A Substation Signal Intelligent Monitoring Method and System Based on Intelligent Agents

Abstract

This application discloses a method and system for intelligent substation signal monitoring based on intelligent agents, relating to the field of substation signal monitoring technology. The method includes: constructing physical anomaly transmission relationships corresponding to physical topology links and signal anomaly transmission relationships corresponding to signal topology links based on the physical operating mechanisms and signal transmission mechanisms of power equipment; grouping power equipment within the monitoring range of the same environmental monitoring equipment into the same monitoring set; obtaining the impact of spatial state deviation based on historical signal data and historical spatial states of the same type of power equipment in different monitoring sets; outputting the comprehensive anomaly probability of each power equipment based on real-time signal data, the impact of spatial state deviation, physical anomaly transmission relationships, and signal anomaly transmission relationships; and outputting anomaly location results based on the comprehensive anomaly probability using the inference mechanism of the intelligent agent. This application balances anomaly perception sensitivity with alarm output reliability.

Description

Technical Field

[0001] This application relates to the field of substation signal monitoring technology, and in particular to a method and system for intelligent substation signal monitoring based on intelligent agents. Background Technology

[0002] As the core hub of power system for energy collection, distribution, and voltage transformation, the operating status of substations directly affects the safety and stability of the power grid. To ensure the safe and reliable operation of substation equipment, online monitoring and control systems are widely used to monitor the operating status of primary and secondary equipment in real time. Existing substation monitoring systems are typically based on data acquisition and monitoring control systems, protection information systems, and various online monitoring devices. They continuously collect electrical and non-electrical signals such as voltage, current, power, temperature, pressure, and vibration from the equipment, and compare the collected signals with preset thresholds. When a signal exceeds a limit, an alarm is triggered, and maintenance personnel then troubleshoot and handle the problem based on the alarm information.

[0003] Existing monitoring technologies generally adopt a single threshold or static threshold judgment mechanism, that is, a fixed upper and lower threshold is set for each monitoring signal, and an alarm is triggered when the signal exceeds the threshold. However, fixed thresholds cannot adapt to equipment aging or environmental fluctuations, and false alarms are prone to occur.

[0004] Furthermore, equipment within a substation is physically coupled through electrical connections, mechanical connections, and heat conduction. An anomaly in one piece of equipment can affect adjacent equipment via these physical paths. Simultaneously, equipment status signals are transmitted to the monitoring system via current transformers, merging units, and communication networks. When an anomaly occurs in any link, current technology struggles to distinguish whether the anomaly stems from actual degradation of the primary equipment itself, distortion in the signal acquisition and transmission link, or a cascading effect caused by anomalies in adjacent equipment transmitted through physical coupling. This limitation restricts the system's accuracy in locating complex anomalies. Summary of the Invention

[0005] This application addresses the problem in related technologies where substation signal monitoring struggles to balance anomaly detection sensitivity and alarm output reliability. It provides an intelligent substation signal monitoring method and system based on intelligent agents. By constructing a comprehensive anomaly probability calculation based on the influence of spatial state deviation, physical anomaly transmission relationships, and signal anomaly transmission relationships, it can both utilize the influence of spatial state deviation to eliminate environmental influences and improve anomaly detection sensitivity, and combine physical anomaly and signal anomaly bidirectional positioning of real abnormal devices to improve alarm output reliability.

[0006] To achieve the above technical objectives, this application provides a technical solution: a substation signal intelligent monitoring method based on intelligent agents, comprising the following steps: determining the physical topology links and signal topology links between power equipment according to the substation topology map; constructing the physical anomaly transmission relationship corresponding to the physical topology links and the signal anomaly transmission relationship corresponding to the signal topology links by combining the physical operation mechanism and signal transmission mechanism of the power equipment; grouping power equipment within the monitoring range of the same environmental monitoring equipment into the same monitoring set; obtaining the spatial state deviation impact based on the historical signal data and historical spatial state of the same type of power equipment in different monitoring sets; acquiring real-time signal data; outputting the comprehensive anomaly probability of each power equipment based on the real-time signal data, the spatial state deviation impact, the physical anomaly transmission relationship, and the signal anomaly transmission relationship; and outputting the anomaly location result based on the comprehensive anomaly probability using the inference mechanism of the intelligent agent.

[0007] Furthermore, the construction of physical anomaly conduction relationships corresponding to physical topology links and signal anomaly conduction relationships corresponding to signal topology links, based on the physical operation mechanisms of electrical conduction, electromagnetic coupling, thermal diffusion, and mechanical linkage between moving devices, includes: obtaining the first anomaly conduction path corresponding to the abnormal action of each moving device on the physical topology link, and the first anomaly conduction weight of each node on the first anomaly conduction path, based on the physical operation mechanisms of electrical conduction, electromagnetic coupling, thermal diffusion, and mechanical linkage between moving devices; constructing physical anomaly conduction relationships using the first anomaly conduction path and the first anomaly conduction weight; and constructing signal anomaly conduction relationships using the complete signal transmission path between moving devices and moving monitoring devices as the second anomaly conduction path, based on the signal transmission mechanisms of time delay distortion and information attenuation during signal transmission.

[0008] Furthermore, determining the physical topology links and signal topology links between power equipment based on the substation topology diagram includes: forming physical topology links based on the primary electrical circuit connection relationships of the operating equipment in the substation topology diagram; and forming signal topology links based on the sampling associations between the operating equipment and the motion monitoring equipment, the signal transmission associations between the motion monitoring equipment, and the interlocking control associations between the motion monitoring equipment and the operating equipment in the substation topology diagram.

[0009] Furthermore, the step of grouping power equipment within the monitoring range of the same environmental monitoring equipment into the same monitoring set, and obtaining the spatial state deviation influence based on historical signal data and historical spatial states of the same type of power equipment in different monitoring sets includes: constructing spatial clustering constraints based on the monitoring range of the environmental monitoring equipment; performing spatial clustering analysis on all power equipment with the deployment location of the environmental monitoring equipment as the cluster center to obtain several clusters, and taking each cluster as a monitoring set; extracting historical signal data and historical spatial states corresponding to the same type of power equipment in different monitoring sets under the same environmental monitoring data conditions; and constructing the spatial state deviation influence based on the offset and fluctuation characteristics of historical signal data as historical spatial states change.

