Power equipment anomaly detection method and device, equipment and storage medium
By constructing a topology weighted graph of power equipment and using a graph partitioning algorithm for dynamic partitioning, combined with multi-level anomaly detection, the problem of unconsidered coupling relationships between equipment in existing technologies is solved, achieving efficient power equipment anomaly detection and regional fault identification, and improving the system's detection efficiency and interpretability.
Patent Information
- Application Number
- CN202511223260.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-29
- Publication Date
- 2025-12-12
AI Technical Summary
Existing power equipment anomaly detection methods based on streaming processing engines fail to effectively consider the electrical coupling relationship and geographical proximity between devices, resulting in high regional fault miss rates, coarse partitioning and computational scheduling, lack of hierarchical detection mechanisms, serious resource waste, and insufficient spatial and topological correlation capabilities.
A weighted topology graph of power equipment is constructed, and a graph partitioning algorithm is used for dynamic partitioning for load awareness. Combined with a multi-level anomaly detection mechanism, abnormal links and regions are identified. The partitioning layout is optimized through the graph partitioning algorithm to achieve load balancing and efficient detection.
It improves the overall throughput and parallel processing capability of the system, reduces resource overhead, enhances the interpretability of abnormal results and the practical value of scheduling, and can identify key abnormal devices and propagation paths, facilitating rapid location and response.
Smart Images

Figure CN121116610A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to power grid technology, and more particularly to a method, apparatus, equipment, and storage medium for detecting abnormalities in power equipment. Background Technology
[0002] Real-time anomaly detection of large-scale power equipment can help managers promptly identify potential faults such as equipment insulation degradation, partial discharge, and local power grid paralysis, enabling them to respond quickly, resolve the problems of the corresponding power equipment, or uncover potential value.
[0003] Currently, to achieve online monitoring and anomaly detection of power equipment, the industry has proposed data processing systems based on streaming frameworks. These typically use platforms such as Flink and Spark Streaming as real-time computing engines, combined with multi-source sensor data streams, to achieve real-time analysis of equipment operating status. The technical approach of such systems is as follows: Figure 1 As shown, it mainly includes the following key modules:
[0004] 1. Data Access and Stream Preprocessing Module. This module receives raw data (including current, voltage, frequency, temperature, and switch status) from substations, meters, and sensors and feeds it into the stream computing platform via a message queue (such as Kafka). Stream preprocessing includes operations such as time alignment, missing value imputation, and format normalization to meet the input requirements of subsequent model processing.
[0005] 2. Data Partitioning. Data generated by different devices is partitioned, and the aggregated results from each unit are sent to each machine according to the partition.
[0006] 3. Rule-based or model-based single-point anomaly detection module. The system typically monitors power equipment operating data in real time based on fixed threshold rules or single-device time series models. For example, an alarm is triggered when the current exceeds a certain value; or a machine learning model predicts the voltage value at the next time point and compares it with the actual value to determine if an anomaly is detected.
[0007] 4. Global Region Aggregation. The recognition results from multiple machines in step 3 are aggregated into one machine by performing a global aggregation operation.
[0008] 5. Anomaly Reporting and Visualization Module. Detected anomalies are transmitted to the visualization system or scheduling backend via a REST interface or database system for manual verification or automatic early warning processing.
[0009] However, existing power equipment anomaly detection methods based on streaming processing engines only analyze the temporal correlation of data streams from individual devices, ignoring electrical coupling relationships and geographical proximity between devices, resulting in a high rate of missed detections for regional faults.
[0010] While the aforementioned technical architecture possesses certain real-time processing and anomaly detection capabilities, it still exhibits the following significant shortcomings when dealing with large-scale device sensing systems, highly coupled topologies, and heterogeneous sensing environments:
[0011] (1) Coarse partitioning and computation scheduling: Most current systems use device number balancing or simple hashing to partition streaming data without considering the differences in device data update frequency and computational complexity. This results in some computing nodes being overloaded, having low processing efficiency, and easily becoming performance bottlenecks.
[0012] (2) Lack of hierarchical detection mechanism: Most existing methods only focus on single device anomalies and do not establish regional anomaly overview and multi-level diagnostic paths, which can easily lead to full model calls, waste of computing resources, and anomalies that are unexplainable and untraceable.
[0013] (3) Insufficient spatial and topological correlation capabilities: Existing technologies analyze equipment as independent individuals, lacking spatial proximity and electrical topology modeling between equipment, and cannot accurately identify regional faults, chain-conduction faults or equipment cluster anomalies. Summary of the Invention
[0014] This invention provides a method, apparatus, device, and storage medium for detecting anomalies in power equipment, thereby improving the overall system throughput and parallel processing capabilities, enhancing overall detection efficiency, and increasing the interpretability and scheduling practical value of anomaly results.
