Electrical fault detection method and electrical fault detection system

Abstract

The invention provides an electrical fault detection method and an electrical fault detection system, and relates to the technical field of data processing, and the method comprises the steps: carrying out the equipment multi-dimensional correlation analysis of a target production environment, and constructing a plurality of dynamic priority detection trees; establishing a communication link between the plurality of root node devices and the fault detection cloud, and establishing a plurality of edge processing units; when the first electrical fault is identified, positioning a first root node device at the plurality of root node devices; and activating the first edge processing unit, performing progressive electrical fault detection on the target production environment by taking the first dynamic priority detection tree as a fault detection constraint, and positioning target risk equipment. The technical problems that in the prior art, electrical fault detection is always static, intelligent priority adjustment is lacked, dynamic changes of equipment operation conditions are difficult to deal with, and the accuracy and efficiency of fault detection are low are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to an electrical fault detection method and an electrical fault detection system. Background Art

[0002] In modern industrial production environments, the health of electrical equipment is one of the key factors to ensure the normal operation of the production process. However, these devices are prone to failures when they are in long-term operation, leading to production interruptions, equipment damage, and even safety accidents. In the prior art, although there are many methods for electrical fault detection, traditional detection methods still have problems. Specifically, existing equipment detection is often static, lacks intelligent priority adjustment, and is difficult to cope with dynamic changes in equipment operation conditions. This static detection mechanism makes it impossible for the system to perform targeted fault detection when resources are limited. The detection process is prone to the risk of wasting resources or missing key equipment, thereby reducing the efficiency of fault detection. Summary of the invention

[0003] The present application provides an electrical fault detection method and an electrical fault detection system, aiming to solve the technical problems that the electrical fault detection in the prior art is often static, lacks intelligent priority adjustment, and is difficult to cope with dynamic changes in equipment operating conditions, resulting in low accuracy and efficiency of fault detection.

[0004] In a first aspect disclosed in the present application, a method for detecting electrical faults is provided, the method comprising: performing a multi-dimensional correlation analysis of equipment in a target production environment, and constructing a plurality of dynamic priority detection trees based on the analysis results; constructing communication links between a plurality of root node devices in the plurality of dynamic priority detection trees and a fault detection cloud, and constructing a plurality of edge processing units in the target production environment according to the plurality of dynamic priority detection trees; when the fault detection cloud identifies a first electrical fault, locating a first root node device among the plurality of root node devices by analyzing the first electrical fault; activating a first edge processing unit according to the first root node device, the first edge processing unit using the first dynamic priority detection tree as a fault detection constraint, performing progressive electrical fault detection on the target production environment, and locating target risk equipment.

[0005] The second aspect disclosed in the present application provides an electrical fault detection system, which is used for the above-mentioned electrical fault detection method, and the system includes: an association analysis module, which is used to perform multi-dimensional association analysis of equipment in the target production environment, and construct multiple dynamic priority detection trees based on the analysis results; a communication link construction module, which is used to construct communication links between multiple root node devices in the multiple dynamic priority detection trees and a fault detection cloud, and construct multiple edge processing units in the target production environment according to the multiple dynamic priority detection trees; a first device positioning module, which is used to locate the first root node device among the multiple root node devices by analyzing the first electrical fault after the fault detection cloud identifies the first electrical fault; an electrical fault detection module, which is used to activate the first edge processing unit according to the first root node device, and the first edge processing unit uses the first dynamic priority detection tree as a fault detection constraint to perform progressive electrical fault detection on the target production environment and locate the target risk device.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0007] By performing multi-dimensional correlation analysis of the equipment in the target production environment, the connection relationship between the equipment can be identified, and multiple dynamic priority detection trees can be constructed to ensure that the complex interaction between the equipment can be considered during fault detection, thereby greatly improving the accuracy of fault detection; communication links between multiple root node devices in multiple dynamic priority detection trees and the fault detection cloud can be constructed to ensure that when an electrical fault occurs, the cloud can instantly obtain device data and analyze the fault. By distributing the detection tasks in the edge processing units, the local edge processing units can be quickly activated for real-time detection and response when a device fails; when an electrical fault is identified, the electrical fault can be analyzed. The fault is located at the fault source device, that is, the root node device, and the fault status of the sub-node device is analyzed based on the root node fault. Through accurate fault location, false alarms are reduced and the effectiveness of the detection process is improved. The edge processing unit performs progressive fault detection on high-priority devices based on the dynamic priority detection tree to locate the target risk devices. This cloud and edge collaboration reduces the delay in data transmission and significantly improves the response speed of fault detection. At the same time, the step-by-step detection mechanism of faults from the root node to the sub-node ensures that the fault propagation path can be tracked in time, thereby improving the detection efficiency. Through accurate fault location, false alarms are reduced and the effectiveness of the detection process is improved.

