Step S15: performing power flow calculation on the power system topology, and constructing a dynamic power system topology according to the power flow calculation results.

3. The power system operation and maintenance method of the intelligent distribution cabinet for weak current control according to claim 2 is characterized in that: Step S13 is specifically as follows: Step S131: classify the voltage levels according to the voltage distribution characteristics of the power distribution cabinet, set the voltage threshold interval to [400V, 220V] to partition the bus nodes, and construct a bus hierarchical structure to obtain bus hierarchical data; Step S132: Calculate the potential difference between different bus levels based on the bus level data, select lines with a potential difference constraint less than or equal to [5V, 20V] as trunk and branch lines, perform topology consistency verification, and generate initial trunk and branch line data; Step S133: constructing a preliminary trunk topology graph with the busbar nodes in the busbar level data as vertices and the initial trunk-branch data as edges, and performing connectivity analysis to remove isolated nodes to generate a preliminary trunk topology structure; Step S134: Analyze the power flow direction based on the voltage distribution characteristics and current flow characteristics of the distribution cabinet, and adjust the branch weights of the preliminary trunk topology according to the power flow direction to obtain an optimized trunk topology; Step S135: performing topological mapping on the optimized backbone topology structure to obtain the power system backbone topology structure.

4. The power system operation and maintenance method of the intelligent distribution cabinet for weak current control according to claim 2, characterized in that: Step S14 is specifically as follows: Step S141: performing weighted directed graph modeling on the current flow characteristics of the power distribution cabinet, setting the current direction threshold to [0.3A, 3.0A], thereby constructing a current flow association graph; Step S142: performing node alignment and edge mapping analysis on the current flow association graph and the power system backbone topology structure, identifying missing connection parts in the power system backbone topology structure, and obtaining the backbone structure node edge data to be completed; Step S143: Perform graph structure reasoning on the edge data of the nodes to be completed in the backbone structure, perform missing connection completion, and obtain global topology prediction data; Step S144: performing edge weight fusion and node uniform coding on the global topology prediction data and the power system backbone topology structure to obtain a preliminary power system topology structure; Step S145: performing consistency check and closed-loop verification on the preliminary power system topology structure to obtain the power system topology structure.

5. The power system operation and maintenance method of the intelligent distribution cabinet for weak current control according to claim 1 is characterized in that: Step S2 is specifically as follows: Step S21: performing spatial mapping and time window slicing on the distribution cabinet sensor data to be analyzed, and performing time window sensor-topology node binding based on the node-branch relationship in the dynamic topology structure of the power system to obtain a spatiotemporal feature map of the power system nodes; Step S22: extracting weighted aggregated energy efficiency indicators based on the spatiotemporal characteristic graph of power system nodes to obtain node energy efficiency indicator data; Step S23: constructing an energy efficiency attribute graph using the node energy efficiency index data, and extracting node energy efficiency behavior characteristics based on the energy efficiency attribute graph; Step S24: clustering energy efficiency behaviors according to the node energy efficiency behavior characteristics to obtain power system energy efficiency model data; Step S25: perform abnormal factor detection and pattern deviation analysis on the power system energy efficiency pattern data, extract potential load abnormality candidate nodes, and perform confidence scoring on the potential load abnormality candidate nodes to obtain power system load abnormality node data.

6. The power system operation and maintenance method of the intelligent distribution cabinet for weak current control according to claim 5 is characterized in that: Step S25 is specifically as follows: Step S251: constructing an energy efficiency behavior residual matrix of each node based on the characteristic vector of each cluster center in the power system energy efficiency model data, and calculating the Euclidean distance between the node energy efficiency behavior characteristics and the energy efficiency behavior residual matrix of the corresponding node to obtain node energy efficiency behavior residual data; Step S252: performing slope fitting and mutation detection based on the node energy efficiency behavior residual data, calculating the abnormal factor score, and obtaining abnormal factor score data; setting the abnormal score dynamic threshold range according to the abnormal factor score data; Step S253: Evaluate the energy efficiency behavior coordination degree of adjacent nodes in the dynamic topology structure of the power system to obtain the energy efficiency behavior coordination degree of the adjacent nodes; Step S254: combining the anomaly score dynamic threshold interval and the adjacent node energy efficiency behavior coordination degree, constructing a weighted anomaly distribution map, identifying and extracting high anomaly factor score nodes, and obtaining potential load anomaly candidate node data; Step S255: perform node confidence evaluation on the potential load anomaly candidate node data, and set the scoring confidence threshold to [0.6, 0.9] to screen high-confidence nodes to obtain power system load anomaly node data.