[0015] In a first aspect, the present invention provides a method for detecting abnormalities in power equipment, comprising:
[0016] Obtain metadata of each power device in the power grid system;
[0017] Based on the topological relationships and load weights of each power device, a power device topology weighted graph is constructed, and a graph partitioning algorithm is used to partition the power device topology weighted graph to obtain a heterogeneous graph including multiple spatial partitions. Each spatial partition includes at least one spatial unit, which is the smallest administrative unit to which the power device belongs, and each spatial unit includes multiple power devices.
[0018] Based on the metadata of each power device, anomaly detection is performed step by step according to spatial partition, spatial unit, and power device to identify the target power device with anomalies.
[0019] The topology search based on each power device includes abnormal links and abnormal regions of the target power device;
[0020] The system will report any abnormal target power equipment, abnormal links, and abnormal areas to the dispatching system and visualization platform.
[0021] Optionally, obtain metadata for each power device in the power grid system, including:
[0022] Real-time acquisition of metadata for various power equipment in the power grid system via Kafka message queue;
[0023] The metadata is standardized to unify field structure, units, and precision;
[0024] Parse the additional information in the metadata, including the exception code and the communication status identifier;
[0025] The metadata is streamed downstream using a unified data structure.
[0026] Optionally, based on the topological relationships and load weights of each power device, a power device topology weighted graph is constructed, and a graph partitioning algorithm is used to partition the power device topology weighted graph to obtain a heterogeneous graph including multiple spatial partitions, including:
[0027] A power equipment topology graph is constructed, with each power equipment as a node and the spatial proximity and electrical connection relationships between the power equipment as edges between the nodes.
[0028] The load weight of the power equipment is calculated based on the data reporting frequency, model calculation complexity, and state space occupied by the power equipment.
[0029] Assign corresponding load weights to each power device in the power equipment topology graph to obtain a power equipment topology weighted graph;
[0030] With the goal of minimizing the number of edges across spatial partitions, a graph partitioning algorithm is used to partition the topological weighted graph of the power equipment to obtain a heterogeneous graph including multiple spatial partitions. Each spatial partition includes at least one spatial unit, which is the smallest administrative unit to which the power equipment belongs, and each spatial unit includes multiple power equipment.
[0031] Optionally, based on the metadata of each power device, anomaly detection is performed step by step according to spatial partition, spatial unit, and power device to identify the target power device with anomalies, including:
[0032] A window aggregation statistical algorithm is used to screen potential risk areas from all spatial partitions;
[0033] A spatial density algorithm is used to identify anomalous risk units from the risk region.
[0034] An inference model is used to infer the time-series data of the power equipment in the risk unit to identify target power equipment with anomalies, wherein each power equipment has a corresponding inference model.
[0035] Optionally, the topology search based on each power device includes abnormal links and abnormal regions of the target power device, including:
[0036] A power equipment topology graph is constructed, with each power equipment as a node and the spatial proximity and electrical connection relationships between the power equipment as edges between the nodes.
[0037] The propagation path weight is calculated based on the node attributes of the power equipment, wherein the node attributes include equipment type, location, whether it is abnormal, and abnormality score;
[0038] For each target power device, starting from the target power device, a heuristic weighted path algorithm is used to find other power devices connected to the target power device in the power device topology graph;
[0039] The target power device and the link formed by other power devices connected to the target power device are designated as abnormal links, and the area where the abnormal link is located is designated as an abnormal area.
[0040] Optionally, the link formed by the target power device and other power devices connected to the target power device is designated as an abnormal link, and the area where the abnormal link is located is designated as an abnormal area, including:
[0041] The target power device and the link formed by other power devices connected to the target power device are considered as potential abnormal links;
[0042] Evaluate the temporal consistency, spatial proximity, and topological distance of nodes in potential anomalous links;
[0043] Based on the evaluation structure, isolated nodes are screened out, abnormal links are identified, and the abnormal regions where the abnormal links are located are determined.
[0044] Optionally, the region where the abnormal link is located is identified as the abnormal region, including:
[0045] Map the abnormal link back to the spatial unit to determine the target spatial unit where the node of the abnormal link design is located;
[0046] The target spatial units are aggregated to form an abnormal region.
[0047] Secondly, the present invention also provides a power equipment anomaly detection device, comprising:
[0048] The data access module is used to acquire metadata of various power equipment in the power grid system;
[0049] The load-aware graph partitioning module is used to construct a power equipment topology weighted graph based on the topological relationship and load weight of each power equipment, and to partition the power equipment topology weighted graph using a graph partitioning algorithm to obtain a heterogeneous graph including multiple spatial partitions. Each spatial partition includes at least one spatial unit, which is the smallest administrative unit to which the power equipment belongs, and each spatial unit includes multiple power equipment.
[0050] The multi-level anomaly detection module is used to perform anomaly detection based on the metadata of each power device, according to spatial partitioning, spatial unit, and power device level, to identify the target power device with anomalies.
[0051] An abnormal link and region identification module is used to search for abnormal links and abnormal regions of the target power equipment based on the topological relationship of each power equipment.