[0008] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 A schematic flow chart of an electrical fault detection method provided in an embodiment of the present application.

[0010] Figure 2 A schematic diagram of the structure of an electrical fault detection system provided in an embodiment of the present application.

[0011] Description of the accompanying drawings: association analysis module 10, communication link construction module 20, first device positioning module 30, electrical fault detection module 40. DETAILED DESCRIPTION

[0012] The embodiment of the present application provides an electrical fault detection method to solve the technical problem that the electrical fault detection in the prior art is often static, lacks intelligent priority adjustment, and is difficult to cope with dynamic changes in equipment operating conditions, resulting in low accuracy and efficiency of fault detection.

[0013] After introducing the basic principles of the present application, various non-limiting implementation methods of the present application will be specifically introduced below in conjunction with the accompanying drawings of the specification. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0014] like Figure 1 As shown, an embodiment of the present application provides an electrical fault detection method, the method comprising:

[0015] Perform multi-dimensional correlation analysis of devices in the target production environment and build multiple dynamic priority detection trees based on the analysis results.

[0016] The multi-dimensional correlation analysis of equipment aims to determine the correlation relationship between different equipment in the production environment. This analysis comprehensively evaluates the potential fault transmission paths and mutual influences between equipment through multiple correlation dimensions, and builds multiple correlation detection trees based on the analysis results. According to the historical operation and maintenance logs in the target production environment, the priority of each node device is calculated, and the equipment risk ranking in the detection process is established to obtain multiple priority topologies. In this way, when detecting faults, priority can be given to those devices with higher fault risks. Based on the aforementioned multi-dimensional correlation analysis of equipment and priority topology, a dynamic priority detection tree is constructed. The key feature of the dynamic priority detection tree is the dynamic adjustment of priority. Through historical data analysis and real-time fault detection, the priority of each node device can be dynamically adjusted, thereby giving priority to those devices that are most likely to fail or have the greatest impact. This improves the efficiency of fault detection and reduces the possibility of false detection or missed detection.

[0017] Communication links between multiple root node devices in the multiple dynamic priority detection trees and a fault detection cloud are constructed, and multiple edge processing units are constructed in the target production environment according to the multiple dynamic priority detection trees.

[0018] The root node device is the top-level device in the dynamic priority detection tree. Each detection tree has one or more root node devices, which are selected through multi-dimensional association analysis and priority evaluation and represent the starting point of the entire detection tree. Through network protocols, such as wireless communication protocols, 5G, LoRa, etc., these root node devices are connected to the fault detection cloud to build a communication link. Through the communication link, the cloud can obtain real-time data of the device and perform fault detection tasks, and at the same time feed back the detection results to the edge processing unit.

[0019] In the target production environment, multiple edge processing units are deployed based on the constructed multiple dynamic priority detection trees. These edge processing units can share the computing tasks in the cloud and perform rapid electrical fault detection and preliminary decision-making locally, thereby improving the detection speed, reducing network latency, and realizing distributed detection.