[0010] Furthermore, the acquisition of real-time signal data and the output of the comprehensive anomaly probability of each power device based on the influence of real-time signal data and spatial state deviation, physical anomaly transmission relationship, and signal anomaly transmission relationship include: responding to a power device anomaly alarm, determining the alarm action device that triggered the anomaly alarm and the corresponding alarm action monitoring device; determining physically associated devices with physical topology association with the alarm action device based on physical anomaly transmission relationship; determining signal associated devices with signal topology association with the alarm action monitoring device based on signal anomaly transmission relationship; calculating the current association deviation based on the inherent association relationship between the real-time signal data collected by the alarm action monitoring device and the alarm action device, and outputting a first anomaly probability based on the degree of matching between the current association deviation and the influence of spatial state deviation; judging the degree of matching between the real-time signal data of the physically associated device and the physical anomaly transmission relationship, and outputting a second anomaly probability; judging the degree of matching between the real-time signal data of the signal associated device and the signal anomaly transmission relationship, and outputting a third anomaly probability; and fusing the first anomaly probability, the second anomaly probability, and the third anomaly probability to obtain the comprehensive anomaly probability.

[0011] Furthermore, the step of determining the degree of matching between the real-time signal data of the physically associated device and the physical anomaly transmission relationship, and outputting the second anomaly probability, includes: outputting the expected signal data of the physically associated device based on the real-time signal data of the alarm action device; and outputting the second anomaly probability based on the degree of matching between the real-time signal data of the physically associated device and the expected signal data, combined with the physical anomaly transmission relationship.

[0012] Furthermore, by using evidence theory to fuse the first, second, and third anomaly probabilities, a comprehensive anomaly probability is obtained.

[0013] Furthermore, determining the physical topology links and signal topology links between power equipment based on the substation topology map also includes: obtaining primary action monitoring devices based on the overlapping nodes of the physical topology links and signal topology links; performing correlation analysis on the historical signal data of all primary action monitoring devices to construct the signal trend correlation relationship of the primary action monitoring devices.

[0014] Furthermore, the step of outputting the comprehensive anomaly probability of each power device based on the influence of real-time signal data and spatial state deviation, physical anomaly transmission relationship, and signal anomaly transmission relationship includes: outputting the signal monitoring reliability of the primary action monitoring device based on the matching degree of the real-time signal data and signal trend correlation of all primary action monitoring devices; using the primary action monitoring device with a signal monitoring reliability greater than the preset signal monitoring reliability benchmark as the matching benchmark, outputting the comprehensive anomaly probability of each power device based on the spatial relationship, physical anomaly transmission relationship, and signal anomaly transmission relationship between the other power devices, action monitoring devices, and primary action monitoring devices; and treating the primary action monitoring device with a signal monitoring reliability less than or equal to the preset signal monitoring reliability benchmark as a monitoring anomaly device, and correcting the alarm result of the alarm action device based on the physical anomaly transmission relationship and signal anomaly transmission relationship between the alarm action device, the alarm action monitoring device, and the monitoring anomaly device.

[0015] Another technical solution provided in this application is a substation signal intelligent monitoring system based on an intelligent agent, which is applicable to the method described above, including: a link derivation module, used to determine the physical topology links and signal topology links between power equipment according to the substation topology map, and to construct the physical anomaly transmission relationship corresponding to the physical topology link and the signal anomaly transmission relationship corresponding to the signal topology link by combining the physical operation mechanism and signal transmission mechanism of the power equipment; a state analysis module, used to group power equipment within the monitoring range of the same environmental monitoring equipment into the same monitoring set, and to obtain the spatial state deviation influence based on the historical signal data and historical spatial state of the same type of power equipment in different monitoring sets; a probability analysis module, used to output the comprehensive anomaly probability of each power equipment based on the real-time signal data and the spatial state deviation influence, the physical anomaly transmission relationship and the signal anomaly transmission relationship; and an intelligent agent, used to output the anomaly location result based on the comprehensive anomaly probability using a reasoning mechanism.

[0016] The beneficial effects of this application are as follows: 1. By constructing a dual-link anomaly transmission relationship between physical topology links and signal topology links, dual anomaly identification of primary hardware physical transmission anomalies and secondary signal interlocking transmission anomalies between power equipment is achieved. This facilitates subsequent identification of hidden anomalies by combining the two types of anomalies, preventing the overlooking of equipment that appears normal in terms of monitoring signals but is actually abnormal, and improving the accuracy of anomaly monitoring. Simultaneously, by dividing the monitoring set within the same environmental monitoring coverage area, and using historical signal data and historical spatial states of similar equipment in different sets for joint analysis, the impact of spatial state deviation is quantified to characterize the signal baseline value shift caused by differences in equipment spatial states. Furthermore, when performing anomaly probability analysis, multi-dimensional constraints of spatial state deviation, physical anomaly transmission, and signal anomaly transmission are integrated to comprehensively calculate the comprehensive anomaly probability of power equipment, achieving a quantitative assessment of the degree of anomaly under multi-constraint coupling. Anomaly location is then performed using the comprehensive anomaly probability, balancing anomaly perception sensitivity and alarm output reliability.

[0017] 2. The current correlation deviation is calculated by comparing the real-time signal data collected by the alarm action monitoring device with the inherent correlation relationship of the alarm action device. The first anomaly probability is output based on the matching degree between the current correlation deviation and the influence of the spatial state deviation. The current performance of the power equipment is compared with the expected performance determined by the physical mechanism of the power equipment itself, and the systematic deviation caused by the difference in spatial state is filtered out. If the current correlation deviation exceeds the influence of the spatial state deviation, it indicates that the signal anomaly of the alarm action device is not caused by environmental anomaly. This avoids the real alarm caused by early equipment deterioration being misjudged as a false alarm caused by environmental factors, and improves the accuracy of anomaly monitoring. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the intelligent monitoring method for substation signals based on intelligent agents proposed in this application.