[0052] The abnormal result output module is used to report abnormal target power equipment, abnormal links, and abnormal areas to the dispatching system and visualization platform.
[0053] Thirdly, the present invention also provides an electronic device, comprising:
[0054] One or more processors;
[0055] Storage device for storing one or more programs;
[0056] When the one or more programs are executed by the one or more processors, the one or more processors implement the power equipment anomaly detection method as provided in the first aspect of the present invention.
[0057] Fourthly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the power equipment anomaly detection method as provided in the first aspect of the present invention.
[0058] The power equipment anomaly detection method provided by this invention, based on a dynamic graph partitioning method with computational load perception, achieves a load-balanced partitioned layout, significantly reducing load imbalance between processing nodes, avoiding hotspot bottlenecks caused by the concentration of high-frequency equipment, improving the overall system throughput and parallel processing capabilities, and meeting the performance requirements of real-time high-frequency data processing. The constructed multi-level anomaly detection mechanism of partition-unit-equipment and a critical equipment fallback detection mechanism triggers detection tasks at each level as needed, supporting lightweight regional-level pre-screening and refined equipment-level identification, effectively reducing system resource overhead, improving overall detection efficiency, and enhancing the ability to identify anomalies in clustered areas and faults in critical nodes, avoiding missed detections of important equipment, and ensuring the integrity and robustness of system detection coverage. By analyzing the anomaly propagation path in the electrical topology and combining it with the spatial location information of equipment to generate anomaly region boundary representations, the method achieves a transformation from point-level identification to chain-level analysis and region-level visualization. It can not only identify critical anomaly equipment but also clearly display the anomaly propagation path and affected areas, facilitating rapid location and response by maintenance personnel, and improving the interpretability and scheduling practical value of anomaly results.
[0059] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description
[0060] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0061] Figure 1 This is a schematic diagram of the technical route for existing power equipment anomaly detection methods.
[0062] Figure 2 A flowchart of a power equipment anomaly detection method provided by the present invention;
[0063] Figure 3 This is a schematic diagram of the technical route of the power equipment anomaly detection method provided by the present invention;
[0064] Figure 4 This is a schematic diagram of the structure of a power equipment anomaly detection device provided by the present invention;
[0065] Figure 5 A schematic diagram illustrating the spatial partitioning process provided by this invention;
[0066] Figure 6 A schematic diagram of multi-level anomaly detection provided by the present invention;
[0067] Figure 7 This is a schematic diagram illustrating the abnormal link and region identification provided by the present invention;
[0068] Figure 8 This is a schematic diagram of the structure of the power equipment anomaly detection device provided by the present invention;
[0069] Figure 9 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention.
[0070] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation
[0071] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.
[0072] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.
[0073] Figure 2 This invention provides a flowchart of a method for detecting abnormalities in power equipment. Figure 3 This is a schematic diagram of the technical route of the power equipment anomaly detection method provided by the present invention. Figure 4This is a schematic diagram of a power equipment anomaly detection device provided by the present invention. This embodiment can be applied to quality control of workpieces during batch processing. The method can be executed by the laser processing quality control device provided by the present invention. This device can be implemented by software and / or hardware, and is usually configured in electronic equipment, such as... Figure 2-4 As shown, the abnormal detection method for power equipment includes the following steps:
[0074] S101. Obtain metadata of each power device in the power grid system.
[0075] In this embodiment of the invention, metadata of various power devices in the power grid system is obtained through a data access module. The data access module is responsible for loading and pre-parsing the metadata of the power sensing devices before the stream computing task starts, and continuously receiving real-time data streams generated by multiple sources of sensors during task execution. It performs format unification, time alignment, and preliminary preprocessing to ensure that subsequent modules can perform efficient processing based on a unified data structure. This module supports the access of various types of devices, including distribution transformers, cable joints, switches, feeder terminals, smart meters, and PMUs, and is compatible with various communication protocols and data publishing formats. The main steps are as follows:
[0076] Step 1: Load power equipment metadata and topology relationships. During the initialization phase, the system loads the basic attributes (such as ID, equipment type, spatial location, feeder to which it belongs, state variable definition, etc.) and electrical connection information of various sensing devices from the external device management platform, and establishes an initial device attribute table and device connection relationship table, providing a data foundation for graph modeling and partitioning.
[0077] Step 2: Integrate real-time data streams. The system subscribes to a Kafka message queue to integrate data streams from different types of sensor devices in real time. Each data packet contains fields such as timestamp, device ID, measurement value, and anomaly flags, and may carry additional metadata (such as sampling frequency and communication status).
[0078] Step 3: Complete data standardization and parsing. This module standardizes the format of the incoming data, unifying field structure, units, and precision, and parses additional information such as exception codes and communication status identifiers. Finally, this module outputs a unified data structure (such as a JSON object or an internal serialized object), including: device ID, acquisition time, spatial location, array of measurement values, status codes, etc., and streams it to the downstream processing module.