7. The power system operation and maintenance method of the intelligent distribution cabinet for weak current control according to claim 1 is characterized in that: The main factors of the load anomaly analysis described in step S3 are specifically: The path tracing depth is set to [3,5] layers to dynamically trace the upstream and downstream paths of the power system load abnormality node data, construct a load abnormality path graph, and extract the structural position of each abnormal node in the dynamic topology of the power system and the correlation degree of adjacent nodes to obtain the load abnormality topology path feature data; Based on the load anomaly topology path feature data and combined with the node energy efficiency index data, a topology structure-energy efficiency composite feature vector matrix is constructed; Perform structural impact factor embedding training on the topology-energy efficiency composite feature vector matrix to generate load abnormal structure impact feature data; Obtain the historical load fluctuation sequence and multi-dimensional sensor data of the distribution cabinet, and combine it with the abnormal load node data of the power system to perform characteristic factor correlation analysis to obtain the weight data of the abnormal load physical factor; The load anomaly structural impact feature data and load anomaly physical factor weight data are integrated to perform multidimensional factor compression and reduction, retaining factor combinations with a cumulative contribution rate threshold greater than or equal to 0.85 to obtain the load anomaly impact factor spectrum; Based on the load anomaly influencing factor spectrum, significance ranking and dominant factor identification are performed. The factor contribution rate threshold is set to 0.15 to extract the dominant factors of abnormal nodes and generate the load anomaly dominant factors.

8. The power system operation and maintenance method of the intelligent distribution cabinet for weak current control according to claim 1 is characterized in that: The three-phase current balance evaluation of the power distribution cabinet described in step S4 is specifically as follows: Extract the three-phase current time series data of the load abnormality-affecting node from the distribution cabinet sensor data, perform period segmentation, missing compensation, and distortion point cleaning on the three-phase current time series data, and obtain the calibrated three-phase current standardized time series data; A three-phase current time-domain waveform tensor model is constructed based on the calibrated three-phase current standardized time series data. The three-phase current inter-phase offset rate, phase sequence inconsistency rate, and time-domain imbalance fluctuation rate are extracted using the three-phase current time-domain waveform tensor model to obtain the three-phase current imbalance characteristic data. Based on the three-phase current imbalance characteristic data and the dynamic topology of the power system, a node-level three-phase current imbalance distribution map is constructed. The current offset tension of each abnormal node is calculated using the node-level three-phase current imbalance distribution map, and the current tension propagation weight matrix is obtained. The current tension propagation weight matrix and the three-phase current imbalance characteristic data are fused to calculate the three-phase current balance index and obtain the abnormal node current balance data.

9. The power system operation and maintenance method of the intelligent distribution cabinet for weak current control according to claim 1, characterized in that: The abnormal node classification described in step S4 is specifically as follows: The abnormal node current balance data is divided into balance intervals to obtain graded current balance label data; Based on the dynamic topology structure of the power system, the topological adjacent information and branch connection relationship of the nodes affected by load anomalies are extracted, and the node-phase connection matrix is constructed. Combined with the node interphase connection matrix and the graded current balance label data, the three-phase structure distribution of each load abnormality-affected node is analyzed to obtain the three-phase phase sequence consistency characteristic data; Perform topological structure projection and constraint rule matching on the three-phase phase sequence consistency feature data, perform wiring pattern recognition, and obtain improper wiring identification data; The hierarchical current balance label data and improper wiring identification data are integrated for joint logical judgment, and the load abnormality-affected nodes are divided into three-phase load offset nodes and three-phase improper wiring nodes.

10. A power system operation and maintenance system for an intelligent distribution cabinet for weak current control, characterized in that: The power system operation and maintenance method for the intelligent power distribution cabinet for weak current control according to claim 1 is executed, and the power system operation and maintenance system for the intelligent power distribution cabinet for weak current control comprises: The topology analysis module is used to obtain the power operation data and sensor data of the distribution cabinet, and perform power system topology analysis based on the power operation data of the distribution cabinet to obtain the dynamic topology of the power system; The potential load anomaly analysis module is used to identify the power system energy efficiency pattern of the distribution cabinet sensor data through the dynamic topology of the power system to obtain the power system energy efficiency pattern data; perform potential load anomaly analysis on the power system energy efficiency pattern data to obtain the power system load anomaly node data; The abnormal impact structure division module is used to analyze the load abnormality dominant factors based on the load abnormality node data of the power system, and divide the load abnormality impact structure of the power system dynamic topology structure according to the load abnormality dominant factors to obtain the load abnormality impact nodes; The three-phase current balance evaluation module is used to evaluate the three-phase current balance of the distribution cabinet at the nodes affected by abnormal loads based on the distribution cabinet sensor data, and classify abnormal nodes based on the current balance of the distribution cabinet to obtain the three-phase load offset nodes and the three-phase improper wiring nodes; The abnormal node power redistribution module is used to reconstruct the power distribution of three-phase power load offset nodes and three-phase power improperly connected nodes, obtain the abnormal node power distribution data, and transmit it to the distribution cabinet management platform to execute the power distribution task.

Images