[0020] Specifically, according to the type and risk characteristics of the equipment in the dynamic priority detection tree, multiple sample fault detection models are extracted from the fault detection library and deployed in the edge processing unit. Each edge processing unit can be responsible for the fault detection task of a specific group of equipment. When a root node device fails, the corresponding edge processing unit is activated to perform the fault detection task based on the dynamic priority detection tree. In this process, each edge processing unit has multiple sample detection models, and the appropriate model is selected according to the device status. Multiple groups of sample fault detection models are stored in the edge processing unit to ensure that each unit can quickly call the corresponding model for fault analysis when receiving a fault alarm. Through cloud-edge collaboration, the entire system can not only ensure real-time performance, but also use the powerful computing power of the cloud to perform more complex analysis, ensuring real-time monitoring and detection of high-risk equipment.

[0021] When the fault detection cloud identifies the first electrical fault, the first root node device is located among the multiple root node devices by analyzing the first electrical fault.

[0022] The fault detection cloud continuously receives monitoring data from multiple root node devices. This data may include device operating status, historical operation and maintenance logs, fault alarms, environmental parameters, etc. In the process of identifying faults, the cloud relies on multiple fault detection models to determine whether there is an electrical fault by analyzing the device's operating data. For example, if the voltage, current and other parameters of the electrical equipment fluctuate abnormally, it is judged that a fault has occurred, the first identified electrical fault is marked, and fault location is initiated.

[0023] First, the first identified electrical fault is parsed to analyze the characteristics of the fault, including the time, location, affected parameters (such as voltage, current, temperature, etc.) of the fault, and the association with other devices. Through the analyzed fault characteristics, the constructed dynamic priority detection tree is queried to find the root node device most related to the fault. In the dynamic priority detection tree, each node device is associated with a fault detection model. Through model comparison and priority evaluation, the node device with the highest match to the current fault characteristics is found. Based on the association analysis, along the node connection path of the detection tree, the possible propagation path of the fault is gradually traced back until the first abnormal device is found, that is, the first root node device.

[0024] The first edge processing unit is activated according to the first root node device. The first edge processing unit uses the first dynamic priority detection tree as a fault detection constraint to perform progressive electrical fault detection on the target production environment and locate the target risk device.

[0025] The cloud sends an activation signal to the first edge processing unit deployed near the device or responsible for monitoring it based on the located first root node device. The activated first edge processing unit uses the first dynamic priority detection tree as a constraint and performs fault detection from high to low according to the preset priority topology structure to ensure reasonable resource allocation and avoid unnecessary low-priority device detection occupying too many resources.

[0026] Progressive detection is a gradually deepening fault detection process, which first detects the most critical and high-risk devices, then gradually expands to other devices, and finally covers the entire fault detection range. Specifically, the first root node device is detected, and its operating parameters such as voltage, current, temperature, etc. are monitored, and compared with historical fault data to determine whether there are further fault signs or hidden dangers. If the root node device itself has a fault, other devices directly associated with it will continue to be detected. According to the topological structure of the dynamic priority detection tree, its neighboring devices are progressively checked along the possible propagation path of the fault. Following the node priority order in the dynamic priority detection tree, the fault detection of the device is performed layer by layer. The devices with higher priority are detected first, and the devices with lower priority are used as subsequent detection targets.

[0027] After progressive detection, devices with high failure risks are identified as target risk devices. Such devices may have already shown signs of failure or are in potential failure propagation paths. After locating the target risk devices, the failure information of the devices is transmitted to the cloud or maintenance personnel are notified, thus achieving real-time, accurate and efficient fault detection.

[0028] Further, constructing communication links between multiple root node devices in the multiple dynamic priority detection trees and the fault detection cloud, and constructing multiple edge processing units in the target production environment according to the multiple dynamic priority detection trees, the method includes:

[0029] Collect true fault data of the target production environment, and aggregate the data collection results according to the device type to obtain N true fault data sets of N types of sample production equipment; construct and obtain N sample fault detection models based on the N true fault data sets; associate and store the N types of sample production equipment and the N sample fault detection models to obtain an electrical fault detection library; after obtaining multiple sample fault detection models from the electrical fault detection library according to the device types of the multiple root node devices, complete the construction of the fault detection cloud by connecting the multiple sample fault detection models in parallel; obtain multiple groups of sample fault detection models from the electrical fault detection library according to the device types that constitute the nodes of the multiple dynamic priority detection trees; after establishing the activation relationship of the multiple groups of sample fault detection models according to the multiple dynamic priority detection trees, store the multiple groups of sample fault detection models to the multiple edge processing units.