[0079] S102. Based on the topological relationship of each power device and the load weight of each power device, construct a power device topology weighted graph, and use a graph partitioning algorithm to partition the power device topology weighted graph to obtain a heterogeneous graph including multiple spatial partitions.
[0080] In this embodiment of the invention, the load-aware graph partitioning module constructs a power equipment topology weighted graph based on the topological relationships and load weights of each power equipment, and then uses a graph partitioning algorithm to partition the power equipment topology weighted graph to obtain a heterogeneous graph including multiple spatial partitions. Each spatial partition includes at least one spatial unit, which is the smallest administrative unit to which a power equipment belongs, and each spatial unit includes multiple power equipment. Figure 5 A schematic diagram of the spatial partitioning process provided by the present invention, as shown below. Figure 5 As shown, the load-aware graph partitioning module constructs a graph structure with load weights (i.e., a power equipment topology weighted graph) based on the topology and computing resource characteristics of power equipment. It then uses a multi-objective graph partitioning algorithm to partition the device network, ultimately providing fundamental support for subsequent data routing and distributed task scheduling. Its core design incorporates indicators such as data reporting frequency, model computational complexity, and state space occupancy into the load assessment system, implementing a computational intensity-aware graph partitioning strategy. This alleviates the node load imbalance problem caused by traditional average partitioning. The main steps include:
[0081] Step 1: Construct the equipment graph. Based on the metadata table and connection relationship table provided by the data access module, the system constructs a power equipment topology graph, abstracting all power equipment as vertices of the graph, and modeling spatial proximity and electrical connection relationships as edges of the graph. Edge weights can be set according to physical distance, resistance, feeder topology level, etc.
[0082] Step 2: Construct evaluation metrics and assign node weights. The system calculates load metrics for each vertex, including but not limited to: data reporting frequency (e.g., 100Hz for PMU, 1Hz for ordinary meters), model computational complexity (e.g., whether deep learning is used), state space occupancy, etc., and sets the weight synthesis strategy according to the configuration. Finally, a single load weight is generated for each node, which is used as the node weight input in the graph partitioning to obtain the power equipment topology weighted graph.
[0083] For example, in this embodiment of the invention, a computational load index (WL) for different devices and different detection methods is defined, as shown in formula (1):
[0084]
[0085] Where F represents the frequency of data reported by the device, and the time complexities of different detection algorithms are assumed to be: O(1) for the threshold-based detection algorithm, O(n) for the time window-based detection algorithm, and O(n) for the machine learning or deep learning model-based detection algorithm. 2In actual deployment, the time complexity parameter in the formula can also be replaced by the inference speed of each detection algorithm. This inference speed can be obtained through pre-experiment statistics, thus more accurately reflecting the computational overhead of different algorithms in a real environment.
[0086] In a spatial proximity graph, the computational load of a vertex is defined as the sum of the computational loads of all devices within that spatial cell, and its specific calculation method is shown in formula (2):
[0087]
[0088] Among them, wl cell Let wl represent the computational load of a vertex (i.e., a spatial cell) in the graph, where n is the number of devices contained within that spatial cell. i Let represent the computational load of the i-th device.
[0089] Step 3: Perform multi-objective graph partitioning. Partition the graph using a graph partitioning algorithm that supports weighted input (such as the improved Metis, Louvain variant, or Streaming Graph Partitioning method). Minimize the number of edges across spatial partitions while maintaining load balancing, thereby improving local processing efficiency and data locality. The partitioning results are recorded as a mapping table of "Device ID → Partition Number," used for data flow routing and task deployment.
[0090] Step 4: Support dynamic updates of graph partitioning. Considering that devices may be dynamically added or removed, or their operating status may change, the system is configured with a timed task scheduling and event triggering mechanism: when a new device is added or the load score of an existing device changes beyond a set threshold, an incremental graph update process will be triggered, performing local repartitioning only on the affected area to avoid system jitter caused by full graph reconstruction.
[0091] Through the above steps, the load perception graph partitioning module can dynamically optimize the partitioning layout of the power perception network based on the runtime computing resource status, thereby achieving reasonable resource scheduling, hotspot dispersion, and high concurrency data processing.
[0092] This invention introduces a computational load-aware graph partitioning algorithm based on multiple factors such as device data frequency, model computational complexity, and state memory usage. It is implemented by the graph partitioning logic unit in the system, replacing the simple partitioning method based on spatial location or fixed ID hashing in the prior art. This significantly improves the resource utilization efficiency and system processing throughput in the stream processing process.
[0093] S103. Based on the metadata of each power device, perform anomaly detection step by step according to spatial partition, spatial unit, and power device to identify the target power device with anomalies.