[0019] Figure 2 This is a schematic diagram illustrating the process of constructing the comprehensive anomaly probability in one embodiment of the intelligent monitoring method for substation signals based on intelligent agents in this application. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description of this application is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely one preferred embodiment of this application and are only used to explain this application. They do not limit the scope of protection of this application. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0021] like Figure 1As shown in the first embodiment of this application, the intelligent monitoring method for substation signals based on intelligent agents includes the following steps: Based on the substation topology diagram, determine the physical topology links and signal topology links between power equipment. Combine the physical operation mechanism and signal transmission mechanism of the power equipment to construct the physical anomaly transmission relationship corresponding to the physical topology link and the signal anomaly transmission relationship corresponding to the signal topology link. Power equipment within the monitoring range of the same environmental monitoring equipment is grouped into the same monitoring set, and the impact of spatial state deviation is obtained based on historical signal data and historical spatial state of the same type of power equipment in different monitoring sets. Acquire real-time signal data, and output the comprehensive anomaly probability of each power device based on the influence of real-time signal data and spatial state deviation, physical anomaly transmission relationship, and signal anomaly transmission relationship. The anomaly localization result is output based on the comprehensive anomaly probability using the inference mechanism of the intelligent agent.

[0022] In this embodiment, by constructing a dual-link anomaly transmission relationship between physical topology links and signal topology links, dual anomaly identification of primary hardware physical transmission anomalies and secondary signal interlocking transmission anomalies between power equipment is achieved. This facilitates subsequent identification of hidden anomalies by combining the two anomaly scenarios, preventing the overlooking of devices that appear normal in monitoring signals but are actually abnormal, thus improving the accuracy of anomaly monitoring. Simultaneously, by dividing the monitoring set within the same environmental monitoring coverage area, historical signal data and historical spatial states of similar devices in different sets are jointly analyzed to quantify the impact of spatial state deviation, characterizing the signal baseline value offset caused by differences in device spatial states. Furthermore, during anomaly probability analysis, multi-dimensional constraints of spatial state deviation, physical anomaly transmission, and signal anomaly transmission are integrated to comprehensively calculate the overall anomaly probability of power equipment. This achieves a quantitative assessment of the anomaly degree under multi-constraint coupling, and anomaly location is performed using the comprehensive anomaly probability, balancing anomaly perception sensitivity and alarm output reliability.

[0023] Specifically, determining the physical topology links and signal topology links between power equipment based on the substation topology diagram includes: Based on the primary electrical circuit connection relationship of the operating equipment in the substation topology diagram, a physical topology link is formed; Based on the sampling association between the operating equipment and the operating monitoring equipment, the signal transmission association between the operating monitoring equipment, and the interlocking control association between the operating equipment and the operating equipment in the substation topology diagram, a signal topology link is formed.

[0024] In this embodiment, the power equipment includes operating devices and operating monitoring devices. Operating devices are high-voltage equipment that directly participates in the generation, transmission, distribution, and conversion of electrical energy, forming the primary electrical circuit of the substation. Operating monitoring devices are devices that collect, transmit, process, judge, and control the operating status of the primary equipment, forming the secondary information system of the substation.

[0025] By dividing physical topology links based on primary electrical circuit connections and signal topology links based on equipment sampling association, monitoring signal transmission association, and interlocking control association, the substation's primary physical hardware links and secondary signal interaction links are decoupled in a hierarchical manner. This distinguishes between the physical hardware transmission relationship between equipment and the monitoring signal transmission and interlocking linkage relationship, determines the propagation source and transmission path of anomalies under the two types of links, and improves the reliability of identifying hidden dangers where the actual equipment itself is abnormal but the apparent monitoring signal is normal.

[0026] Furthermore, by combining the physical operating mechanism and signal transmission mechanism of power equipment, the physical anomaly propagation relationship corresponding to the physical topology link and the signal anomaly propagation relationship corresponding to the signal topology link are constructed, including: Based on the physical operating mechanism of electrical conduction, electromagnetic coupling, thermal diffusion and mechanical linkage between the actuators, the first abnormal conduction path corresponding to the abnormal action of each actuator on the physical topology link is obtained, as well as the first abnormal conduction weight of each node on the first abnormal conduction path. The physical abnormal conduction relationship is constructed with the first abnormal conduction path and the first abnormal conduction weight. Based on the signal transmission mechanism of time delay distortion and information attenuation during signal transmission, the complete signal transmission path between the motion device and the motion monitoring device is used as the second abnormal transmission path. The second abnormal transmission weight of each node on the second abnormal transmission path is obtained, and the signal abnormal transmission relationship is constructed using the second abnormal transmission path and the second abnormal transmission weight.

[0027] By combining multiple physical mechanisms such as electrical conduction, electromagnetic coupling, thermal diffusion, and mechanical linkage, a first anomaly conduction path and corresponding first anomaly conduction weight are constructed to quantify the propagation range, conduction intensity, and impact of each anomalous device's own anomaly in the primary hardware topology. Simultaneously, a second anomaly conduction path and second anomaly conduction weight are constructed by combining signal transmission delay distortion and information attenuation mechanisms to characterize the transmission law and information loss characteristics of the monitoring signal in the secondary transmission link. By quantifying the attenuation of anomaly energy propagating along the physical path and the accumulation of distortion along the signal transmission path, the apparent impact of anomalous devices and motion monitoring devices on other devices is quantified from both hardware conduction and software transmission perspectives.

[0028] Specifically, electrical conduction refers to the direct transmission of electrical energy through conductors; electromagnetic coupling refers to the energy transfer between transformer windings or between the primary and secondary sides of instrument transformers via magnetic fields; thermal diffusion refers to the heat conduction from abnormal temperature rise hotspots along busbars, equipment chambers, and metal frames; and mechanical linkage refers to the transmission of mechanical forces between circuit breaker operating mechanisms and disconnector transmission rods. Time delay distortion refers to message delay, time synchronization error, sampling value jitter, and harmonic aliasing that exist during signal transmission; and information attenuation refers to signal amplitude loss and message loss that exist during signal transmission.