[0030] True faults refer to real fault data that occurs in the actual production process of the equipment. These faults include electrical faults, control system faults, mechanical faults, etc., which are related to the actual operation of the equipment. In the target production environment, various types of production equipment are monitored in real time, and electrical parameters such as voltage, current, power, etc., as well as fault-related data such as fault codes and equipment status when the fault occurs are collected during the operation of the equipment.

[0031] There are usually many different types of equipment in a production environment. Each type of equipment has different failure modes and characteristics. According to the equipment type, such as motors, transformers, switches, etc., the collected data is classified and aggregated to obtain N true fault data sets corresponding to N types of sample production equipment. Each data set aggregates the fault data of a specific equipment type.

[0032] Based on the collected N true fault data sets, fault detection models are constructed for each type of equipment. These models will be used to detect electrical faults that may occur in the equipment during operation. Specifically, the appropriate model type is selected according to the equipment type and fault characteristics. Common fault detection model types include decision trees, random forests, support vector machines, etc. The equipment failure mode is learned through training data, and the model is trained using the true fault data set of each equipment type. By learning the difference between the normal operating state and the fault state of the equipment, the model can identify possible failures of the equipment in the future. After the model training is completed, the model is evaluated through the test data set to ensure its accuracy and reliability. Finally, N sample fault detection models for N types of equipment are obtained. These models can perform effective fault detection based on the true fault data of the equipment.

[0033] N types of sample production equipment and their corresponding N sample fault detection models are associated and stored to build an electrical fault detection library. The fault detection library is a comprehensive database that includes the type of equipment, its corresponding fault detection model, and related historical fault data. Through this library, the detection model of a specific device can be quickly accessed to perform fault detection.

[0034] For multiple root node devices, each root node device has its corresponding device type. First, the corresponding fault detection model is called from the electrical fault detection library according to these device types to obtain multiple sample fault detection models. These models are integrated in parallel. Parallel means that these fault detection models can process input data and make judgments at the same time. The multiple sample fault detection models after parallel connection are integrated into the cloud detection system to form a complete fault detection cloud architecture, which can process data from root node devices in real time and quickly identify possible electrical faults.

[0035] The dynamic priority detection tree is composed of multiple nodes, each node represents a device, and each device type may be different. For each node device, the corresponding sample detection model is called from the fault detection library according to its device type. Each group of devices may have different fault characteristics, so it is necessary to call the corresponding detection model in a targeted manner. Each node device corresponds to a group of sample fault detection models. The various node devices of the entire dynamic priority detection tree form multiple groups of sample fault detection models. These model groups are used in the subsequent fault detection process to ensure that the fault detection model of each device is highly matched with its device type.

[0036] The edge processing unit is located at the production site and can process the operation data of the equipment in real time. Therefore, multiple sets of sample fault detection models are stored in the edge processing unit so that they can be quickly called and executed during fault detection. Each edge processing unit stores the sample fault detection model of the equipment it is responsible for. In this way, the edge processing unit can dynamically activate the corresponding detection model according to the fault status to achieve fast and real-time fault detection.

[0037] Furthermore, a multi-dimensional correlation analysis of equipment is performed on the target production environment, and multiple dynamic priority detection trees are constructed based on the analysis results. The method includes:

[0038] Perform multi-dimensional equipment association analysis on the target production environment, and build multiple association detection trees based on the analysis results; use the historical operation and maintenance logs of the target production environment to calculate the node priorities of the multiple association detection trees to obtain multiple priority topologies; pre-allocate detection computing power for the multiple association detection trees according to the multiple priority topologies to obtain the multiple dynamic priority detection trees.