[0094] In this embodiment of the invention, the multi-level anomaly detection module performs anomaly detection based on the metadata of each power device, proceeding step by step according to spatial partitions, spatial units, and power devices, to identify the target power devices with anomalies. Based on the graph partitioning results, the multi-level anomaly detection module constructs a three-level anomaly detection structure of "spatial partition → spatial unit → device" within each partition. It employs a step-by-step triggering mechanism and a dynamic indicator system based on historical modeling to identify macro-level regional risks and micro-level device faults, ensuring the efficiency, accuracy, and coverage integrity of anomaly detection. Figure 6 This is a schematic diagram of the multi-level anomaly detection provided by the present invention, as shown below. Figure 6 As shown, the main steps include:
[0095] Step 1: The system pre-defines an evaluation index system and anomaly judgment framework covering the zoning, unit, and equipment levels. Evaluation indicators include absolute limits and sliding window statistical indicators for basic monitoring values such as voltage, current, and frequency; time series characteristics such as rate of change, slope abrupt changes, and anomaly duration; and power business-related indicators such as voltage imbalance, load fluctuation rate, and anomaly density. The system sets structured calculation methods and anomaly scoring mechanisms for each type of indicator and establishes hierarchical triggering rules: zoning-level indicators trigger unit-level detection, and unit-level anomalies trigger equipment-level detection, thereby constructing an efficient hierarchical linkage detection path.
[0096] Step 2: The system uses historical operating data to statistically model the above indicators, extracting the normal fluctuation range and confidence interval for each indicator. Dynamic reference baselines are established at the daily, load, and equipment category levels using methods such as quantile estimation, sliding window kernel density fitting, or box-line analysis. These statistical results provide adaptive judgment thresholds for real-time detection, improving the system's adaptability to different operating environments and load scenarios, and avoiding false alarms or missed alarms caused by fixed thresholds.
[0097] Step 3: The system performs step-by-step inspection of the real-time data stream according to the evaluation indicators and judgment framework. At the zone level, the system quickly screens potential risk areas based on window aggregation statistics (such as anomaly ratio, standard deviation, and outlier percentage); once an anomaly is detected, the inspection task of its subordinate spatial units is triggered. Unit-level inspection uses the smallest administrative unit (such as a street block) to which the power equipment belongs as the granularity, and uses algorithms such as local clustering and spatial density estimation to identify regional anomaly clusters; if the density anomaly of a unit exceeds a set threshold, the equipment-level inspection process within that unit is further triggered. Equipment-level inspection then uses inference models such as rule engines, isolated forests, and LSTM to conduct in-depth analysis of the time-series data of individual equipment, evaluate its short-term volatility, prediction residuals, and confidence scores, and generate the final equipment anomaly judgment result.
[0098] Step 4: To avoid missing critical equipment due to the hierarchical triggering mechanism, the system is designed with a fallback detection mechanism for critical nodes. Maintenance personnel can set a whitelist of critical equipment based on the power grid structure, such as backbone line equipment, heavily loaded nodes, and power injection points. The system will periodically perform independent checks on these devices without relying on upper-level triggering logic, ensuring that their abnormal states can be identified immediately. The fallback detection task uses a lightweight model or multi-threaded asynchronous processing method to run without interfering with the main detection process, ensuring the system's real-time performance and detection integrity.
[0099] This invention proposes a three-level anomaly detection structure of partition-unit-device, and implements step-by-step triggering and indicator evaluation by multi-level anomaly detection sub-modules running on each distributed node. Unlike the "planar" strategy of performing point-by-point detection on all devices in existing methods, this invention significantly reduces the overall computational burden of the system and improves the detection coverage and response efficiency of regional faults.
[0100] S104. The topology relationship search based on each power device includes abnormal links and abnormal regions of the target power device.
[0101] In this embodiment of the invention, the abnormal link and region identification module searches for abnormal links and abnormal regions of the target power equipment based on the topological relationships of each power device. Based on the device-level anomaly results, the module combines the electrical connections and spatial relationships of the power system to construct anomaly propagation links, identify potential anomaly diffusion paths and regional ranges, and output the final results in a formatted format for use by the operation and maintenance scheduling system and GIS platform. Figure 7 This is a schematic diagram of abnormal link and region identification provided by the present invention, as shown below. Figure 7 As shown, the main steps include:
[0102] Step 1: The system base data access module receives device connection information, and the load perception graph partitioning module constructs a device graph structure. An electrical topology graph is initialized, with all sensing devices as nodes and physical or logical connections as edges, forming a basic topology network for anomaly propagation analysis. In this graph structure, node attributes include device type, location, whether it is abnormal, and anomaly score. Edges can have auxiliary attributes such as impedance, voltage level, and feeder number attached for propagation path weight evaluation.
[0103] Step 2: Within each detection cycle, the system performs a topology traversal operation on all device nodes identified as anomalous. Starting with each anomalous device, it uses breadth-first search (BFS) or a heuristic weighted path algorithm to find other anomalous nodes directly or indirectly connected to it in the topology graph, forming a preliminary set of anomalous links. The system evaluates the temporal consistency (e.g., time difference), spatial proximity, and topological distance between anomalous nodes in each path, filtering out potentially isolated or false alarm nodes and retaining highly coupled anomalous propagation links.