[0029] In this embodiment, the physical operation mechanism is obtained through analysis based on the inherent structural parameters of the operating equipment and historical fault transmission data. The signal transmission mechanism is obtained through analysis based on the link communication transmission characteristics, the inherent attributes of the communication protocol, historical signal transmission timing data, and message transmission data.

[0030] It is understandable that a complete signal transmission path is a full-link signal flow channel formed after the operating signal of the motion monitoring equipment is sampled and shaped by the motion monitoring equipment. It includes at least the initial signal transmission from the motion monitoring equipment, the horizontal transmission between multiple motion monitoring equipment, the uplink convergence transmission of signals, the downlink distribution transmission of control commands, and the end transmission of interlocking control commands.

[0031] Power equipment within the same environmental monitoring equipment's monitoring range is grouped into the same monitoring set. The impact of spatial state deviation is obtained based on historical signal data and historical spatial states of the same type of power equipment in different monitoring sets, including: Spatial clustering constraints are constructed based on the monitoring range of environmental monitoring equipment. Spatial clustering analysis is performed on all power equipment with the deployment location of environmental monitoring equipment as the cluster center to obtain several clusters. Each cluster is used as a monitoring set. Historical signal data and historical spatial status of the same type of power equipment under the same environmental monitoring data conditions are extracted from different monitoring sets. Based on the offset and fluctuation characteristics of historical signal data as historical spatial status changes, the influence of spatial status deviation is constructed.

[0032] In this embodiment, the environmental monitoring equipment includes at least a temperature and humidity sensor. Using the three-dimensional coordinates of the temperature and humidity sensor as the cluster center, if the spatial distance between the power equipment and the temperature and humidity sensor is within the monitoring range of the sensor, then the power equipment is assigned to the monitoring set corresponding to that temperature and humidity sensor.

[0033] Spatial clustering constraints include at least a spatial distance threshold and an isolation threshold. The spatial distance threshold is set based on the monitoring radius of the environmental monitoring equipment. The isolation threshold is 0, meaning that the power equipment belongs to the monitoring set corresponding to the environmental monitoring equipment only when the spatial distance between the power equipment and the environmental monitoring equipment is within the monitoring range of the environmental monitoring equipment and there is no isolation between them.

[0034] Among them, the clustering algorithm can be either the constrained K-Means algorithm or the DBSCAN algorithm (density-based spatial clustering algorithm).

[0035] It is understandable that spatial clustering constraints can be set according to the monitoring needs of environmental monitoring equipment. When a power device falls within the monitoring range of multiple environmental monitoring devices at the same time, it will be assigned to the corresponding monitoring set with the highest monitoring accuracy.

[0036] Spatial conditions include at least the cabinet's orientation, ventilation conditions, degree of obstruction, altitude, whether it is against a wall, and IP protection level. Historical signal data and historical spatial conditions for the same type of electrical equipment from different monitoring sets are extracted under the same environmental monitoring data conditions. Based on the offset and fluctuation characteristics of historical signal data as a function of historical spatial conditions, the influence of spatial condition deviations is constructed as follows: Based on the same environmental monitoring data conditions, historical signal data and historical spatial status data corresponding to the same type of power equipment in different monitoring sets are grouped into the same comparison sample set; Perform inherent correlation analysis between historical input signals and historical output signals on the same set of comparison samples to obtain the correlation deviation of each comparison sample. Combine the correlation deviation with historical spatial state data to perform offset fluctuation characteristic analysis and obtain the influence of spatial state deviation.

[0037] Since the same comparison sample set contains historical signal data and historical spatial state data of the same type of power equipment, and the environmental monitoring data conditions corresponding to each sample are the same, ideally, the historical input signals and historical output signals of each power equipment within the same comparison sample set should exhibit the same inherent correlation characteristics. However, due to the spatial state differences of the monitoring sets to which each power equipment belongs, the actual output signals of different equipment under the same input signal conditions exhibit a certain degree of correlation deviation relative to the inherent correlation characteristics. Based on this, an inherent correlation relationship is first established based on the overall historical input signals and historical output signals within the same comparison sample set. Then, the correlation deviation of historical signal data relative to the inherent correlation relationship under each spatial state condition is calculated separately. Based on the offset fluctuation characteristics of the historical spatial state changes of this correlation deviation value, the influence of spatial state deviation is constructed.

[0038] By establishing a comparison sample set under fixed environmental conditions and selecting equipment of the same type across sets, and relying on the inherent global correlation patterns of power equipment input and output signals, the inherent operating characteristics of the equipment are stripped away, and the pure correlation deviation caused solely by spatial state differences is extracted. The fluctuation characteristics of the pure correlation deviation with spatial state data are used to construct a spatial state deviation model corresponding to different types of power equipment, eliminating non-fault signal offsets caused by differences in the spatial layout of power equipment, and avoiding misjudgments of equipment anomalies due to spatial differences.

[0039] Specifically, the correlation bias is the residual between the actual historical output signal and the predicted output signal corresponding to the inherent correlation.

[0040] This process is performed before the comprehensive anomaly probability calculation, and continuously outputs the anomaly monitoring results of all environmental monitoring equipment based on real-time environmental monitoring data from all environmental monitoring equipment.

[0041] Although environmental monitoring equipment at different locations within the same substation may display different readings, the deviations are not excessive. Therefore, the historical environmental monitoring data of all equipment is obtained, showing the changing patterns as the equipment's location changes. Based on whether the continuity of real-time environmental monitoring data conforms to these patterns, it is determined whether any current environmental monitoring equipment malfunctions.