[0039] The associated features of the devices in the target production environment are aggregated. The associated features between devices include physical connection, functional correlation, fault history similarity, etc. By aggregating these features, multi-dimensional association types are obtained. Based on the fault transmission path, the propagation relationship of faults between devices is analyzed. In this analysis, the direction of fault transmission is carried out in the form of out-degree connection, that is, how the fault of one device affects other devices. Based on the aggregated multi-dimensional association types, multiple groups of device association trees are constructed. In order to improve the accuracy of fault detection, cross-dimensional fusion is performed on the association trees of different devices. During the fusion process, a preset association scale is set, and these device association trees are cross-fused to finally generate multiple association detection trees. By introducing multiple association dimensions and cross-dimensional fusion, a more comprehensive and complex device association model can be formed to ensure that the interaction of different association dimensions is fully reflected.

[0040] From the historical operation and maintenance log, call the fault risk parameters related to the node equipment of the associated detection tree. The risk parameters include the equipment's fault frequency, loss mean, fault level, associated fault coefficient, etc. Based on the extracted fault risk parameters, evaluate the risk value of each node equipment, fit these risk values ​​into the associated detection tree, and form the equipment risk ranking. In order to build a priority topology structure, the risk value of the equipment in the associated detection tree is cumulatively calculated across levels. In this way, each layer of the entire associated detection tree can be comprehensively evaluated and the relative priority of each device can be obtained. Similarly, node priority calculations are performed on multiple associated detection trees to obtain multiple priority topologies. The priority of each device is determined by its fault risk value, and devices with higher priorities will be given priority during the detection process.

[0041] According to multiple priority topologies, corresponding detection computing power is allocated to each associated detection tree, which means that during fault detection, detection resources are adjusted according to the priority of the device to ensure that high-risk devices get more resources. Finally, based on the device priority and computing power allocation results, dynamic priority detection trees are generated. These detection trees can be dynamically adjusted as the device status, production environment, and fault risk change, thereby providing a more efficient and targeted detection path in actual fault detection.

[0042] Furthermore, a multi-dimensional correlation analysis of equipment is performed on the target production environment, and multiple correlation detection trees are constructed based on the analysis results. The method includes:

[0043] Aggregate equipment association features of the target production environment to obtain a multi-dimensional association type; use fault conduction as the out-degree connection direction, adopt the multi-dimensional association type to perform equipment fault conduction analysis in the target production environment, and complete the construction of multiple groups of equipment association trees; preset an association scale, and use the association scale as a fusion constraint to perform cross-dimensional cross-fusion on the multiple groups of equipment association trees to obtain the multiple association detection trees.

[0044] By aggregating different types of equipment association features, the multi-dimensional association relationship between equipment is identified. Specifically, equipment association features can come from data of multiple dimensions, such as electrical association, control association, mechanical association, and process flow association. The association information from different dimensions is integrated to obtain a multi-dimensional association type, which represents the association between each device and other devices in multiple dimensions.

[0045] Based on multi-dimensional association types, the fault conduction paths between devices are analyzed. Fault conduction refers to how a device may affect other devices associated with it when it fails. Specifically, the direction of fault conduction is used as the out-degree connection. Specifically, the device fault is transmitted from the fault source device to the external device, forming a connection relationship "from the fault-occurring device to the affected device". The possible propagation path of the fault is determined through conduction analysis. Based on different association types, the following four association detection trees are constructed respectively: electrical association detection tree, which is mainly used to analyze fault propagation in electrical systems. The root node is the main electrical equipment, such as transformers, switches, etc., and the child nodes are the devices electrically connected to them; control association detection tree, which is used to analyze device associations in control systems. The root node is the control signal source device, and the child nodes are the controlled devices; mechanical association detection tree, which is used to analyze the mechanical dependency between devices. The root node is the key mechanical transmission equipment, such as conveyor belts, robotic arms, etc.; process flow association detection tree, which is used to analyze the association of equipment in the production process. The root node is the key equipment in the process, and the child nodes are its downstream equipment. Each association detection tree shows how device failures are transmitted within that dimension. Through these association detection trees, possible failure transmission paths between devices can be identified, and possible chain reactions caused by device failures can be predicted in advance.

[0046] The preset association scale is to control the depth of the association detection tree. For example, the maximum level of an association detection tree is set to 5 layers. If the detection tree exceeds 5 layers, the root node is reset. This is to improve the detection response speed of electrical faults and prevent the association chain from being too long, which leads to a decrease in detection efficiency.