[0104] Step 3: The system maps each identified abnormal link back to a spatial cell, extracts the spatial cells of the devices involved, and constructs the boundary contour of the abnormal region by aggregating the spatial cells, generating a spatial abnormal region with geometric semantics. If multiple links intersect or are adjacent, forming spatial overlap, the system further performs a region merging operation to ensure the continuity and consistency of the abnormal region description.
[0105] This invention uses the joint analysis of device connection diagrams and spatial indexes, with the topology chain identification module and the region merging module working together to identify fault propagation paths and extract spatial anomaly region contours. Compared with existing solutions that only provide alarms for single device anomalies or have no spatial output capability, this invention supports the structured and visual representation of power grid anomalies and has stronger interpretability and scheduling linkage capabilities.
[0106] S105. Report any abnormal target power equipment, abnormal links, and abnormal areas to the dispatching system and visualization platform.
[0107] In this embodiment of the invention, the abnormal result output module structures and encapsulates the abnormal devices, abnormal links, and spatial abnormal areas identified by the preceding module, and outputs the results to the scheduling system and visualization platform, supporting alarm linkage and operation and maintenance response. Specifically, this may include the following steps:
[0108] Step 1: The system encapsulates the results of each detection cycle into a standard data structure, which includes information such as a list of abnormal devices, abnormal link paths, spatial area boundaries, alarm levels, and timestamps.
[0109] Step 2: The system synchronizes the results to the scheduling platform through interface push or message queue publishing, so as to realize real-time visualization and closed-loop handling support for abnormal states.
[0110] The power equipment anomaly detection method provided by this invention, based on a dynamic graph partitioning method with computational load perception, achieves a load-balanced partitioned layout, significantly reducing load imbalance between processing nodes, avoiding hotspot bottlenecks caused by the concentration of high-frequency equipment, improving the overall system throughput and parallel processing capabilities, and meeting the performance requirements of real-time high-frequency data processing. The constructed multi-level anomaly detection mechanism of partition-unit-equipment and a critical equipment fallback detection mechanism triggers detection tasks at each level as needed, supporting lightweight regional-level pre-screening and refined equipment-level identification, effectively reducing system resource overhead, improving overall detection efficiency, and enhancing the ability to identify anomalies in clustered areas and faults in critical nodes, avoiding missed detections of important equipment, and ensuring the integrity and robustness of system detection coverage. By analyzing the anomaly propagation path in the electrical topology and combining it with the spatial location information of equipment to generate anomaly region boundary representations, the method achieves a transformation from point-level identification to chain-level analysis and region-level visualization. It can not only identify critical anomaly equipment but also clearly display the anomaly propagation path and affected areas, facilitating rapid location and response by maintenance personnel, and improving the interpretability and scheduling practical value of anomaly results.
[0111] Figure 8 This is a schematic diagram of the power equipment anomaly detection device provided by the present invention, as shown below. Figure 8 As shown, the power equipment anomaly detection device includes:
[0112] Data access module 201 is used to acquire metadata of various power equipment in the power grid system;
[0113] The load-aware graph partitioning module 202 is used to construct a power equipment topology weighted graph based on the topology relationship of each power equipment and the load weight of each power equipment, and to partition the power equipment topology weighted graph using a graph partitioning algorithm to obtain a heterogeneous graph including multiple spatial partitions. The spatial partition includes at least one spatial unit, which is the smallest administrative unit to which the power equipment belongs, and the spatial unit includes multiple power equipment.
[0114] The multi-level anomaly detection module 203 is used to perform anomaly detection based on the metadata of each power device, according to spatial partition, spatial unit, and power device level, to identify the target power device with anomalies.
[0115] The abnormal link and region identification module 204 is used to search for abnormal links and abnormal regions of the target power equipment based on the topological relationship of each power equipment.
[0116] The abnormal result output module 205 is used to report the abnormal target power equipment, abnormal links and abnormal areas to the dispatching system and visualization platform.
[0117] In some embodiments of the present invention, the data access module 201 includes:
[0118] The metadata acquisition submodule is used to acquire metadata of various power equipment in the power grid system in real time through the Kafka message queue;
[0119] The normalization processing submodule is used to normalize the metadata, unifying the field structure, units, and precision.
[0120] The information parsing submodule is used to parse the additional information in the metadata, including the exception code and the communication status identifier.
[0121] The streaming delivery submodule is used to send the metadata to the downstream in a unified data structure.
[0122] In some embodiments of the present invention, the load-aware graph partitioning module 202 includes:
[0123] The first topology graph construction submodule is used to construct a power equipment topology graph with each power equipment as a node and the spatial proximity and electrical connection relationships of the power equipment as edges between the nodes.