[0042] like Figure 2 As shown, in one embodiment, the substation signal intelligent monitoring system is integrated into the existing equipment anomaly alarm system of the substation. Real-time signal data acquisition is only performed when the substation outputs a power equipment anomaly alarm. Correspondingly, the real-time signal data is acquired, and the comprehensive anomaly probability of each power device is output based on the influence of the real-time signal data and spatial state deviation, the physical anomaly propagation relationship, and the signal anomaly propagation relationship. This includes: In response to an abnormal alarm from a power equipment, determine the alarm-acting device that triggered the abnormal alarm and the corresponding alarm-acting monitoring device; Based on the physical anomaly propagation relationship, identify the physically associated devices that have a physical topological association with the alarm action devices; Based on the abnormal signal propagation relationship, identify signal-related devices that have a signal topology association with the alarm action monitoring device; The current correlation deviation is calculated based on the real-time signal data collected by the alarm action monitoring device and the inherent correlation relationship between the alarm action device and the device. The first anomaly probability is output based on the degree of matching between the current correlation deviation and the spatial state deviation. Based on the real-time signal data of the physically associated devices, determine the degree of matching between the data and the physical anomaly transmission relationship, and output the second anomaly probability. Based on the real-time signal data of the signal association device, determine the degree of matching between the signal and the abnormal signal transmission relationship, and output the third abnormal probability; The first, second, and third anomaly probabilities are combined to obtain the comprehensive anomaly probability.

[0043] In this embodiment, by simultaneously identifying the alarm-initiating device and its corresponding alarm-initiating monitoring device when responding to an abnormal alarm of power equipment, and splitting the subsequent monitoring and verification path into physical link tracing starting from the primary device and signal link tracing starting from the secondary monitoring device, dual-entry decoupled analysis is achieved. The authenticity of the alarm is independently verified from the energy transmission dimension and the information transmission dimension, respectively, thereby improving the accuracy of abnormal monitoring.

[0044] Meanwhile, the current correlation deviation is calculated by comparing the real-time signal data collected by the alarm action monitoring device with the inherent correlation relationship of the alarm action device. The first anomaly probability is output based on the matching degree between the current correlation deviation and the influence of the spatial state deviation. The current performance of the power equipment is compared with the expected performance determined by the physical mechanism of the power equipment itself, and the systematic deviation caused by the difference in spatial state is filtered out. If the current correlation deviation exceeds the influence of the spatial state deviation, it indicates that the signal anomaly of the alarm action device is not caused by environmental anomaly. This avoids the real alarm caused by early equipment deterioration being misjudged as a false alarm caused by environmental factors, and improves the accuracy of anomaly monitoring.

[0045] Specifically, the spatial state deviation influence range is obtained based on the current spatial state of the alarm action device and the spatial state deviation influence range. If the current associated deviation is within the spatial state deviation influence range, the first anomaly probability is 0. If the current associated deviation is not within the spatial state deviation influence range, the minimum absolute difference between the current associated deviation and the spatial state deviation influence range is calculated, and the first anomaly probability is obtained by the ratio of the minimum absolute difference to the width of the spatial state deviation influence range.

[0046] By combining the spatial status of the alarm action device to determine the spatial status deviation influence range, when the signal deviation, i.e. the current associated deviation, is within the spatial status deviation influence range, the abnormal probability is directly set to zero, eliminating the systematic signal offset caused by the spatial location environment. When the signal deviation exceeds the spatial status deviation influence range, the abnormal probability of the device itself is quantified according to the degree of exceedance, distinguishing between spatial environment interference deviation and power equipment abnormality, and avoiding misjudgment of occasional real alarms caused by early hidden degradation of the equipment.

[0047] Furthermore, based on the real-time signal data of the physically associated devices, the degree of matching between the data and the physical anomaly transmission relationship is determined, and the second anomaly probability is output, including: Output the desired signal data of the physically associated devices based on the real-time signal data of the alarm action devices; Based on the degree of matching between the real-time signal data and the expected signal data of the physically associated devices, and combined with the physical anomaly transmission relationship, a second anomaly probability is output.

[0048] In this embodiment, physically associated devices are obtained based on the first abnormal transmission path containing alarm-acting devices in the physical abnormal transmission relationship. Expected signal data for the physically associated devices is output based on the hardware action relationship between the alarm-acting devices and the physically associated devices. The degree of action deviation of each physically associated device is calculated using the expected signal data and real-time signal data; that is, the absolute difference between the expected signal data and the real-time signal data is calculated, and the ratio of the absolute difference to the expected signal data is used as the degree of action deviation. The comprehensive deviation is obtained by weighting the degree of action deviation with the first abnormal transmission weight. A monotonically decreasing mapping function is used to map the comprehensive deviation to a second abnormal probability. The larger the comprehensive deviation, the greater the difference between the actual operation of the physically associated device and the physical abnormal transmission relationship, the less support the alarm has from real physical transmission, the lower the anomaly confidence, and the smaller the second abnormal probability value.

[0049] Correspondingly, based on the real-time signal data of the signal-associated device, the degree of matching between the data and the abnormal signal transmission relationship is determined, and the third anomaly probability is output, including: Based on the real-time signal data of the alarm action monitoring device and the relationship between signal anomaly transmission, the expected signal data of the signal association device is output. Based on the degree of matching between the real-time signal data of the signal-associated device and the expected signal data, and combined with the signal anomaly transmission relationship, a third anomaly probability is output.

[0050] The signal deviation of each signal-associated device is calculated based on the expected signal data and the real-time signal data. This involves calculating the absolute difference between the expected signal data and the real-time signal data, and using the ratio of the absolute difference to the expected signal data as the signal deviation. The signal deviation of each device is then weighted and summed with the corresponding second anomaly propagation weight to obtain the comprehensive deviation. A monotonically decreasing mapping function is used to map the comprehensive deviation to the third anomaly probability. The larger the comprehensive deviation, the lower the degree of matching between the transmission characteristics of the signal-associated device and the signal anomaly propagation relationship, the lower the reliability of the alarm anomaly, and the smaller the third anomaly probability.

[0051] As a feasible implementation, the first anomaly probability, the second anomaly probability, and the third anomaly probability are weighted and fused according to a preset confidence weight to obtain a comprehensive anomaly probability.

[0052] As another feasible implementation, the first anomaly probability, the second anomaly probability, and the third anomaly probability are fused using evidence theory to obtain the comprehensive anomaly probability.