[0047] Different types of associated detection trees may intersect with each other. For example, when an electrical device fails, it may not only affect the electrical system, but also the control system and mechanical system. Multiple associated detection trees with cross-devices can be integrated with each other to interactively analyze the transmission paths between different fault types. During the fusion process, the fault transmission paths between different detection trees are interactively analyzed. For example, when a fault in an electrical device is detected, not only its electrical associated path is analyzed, but also the status of the equipment associated with the mechanical, control, and process flow of the device is checked across dimensions. Through multi-dimensional fusion analysis, the possible transmission paths of the fault can be more comprehensively grasped. After cross-dimensional cross-fusion, multiple associated detection trees are generated. These fused detection trees not only show the fault conduction of the device in a single dimension, but also reflect the fault transmission path across dimensions, which makes fault detection more comprehensive and intelligent.

[0048] Furthermore, the historical operation and maintenance logs of the target production environment are used to calculate the node priorities of the multiple associated detection trees to obtain multiple priority topologies. The method includes:

[0049] According to the device composition of the first associated detection tree, multiple fault risk parameters of multiple node constituent devices are obtained from the historical operation and maintenance log call, wherein the device composition of the first associated detection tree includes the first root node device and multiple node constituent devices; the multiple fault risk parameters are used to evaluate the multiple node constituent devices to obtain multiple device risk values; according to the positional relationship of the multiple node constituent devices in the first associated detection tree, the multiple device risk values ​​are fitted to the first associated detection tree; cross-level value accumulation calculation is performed on the first associated detection tree to obtain a first priority topology; and by analogy, node priority calculation is performed on the multiple associated detection trees to obtain the multiple priority topologies.

[0050] In the first associated detection tree, the equipment composition is clearly defined. Each detection tree contains a root node device and several child node devices associated with it. From the historical operation and maintenance logs, fault-related data of these devices are extracted separately, including fault frequency, frequency affected by root node faults, fault risk level distribution, economic impact, etc., to form multiple fault risk parameters. These data include not only the historical fault conditions of the equipment itself, but also the fault propagation of the equipment in the associated environment and the economic impact of the fault on production.

[0051] The risk value of each device is comprehensively evaluated by combining its historical failure frequency, the failure frequency affected by the root node device, the failure risk level, and economic impact. The risk weighting method can be used to weight the sum of different parameters to obtain the device risk value through calculation.

[0052] When performing fault traversal detection, it is necessary to reasonably allocate computing resources according to the risk value of each device. High-risk devices require more computing resources to ensure the accuracy and real-time performance of their detection, while low-risk devices can be allocated fewer resources. By evaluating the risk value of the device, the detection order and resource allocation can be dynamically adjusted. For example, devices that have frequently failed in the past and have a high risk of fault propagation will be tested first, while those with a low risk of failure can be tested later or simplified when resources are insufficient. In this way, detection computing resources will be reasonably utilized to ensure maximum detection efficiency.

[0053] The risk value of each node constituting the device is mapped according to its position in the associated detection tree to ensure that the risk value is consistent with the hierarchical relationship of the device in the tree structure. Specifically, based on the device risk value calculation result, multiple device risk values ​​are allocated to the corresponding node devices in the first associated detection tree. For example, if in the first associated detection tree, the risk value of a root node device is 75, and the risk values ​​of its two directly associated child node devices are 25 and 50 respectively, then these values ​​will be fitted to the corresponding nodes in the associated detection tree respectively. In this way, each node in the tree structure contains the risk information of the device.

[0054] According to the risk value of the device, a cross-level risk accumulation calculation is performed for the allocation of computing power resources. For example, if the risk value of the parent node is 12, and the risk values ​​of the two child nodes are 18 and 27, then the cumulative risk value of the parent node is 12+18+27=57. Similarly, the risk value is accumulated between the hierarchical nodes. The risk value ratio of two child nodes at the same level is the computing power allocation ratio. The rule for allocating the computing power allocated to the parent node to the child nodes is also the risk value ratio of the child nodes at the same level. In general, the key to cross-level cumulative calculation is to ensure that the risk value of each device is reflected in its own detection priority and also affects the allocation of overall computing power resources through the parent-child relationship. Through this cross-level risk value accumulation calculation, the first priority topology of the first associated detection tree is obtained, which clarifies how to allocate computing power resources according to the risk value of the device.