[0124] The load weight calculation submodule is used to calculate the load weight of the power equipment based on the data reporting frequency, model calculation complexity, and state space occupied by the power equipment.
[0125] The weighted graph construction submodule is used to assign corresponding load weights to each power device in the power equipment topology graph to obtain a power equipment topology weighted graph.
[0126] The spatial partitioning submodule is used to partition the topological weighted graph of the power equipment with the goal of minimizing the number of edges across spatial partitions, and to obtain a heterogeneous graph including multiple spatial partitions. The spatial partition includes at least one spatial unit, which is the smallest administrative unit to which the power equipment belongs, and the spatial unit includes multiple power equipment.
[0127] In some embodiments of the present invention, the multi-level anomaly detection module 203 includes:
[0128] The risk area screening submodule is used to screen potential risk areas from all spatial partitions using a window aggregation statistical algorithm;
[0129] The risk unit identification submodule is used to identify abnormal risk units from the risk area using a spatial density algorithm.
[0130] The abnormal equipment identification submodule is used to infer the time-series data of the power equipment in the risk unit using an inference model to identify the target power equipment that is abnormal, wherein each power equipment has a corresponding inference model.
[0131] In some embodiments of the present invention, the abnormal link and region identification module 204 includes:
[0132] The second topology graph construction submodule is used to construct a power equipment topology graph with each power equipment as a node and the spatial proximity and electrical connection relationships of the power equipment as edges between the nodes.
[0133] The propagation path weight calculation submodule is used to calculate the propagation path weight based on the node attributes of the power equipment, wherein the node attributes include equipment type, location, whether it is abnormal, and abnormality score;
[0134] The search submodule is used to, for each target power device, use a heuristic weighted path algorithm to search for other power devices connected to the target power device in the power device topology graph, starting from the target power device;
[0135] The link and region determination submodule is used to identify the target power equipment and other power equipment connected to the target power equipment as abnormal links, and to determine the region where the abnormal links are located as abnormal regions.
[0136] In some embodiments of the present invention, the link and region determination submodule includes:
[0137] A potential abnormal link determination unit is used to identify the target power device and the link formed by other power devices connected to the target power device as potential abnormal links.
[0138] The node evaluation unit is used to evaluate the temporal consistency, spatial proximity, and topological distance of nodes in a potentially anomalous link.
[0139] The link and region determination unit is used to filter out isolated nodes based on the evaluation structure, determine abnormal links, and identify the abnormal regions where the abnormal links are located.
[0140] In some embodiments of the present invention, the link and region determination submodule further includes:
[0141] A mapping unit is used to map the abnormal link back to a spatial unit and determine the target spatial unit where the node of the abnormal link design is located.
[0142] An aggregation unit is used to aggregate the target spatial units to form an abnormal region.
[0143] The aforementioned power equipment anomaly detection device can execute the power equipment anomaly detection method provided in the foregoing embodiments of the present invention, and has the corresponding functional modules and beneficial effects for executing the power equipment anomaly detection method.
[0144] Figure 9 This is a schematic diagram of an electronic device provided for an embodiment of the present invention. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (such as helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.
[0145] like Figure 9 As shown, the electronic device includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded into the RAM 13 from storage unit 18. The RAM 13 can also store various programs and data required for the operation of the electronic device. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.
[0146] Multiple components in the electronic device are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, optical disk, etc.; and a communication unit 19, such as a network card, modem, wireless transceiver, etc. The communication unit 19 allows the electronic device to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.
[0147] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as power equipment anomaly detection methods.
[0148] In some embodiments, the power equipment anomaly detection method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on an electronic device via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the power equipment anomaly detection method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the power equipment anomaly detection method by any other suitable means (e.g., by means of firmware).
[0149] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.
[0150] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.
[0151] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.
[0152] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).
[0153] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.
[0154] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.
[0155] This invention also provides a computer program product, including a computer program that, when executed by a processor, implements the power equipment anomaly detection method provided in any embodiment of this application.
[0156] In implementing the computer program product, computer program code for performing the operations of this invention can be written in one or more programming languages or a combination thereof. Programming languages include object-oriented programming languages such as Java, Smalltalk, and C++, as well as conventional procedural programming languages such as C or similar languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).
[0157] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.
[0158] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.
Claims
1. A method for detecting abnormalities in power equipment, characterized in that, include: Obtain metadata of each power device in the power grid system; Based on the topological relationships and load weights of each power device, a power device topology weighted graph is constructed, and a graph partitioning algorithm is used to partition the power device topology weighted graph to obtain a heterogeneous graph including multiple spatial partitions. Each spatial partition includes at least one spatial unit, which is the smallest administrative unit to which the power device belongs, and each spatial unit includes multiple power devices. Based on the metadata of each power device, anomaly detection is performed step by step according to spatial partition, spatial unit, and power device to identify the target power device with anomalies. The topology search based on each power device includes abnormal links and abnormal regions of the target power device; The system will report any abnormal target power equipment, abnormal links, and abnormal areas to the dispatching system and visualization platform.