[0053] Specifically, construct the recognition framework: ; in, Represents the recognition framework; This indicates that the alarm was caused by a real abnormality in the alarm action device itself, and that the corresponding alarm action monitoring device is working normally. This indicates that the alarm was caused by a sampling or transmission abnormality in the alarm action monitoring device itself, and the alarm action device itself is normal. This indicates that the alarm was caused by a real anomaly in the upstream equipment being transmitted along the physical link to the alarm-acting device. The alarm-acting device itself is normal, and its alarm-acting monitoring equipment is working normally.

[0054] Constructing the basic probability assignment function for the first anomaly probability: ; in, The basic probability assignment function representing the first anomaly probability. This represents the probability of the first anomaly.

[0055] Constructing the basic probability assignment function for the second anomaly probability: ; ; in, The basic probability assignment function representing the second anomaly probability. This indicates the probability of the second anomaly.

[0056] Constructing the basic probability assignment function for the third anomaly probability: ; ; in, The basic probability assignment function representing the probability of the third anomaly. This indicates the probability of the third anomaly.

[0057] Constructing the conflict coefficient: ; in, This represents the conflict coefficient.

[0058] Constructing the basic probability assignment for synthesis: ; ; ; in, express The basic probability of synthesis, express The basic probability of synthesis, express The basic probability of synthesis.

[0059] At this point, the overall anomaly probability is... , as well as .

[0060] In this embodiment, a first anomaly probability is constructed by combining the spatial state deviation influence interval with the current correlation deviation. This distinguishes whether signal anomalies can be explained by the spatial location of the equipment or differences in the local environment, intercepting false alarms caused by environmental differences and preventing power equipment from being falsely alarmed when there are no real anomalies. Furthermore, an identification framework is constructed that includes three types of causes: equipment-related anomalies, anomalies of the monitoring equipment itself, and anomalies propagated from upstream equipment. Basic probability allocation and multi-evidence fusion calculation are performed by combining the anomaly probabilities of each type. The global confidence level of the three types of anomalies is determined by the first anomaly probability. Through conflict coefficient correction and normalization fusion calculation, the true cause of the alarm is distinguished, improving the accuracy of anomaly monitoring.

[0061] Then, the agent performs anomaly localization based on the maximum value among the comprehensive anomaly probabilities. When At its maximum, it indicates a genuine anomaly in the alarm action itself and outputs the physical associated device with a synchronization anomaly; when At its maximum, the alarm action monitoring device is identified as having a sampling or transmission anomaly. The intelligent agent traces downstream along the signal topology link, marking all related devices that depend on the monitoring device's signal and exhibit logical inconsistencies as synchronization anomalies. When the maximum value is reached, the anomaly is located as being transmitted from an upstream device to this device along the physical link. The agent searches upstream in reverse along the physical topology link, identifies the node with the highest overall anomaly probability on the first transmission path that has not been interpreted by other upstream devices as the true root cause of the anomaly, and marks the alarm action device as the device that affects the alarm.

[0062] In another scenario of this embodiment, the substation signal intelligent monitoring system adopts a full-domain, normalized real-time monitoring mode. It does not rely on existing equipment anomaly alarms within the substation for triggering, but actively collects real-time signal data from all power equipment. In this case, based on the acquired real-time signal data and the influence of spatial state deviations, physical anomaly propagation relationships, and signal anomaly propagation relationships, the system outputs the comprehensive anomaly probability for each power device, including: The real-time signal data of each power device is corrected based on the influence of real-time environmental monitoring data and spatial state deviation. The comprehensive anomaly probability of each power device is output based on the degree of matching between the corrected real-time signal data and the physical anomaly transmission relationship and the signal anomaly transmission relationship.

[0063] In this scenario, real-time environmental monitoring data and the spatial state deviation influence range corresponding to each power device are first used to perform pre-deviation correction on the original real-time signal data of each power device. This process eliminates systematic signal offsets caused by differences in device spatial location and regional differences between the local environment and environmental monitoring devices, restoring corrected real-time signal data that reflects the true operating state of the power devices. Based on the degree of matching between the corrected real-time signal data and the physical anomaly transmission relationship and the signal anomaly transmission relationship, the comprehensive anomaly probability of each link is output.

[0064] Specifically, the hardware conduction offset is obtained by using the corrected real-time signal data of each power device on the first abnormal conduction path and the first abnormal conduction weight. The signal conduction offset is obtained by using the corrected real-time signal data of each power device on the second abnormal conduction path and the second abnormal conduction weight. For the same power device, the hardware conduction offset and the signal conduction offset are combined for calculation to output the comprehensive abnormal probability of each power device.

[0065] Then, the agent includes power devices whose overall anomaly probability exceeds a preset screening threshold into a candidate anomaly set. Subsequently, for power devices in the candidate anomaly set, it determines whether to perform source tracing inference based on whether multiple power devices are on the same physical topology link or the same signal topology link. If multiple power devices are on the same physical topology link or the same signal topology link, the source device of the anomaly is calculated based on the overall anomaly probability and anomaly weight, and the source anomaly location result is output.

[0066] As a second embodiment of this application, since some equipment in the substation lines possesses both power equipment characteristics and motion monitoring equipment characteristics, determining the physical topology links and signal topology links between power equipment based on the substation topology diagram further includes: A single action monitoring device is obtained based on the overlapping nodes of the physical topology link and the signal topology link; A correlation analysis was performed on the historical signal data of all primary motion monitoring devices to construct the signal trend correlation relationship of the primary motion monitoring devices.

[0067] At this point, after acquiring the real-time signal data, execute: Based on the degree of matching between the real-time signal data and signal trend correlation of all primary motion monitoring devices, the signal monitoring reliability of the primary motion monitoring devices is output. Using a primary action monitoring device whose signal monitoring reliability is greater than the preset signal monitoring reliability benchmark as the matching benchmark, the comprehensive anomaly probability of each power device is output based on the spatial relationship, physical anomaly transmission relationship and signal anomaly transmission relationship between the other power devices, the action monitoring devices and the primary action monitoring devices; A single action monitoring device whose signal monitoring reliability is less than or equal to the preset signal monitoring reliability benchmark is considered an abnormal monitoring device. The alarm result of the alarm action device is corrected based on the physical abnormal transmission relationship and signal abnormal transmission relationship between the alarm action device, the alarm action monitoring device and the abnormal monitoring device.