[0055] According to the calculation method in the first association detection tree, the risk value is fitted for each node of the association detection tree, and cross-level cumulative calculations are performed according to the hierarchical relationship of the equipment. The nodes in each tree dynamically adjust the computing power allocation according to their relative importance and risk value, and finally obtain multiple priority topologies to achieve comprehensive priority resource allocation.

[0056] Further, according to the device composition of the first association detection tree, multiple fault risk parameters of multiple node composition devices are obtained from the historical operation and maintenance log call, and the method includes:

[0057] In the first association detection tree, a first hierarchical path node of a first node constituent device is obtained by tracing back along the node connection relationship, wherein the first hierarchical path node includes M node constituent devices, and the first node constituent device is any one of the multiple node constituent devices; after enumerating the first node constituent device and the M node constituent devices in combination to obtain M groups of device combinations, the M groups of device combinations are used to traverse the historical operation and maintenance log to obtain M associated fault frequencies; weighted calculation of the M associated fault frequencies is performed according to the hierarchical relationship between the first node constituent device and the M node constituent devices to obtain a first associated fault coefficient; a first device failure frequency, a first fault loss mean, and a first fault level mode of the first node constituent device are obtained from the historical operation and maintenance log, and the first device failure frequency, the first fault loss mean, the first fault level mode, and the first associated failure coefficient constitute a first fault risk parameter; and so on, the multiple fault risk parameters are obtained from the historical operation and maintenance log according to the device composition of the first association detection tree.

[0058] In the association detection tree, device nodes are connected through fault association relationships. For example, the fault of electrical equipment may be transmitted to the equipment it directly controls or the equipment affected by it. The tracing process starts from a low-level node and searches upward step by step for the upper-level nodes directly associated with it until a complete path is found. It is assumed that the first association detection tree is composed of a root node, child nodes, and multi-layer child nodes. In this tree structure, the first node device is any low-level node (the device closest to the bottom layer). By tracing, the upper-level node connected to it is found to form a complete path. Along the node connection relationship, the complete hierarchical path of the first node device is obtained. This path contains multiple nodes from the low level to the high level, marked as M nodes constituting the device.

[0059] The first node constituent device and the M node constituent devices are combined and enumerated to generate M different device combinations. Based on these device combinations, the fault data corresponding to these device combinations are queried in the historical operation and maintenance logs to obtain the associated fault frequencies under different combinations. The associated fault frequencies reflect the number of times other devices are affected by a device failure and fail. These frequencies are used to evaluate the fault conduction correlation between devices.

[0060] The hierarchical relationship of the equipment in the association detection tree will affect the weight of fault transmission. Generally, the equipment closer to the root node has a greater impact on the fault of the lower-level equipment, so the upper-level node will obtain a higher weight. According to the hierarchical relationship and the historical fault frequency, these fault frequencies are calculated using a weighted method. For example, assuming there are two combinations, the association fault frequency of combination 1 (A and B) is 10 times, and the association fault frequency of combination 2 (A and C) is 5 times. In addition, A and B are at a low level with a weight of 0.6, and the weight between A and C is 0.8. Then the weighted calculation is F = 0.6*10 + 0.8*5 = 10, and F represents the overall association fault coefficient between A, B, and C. The first association fault coefficient obtained by calculation is used to evaluate the transmission relationship of faults between nodes, providing a basis for subsequent fault detection and computing resource allocation.