2. The method for detecting abnormalities in power equipment according to claim 1, characterized in that, Obtain metadata for each power device in the power grid system, including: Real-time acquisition of metadata for various power equipment in the power grid system via Kafka message queue; The metadata is standardized to unify field structure, units, and precision; Parse the additional information in the metadata, including the exception code and the communication status identifier; The metadata is streamed downstream using a unified data structure.
3. The method for detecting abnormalities in power equipment according to claim 1, characterized in that, Based on the topological relationships and load weights of each power device, a power device topology weighted graph is constructed. A graph partitioning algorithm is then used to divide the power device topology weighted graph, resulting in a heterogeneous graph comprising multiple spatial partitions, including: A power equipment topology graph is constructed, with each power equipment as a node and the spatial proximity and electrical connection relationships between the power equipment as edges between the nodes. The load weight of the power equipment is calculated based on the data reporting frequency, model calculation complexity, and state space occupied by the power equipment. Assign corresponding load weights to each power device in the power equipment topology graph to obtain a power equipment topology weighted graph; With the goal of minimizing the number of edges across spatial partitions, a graph partitioning algorithm is used to partition the topological weighted graph of the power equipment to obtain a heterogeneous graph including multiple spatial partitions. Each spatial partition includes at least one spatial unit, which is the smallest administrative unit to which the power equipment belongs, and each spatial unit includes multiple power equipment.
4. The method for detecting abnormalities in power equipment according to claim 1, characterized in that, Based on the metadata of each power device, anomaly detection is performed step by step according to spatial partition, spatial unit, and power device to identify target power devices with anomalies, including: A window aggregation statistical algorithm is used to screen potential risk areas from all spatial partitions; A spatial density algorithm is used to identify anomalous risk units from the risk region. An inference model is used to infer the time-series data of the power equipment in the risk unit to identify target power equipment with anomalies, wherein each power equipment has a corresponding inference model.
5. The method for detecting abnormalities in power equipment according to any one of claims 1-4, characterized in that, The topology search based on each power device includes abnormal links and abnormal regions of the target power device, including: A power equipment topology graph is constructed, with each power equipment as a node and the spatial proximity and electrical connection relationships between the power equipment as edges between the nodes. The propagation path weight is calculated based on the node attributes of the power equipment, wherein the node attributes include equipment type, location, whether it is abnormal, and abnormality score; For each target power device, starting from the target power device, a heuristic weighted path algorithm is used to find other power devices connected to the target power device in the power device topology graph; The target power device and the link formed by other power devices connected to the target power device are designated as abnormal links, and the area where the abnormal link is located is designated as an abnormal area.
6. The method for detecting abnormalities in power equipment according to claim 5, characterized in that, The target power device and the link formed by other power devices connected to the target power device are designated as abnormal links, and the area where the abnormal link is located is designated as an abnormal area, including: The target power device and the link formed by other power devices connected to the target power device are considered as potential abnormal links; Evaluate the temporal consistency, spatial proximity, and topological distance of nodes in potential anomalous links; Based on the evaluation structure, isolated nodes are screened out, abnormal links are identified, and the abnormal regions where the abnormal links are located are determined.
7. The method for detecting abnormalities in power equipment according to claim 5, characterized in that, The region where the abnormal link is located is identified as the abnormal region, including: Map the abnormal link back to the spatial unit to determine the target spatial unit where the node of the abnormal link design is located; The target spatial units are aggregated to form an abnormal region.
8. A device for detecting abnormalities in power equipment, characterized in that, include: The data access module is used to acquire metadata of various power equipment in the power grid system; The load-aware graph partitioning module is used to construct a power equipment topology weighted graph based on the topological relationship and load weight of each power equipment, and to partition the power equipment topology weighted graph using a graph partitioning algorithm to obtain a heterogeneous graph including multiple spatial partitions. Each spatial partition includes at least one spatial unit, which is the smallest administrative unit to which the power equipment belongs, and each spatial unit includes multiple power equipment. The multi-level anomaly detection module is used to perform anomaly detection based on the metadata of each power device, according to spatial partitioning, spatial unit, and power device level, to identify the target power device with anomalies. An abnormal link and region identification module is used to search for abnormal links and abnormal regions of the target power equipment based on the topological relationship of each power equipment. The abnormal result output module is used to report abnormal target power equipment, abnormal links, and abnormal areas to the dispatching system and visualization platform.
9. An electronic device, characterized in that, include: One or more processors; Storage device for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the power equipment anomaly detection method as described in any one of claims 1-7.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by the processor, the program implements the power equipment anomaly detection method as described in any one of claims 1-7.
Citation Information
Cited By
Abnormal alarm method and device, electronic equipment, storage medium and program product
CN122157462A
Abnormality alarm method and device, electronic equipment, storage medium and program product
CN122157462B