[0068] In this embodiment, the primary motion monitoring device can be a current transformer. During the real-time monitoring phase, based on the degree of matching between the real-time signal data of each primary motion monitoring device and the expected value derived from the group of devices in the correlation between the signal trend, the signal monitoring reliability of each device is calculated, and they are divided into two categories, high-reliability devices and low-reliability devices, according to a preset reliability benchmark.

[0069] For a high-reliability primary motion monitoring device, historical correlation deviations under its current environmental monitoring data are retrieved. These historical correlation deviations are then sorted according to the matching degree between historical and real-time signal data. Historical correlation deviations of other power devices are retrieved in the same time sequence. Based on the sorting, the historical correlation deviations of the remaining power devices are weighted and calculated to obtain a baseline correlation deviation value. The first anomaly probability is output based on the difference between the current correlation deviation of the remaining power devices and the baseline correlation deviation value. Furthermore, the second and third anomaly probabilities are output based on the matching degree between the actual physical conduction relationship and the physical anomaly conduction relationship, and between the actual signal anomaly conduction relationship and the signal anomaly conduction relationship between the remaining power devices and the primary motion monitoring device.

[0070] For a low-reliability primary motion monitoring device, it is marked as an abnormal monitoring device, and all signal-related devices affected by its signal are traced downstream along the signal topology link. The abnormal judgment results based on the unreliable signal are corrected or revoked. At the same time, the response of neighboring devices in the physical topology link is combined to correct the root cause from the primary device abnormality to the monitoring device itself abnormality, thereby suppressing false alarms caused by the distortion of the monitoring source and accurately locating the monitoring device fault.

[0071] The identification of primary motion monitoring devices is achieved through overlapping nodes in dual-topology links. Signal trend correlations are constructed based on historical data of the primary motion monitoring device group, and the reliability of device signal monitoring is graded and assessed by combining the expected value of the group projection. For high-reliability primary motion monitoring devices, a weighted calculation is performed based on their temporal historical correlation deviation to obtain a deviation benchmark value, achieving accurate quantification of the first anomaly probability. The second and third anomaly probabilities are calculated by combining physical and signal dual-dimensional transmission matching characteristics, ensuring reliable signal benchmark support for probability calculations and improving the accuracy of anomaly quantification results. For low-reliability primary motion monitoring devices with distorted signals, all associated devices affected by signal contamination are traced along the signal topology link. Related anomaly judgment results are corrected, and cross-validation is performed by combining the operational responses of adjacent devices in the physical topology. This re-locates the misjudgment of primary device anomalies caused by source distortion as a fault in the monitoring device itself. This suppresses cascading false alarms caused by sampling distortion and signal anomaly contamination of primary monitoring devices, improving the accuracy of station-wide anomaly monitoring and root cause localization.

[0072] As a third embodiment of this application, a substation signal intelligent monitoring system based on intelligent agents includes: The link derivation module is used to determine the physical topology links and signal topology links between power equipment based on the substation topology diagram, and to construct the physical anomaly transmission relationship corresponding to the physical topology link and the signal anomaly transmission relationship corresponding to the signal topology link by combining the physical operation mechanism and signal transmission mechanism of the power equipment. The status analysis module is used to group power equipment within the monitoring range of the same environmental monitoring equipment into the same monitoring set, and to obtain the impact of spatial status deviation based on historical signal data and historical spatial status of the same type of power equipment in different monitoring sets. The probability analysis module is used to output the comprehensive anomaly probability of each power device based on the influence of real-time signal data and spatial state deviation, physical anomaly transmission relationship, and signal anomaly transmission relationship. An intelligent agent is used to output anomaly localization results based on the anomaly probability by using an inference mechanism.

[0073] In this embodiment, the link deduction module and the state analysis module are connected to the probability analysis module, and the link deduction module and the probability analysis module are connected to the intelligent agent.

[0074] It is understood that the agent in this embodiment is a decision-making unit with an embedded reasoning mechanism.

[0075] The specific embodiments described above are preferred embodiments of the intelligent monitoring method and system for substation signals based on intelligent agents in this application, and are not intended to limit the specific scope of this application. The scope of this application includes but is not limited to the specific embodiments described above. All equivalent changes made in accordance with the shape and structure of this application are within the protection scope of this application.

Claims

1. A method for intelligent monitoring of substation signals based on intelligent agents, characterized in that: Includes the following steps: Based on the substation topology diagram, determine the physical topology links and signal topology links between power equipment. Combine the physical operation mechanism and signal transmission mechanism of the power equipment to construct the physical anomaly transmission relationship corresponding to the physical topology link and the signal anomaly transmission relationship corresponding to the signal topology link. Power equipment within the monitoring range of the same environmental monitoring equipment is grouped into the same monitoring set, and the impact of spatial state deviation is obtained based on historical signal data and historical spatial state of the same type of power equipment in different monitoring sets. Acquire real-time signal data, and output the comprehensive anomaly probability of each power device based on the influence of real-time signal data and spatial state deviation, physical anomaly transmission relationship, and signal anomaly transmission relationship. The anomaly localization result is output based on the comprehensive anomaly probability using the agent's reasoning mechanism. The process of acquiring real-time signal data and outputting the comprehensive anomaly probability of each power device based on the influence of real-time signal data and spatial state deviation, the physical anomaly transmission relationship, and the signal anomaly transmission relationship includes: In response to an abnormal alarm from a power equipment, determine the alarm-acting device that triggered the abnormal alarm and the corresponding alarm-acting monitoring device; Based on the physical anomaly propagation relationship, identify the physically associated devices that have a physical topological association with the alarm action devices; Based on the abnormal signal propagation relationship, identify signal-related devices that have a signal topology association with the alarm action monitoring device; The current correlation deviation is calculated based on the real-time signal data collected by the alarm action monitoring device and the inherent correlation relationship between the alarm action device and the device. The first anomaly probability is output based on the degree of matching between the current correlation deviation and the spatial state deviation. Based on the real-time signal data of the physically associated devices, determine the degree of matching between the data and the physical anomaly transmission relationship, and output the second anomaly probability. Based on the real-time signal data of the signal association device, determine the degree of matching between the signal and the abnormal signal transmission relationship, and output the third abnormal probability; The first, second, and third anomaly probabilities are combined to obtain the comprehensive anomaly probability.

2. The intelligent substation signal monitoring method based on intelligent agents as described in claim 1, characterized in that: The construction of physical anomaly propagation relationships for physical topology links and signal anomaly propagation relationships for signal topology links, combining the physical operation mechanism and signal transmission mechanism of power equipment, includes: Based on the physical operating mechanism of electrical conduction, electromagnetic coupling, thermal diffusion and mechanical linkage between the actuators, the first abnormal conduction path corresponding to the abnormal action of each actuator on the physical topology link is obtained, as well as the first abnormal conduction weight of each node on the first abnormal conduction path. The physical abnormal conduction relationship is constructed with the first abnormal conduction path and the first abnormal conduction weight. Based on the signal transmission mechanism of time delay distortion and information attenuation during signal transmission, the complete signal transmission path between the motion device and the motion monitoring device is used as the second abnormal transmission path. The second abnormal transmission weight of each node on the second abnormal transmission path is obtained, and the signal abnormal transmission relationship is constructed using the second abnormal transmission path and the second abnormal transmission weight.

3. The intelligent substation signal monitoring method based on intelligent agents as described in claim 1 or 2, characterized in that: The process of determining the physical topology links and signal topology links between power equipment based on the substation topology map includes: Based on the primary electrical circuit connection relationship of the operating equipment in the substation topology diagram, a physical topology link is formed; Based on the sampling association between the operating equipment and the operating monitoring equipment, the signal transmission association between the operating monitoring equipment, and the interlocking control association between the operating equipment and the operating equipment in the substation topology diagram, a signal topology link is formed.

4. The intelligent substation signal monitoring method based on intelligent agents as described in claim 1, characterized in that: The step of grouping power equipment within the same environmental monitoring equipment's monitoring range into the same monitoring set, and obtaining the spatial state deviation impact based on historical signal data and historical spatial states of the same type of power equipment in different monitoring sets, includes: Spatial clustering constraints are constructed based on the monitoring range of environmental monitoring equipment. Spatial clustering analysis is performed on all power equipment with the deployment location of environmental monitoring equipment as the cluster center to obtain several clusters. Each cluster is used as a monitoring set. Historical signal data and historical spatial status of the same type of power equipment under the same environmental monitoring data conditions are extracted from different monitoring sets. Based on the offset and fluctuation characteristics of historical signal data as historical spatial status changes, the influence of spatial status deviation is constructed.

5. The intelligent substation signal monitoring method based on intelligent agents as described in claim 1, characterized in that: The step of determining the degree of matching between the real-time signal data of the physically associated devices and the physical anomaly transmission relationship, and outputting the second anomaly probability, includes: Output the desired signal data of the physically associated devices based on the real-time signal data of the alarm action devices; Based on the degree of matching between the real-time signal data and the expected signal data of the physically associated devices, and combined with the physical anomaly transmission relationship, a second anomaly probability is output.

6. The intelligent substation signal monitoring method based on intelligent agents as described in claim 1, characterized in that: By using evidence theory to fuse the first, second, and third anomaly probabilities, a comprehensive anomaly probability is obtained.

7. The intelligent substation signal monitoring method based on intelligent agents as described in claim 3, characterized in that: The process of determining the physical topology links and signal topology links between power equipment based on the substation topology map also includes: A single action monitoring device is obtained based on the overlapping nodes of the physical topology link and the signal topology link; A correlation analysis was performed on the historical signal data of all primary motion monitoring devices to construct the signal trend correlation relationship of the primary motion monitoring devices.

8. The intelligent substation signal monitoring method based on intelligent agents as described in claim 7, characterized in that: The comprehensive anomaly probability of each power device output based on the influence of real-time signal data and spatial state deviation, the physical anomaly transmission relationship, and the signal anomaly transmission relationship includes: Based on the degree of matching between the real-time signal data and signal trend correlation of all primary motion monitoring devices, the signal monitoring reliability of the primary motion monitoring devices is output. Using a primary action monitoring device whose signal monitoring reliability is greater than the preset signal monitoring reliability benchmark as the matching benchmark, the comprehensive anomaly probability of each power device is output based on the spatial relationship, physical anomaly transmission relationship and signal anomaly transmission relationship between the other power devices, the action monitoring devices and the primary action monitoring devices; A single action monitoring device whose signal monitoring reliability is less than or equal to the preset signal monitoring reliability benchmark is considered an abnormal monitoring device. The alarm result of the alarm action device is corrected based on the physical abnormal transmission relationship and signal abnormal transmission relationship between the alarm action device, the alarm action monitoring device and the abnormal monitoring device.

9. A substation signal intelligent monitoring system based on intelligent agents, used to implement the method as described in any one of claims 1 to 8, characterized in that: include: The link derivation module is used to determine the physical topology links and signal topology links between power equipment based on the substation topology diagram, and to construct the physical anomaly transmission relationship corresponding to the physical topology link and the signal anomaly transmission relationship corresponding to the signal topology link by combining the physical operation mechanism and signal transmission mechanism of the power equipment. The status analysis module is used to group power equipment within the monitoring range of the same environmental monitoring equipment into the same monitoring set, and to obtain the impact of spatial status deviation based on historical signal data and historical spatial status of the same type of power equipment in different monitoring sets. The probability analysis module is used to output the comprehensive anomaly probability of each power device based on the influence of real-time signal data and spatial state deviation, physical anomaly transmission relationship, and signal anomaly transmission relationship. An intelligent agent is used to output anomaly localization results based on the anomaly probability by using an inference mechanism.

Images