[0061] Extract key fault parameters related to the first node component equipment from the historical operation and maintenance log, including the first equipment failure frequency, the first fault loss mean and the first fault level mode. The first equipment failure frequency reflects the number of times the equipment has failed in history. The higher the equipment failure frequency, the more likely the equipment is to fail and the higher the risk. The first fault loss mean represents the average loss caused by each failure, such as maintenance cost, production downtime loss, etc. This parameter is used to evaluate the economic impact of equipment failure. The first fault level mode is the most common value in the equipment failure level in history, which is used to reflect the severity of the equipment failure. The fault level is usually divided into several levels, such as minor, medium and severe. These values, combined with the first associated fault coefficient, constitute the first fault risk parameter, which serves as the complete fault risk parameter of the first node component equipment.

[0062] Based on the device that constitutes the first node, the same extraction process is applied to other nodes in the associated detection tree. For each device, its failure frequency, failure loss mean, failure level mode, and associated failure coefficient are extracted from the historical operation and maintenance log to form the corresponding failure risk parameters. All nodes in the associated detection tree are traversed to extract the corresponding failure risk parameters for each device. These parameters provide a data basis for subsequent risk analysis and fault detection.

[0063] Furthermore, the method uses the multiple fault risk parameters to evaluate the multiple node constituent devices to obtain multiple device risk values, and includes:

[0064] A device value evaluation function is pre-constructed, and the device value evaluation function is as follows:

[0065]

[0066] Among them, I is the equipment risk value, F is the equipment failure frequency, L is the mean failure loss, K is the scaling factor, M is the mode of the failure level, and C is the associated failure coefficient; the multiple failure risk parameters are substituted into the equipment value evaluation function to obtain the multiple equipment risk values.

[0067] Specifically, the equipment value evaluation function is as follows:

[0068]

[0069] I is the equipment risk value, which indicates the overall failure risk value of the equipment; F is the equipment failure frequency, which reflects the number of times the equipment has failed in history. The higher the failure frequency, the greater the risk; L is the mean failure loss, that is, the average loss caused by each equipment failure, such as economic loss, production downtime, etc. The greater the loss, the higher the risk of the equipment; K is the scaling factor, which is a constant used to adjust the failure loss. The starting loss of production equipment is several thousand, which is not in the same order of magnitude as other data; M is the mode of the failure level, which refers to the most common level value in the equipment failure level in historical failures, such as minor, moderate, severe, etc. The higher the failure level, the greater the potential risk of the equipment; C is the correlation failure coefficient, which indicates the degree of correlation between the equipment failure and other equipment. The larger the correlation failure coefficient, the smaller the impact of the equipment failure on other equipment.

[0070] The function provides a comprehensive risk assessment by combining equipment failure frequency, failure loss, failure level and the failure correlation between the equipment and other equipment. Through the combination of square roots and proportional coefficients, the function can balance the contribution of different risk parameters to the overall risk of the equipment.

[0071] Substitute the failure risk parameters of each device into the evaluation function, calculate the specific risk value of the device, and obtain multiple device risk values. The risk value is used to determine the priority detection level and resource allocation of the device. The higher the risk value of the device, the higher the fault detection priority, indicating that the device requires closer monitoring and higher resource allocation. Through these risk values, the failure risk of the equipment can be effectively determined, and high-risk equipment can be given priority in detection and repair to avoid losses to production caused by equipment failure.

[0072] In summary, the electrical fault detection method provided by the embodiment of the present application has the following technical effects:

[0073] By performing multi-dimensional correlation analysis of the equipment in the target production environment, the connection relationship between the equipment can be identified, and multiple dynamic priority detection trees can be constructed to ensure that the complex interaction between the equipment can be considered during fault detection, thereby greatly improving the accuracy of fault detection; communication links between multiple root node devices in multiple dynamic priority detection trees and the fault detection cloud can be constructed to ensure that when an electrical fault occurs, the cloud can instantly obtain device data and analyze the fault. By distributing the detection tasks in the edge processing units, the local edge processing units can be quickly activated for real-time detection and response when a device fails; when an electrical fault is identified, the electrical fault can be analyzed. The fault is located at the fault source device, that is, the root node device, and the fault status of the sub-node device is analyzed based on the root node fault. Through accurate fault location, false alarms are reduced and the effectiveness of the detection process is improved. The edge processing unit performs progressive fault detection on high-priority devices based on the dynamic priority detection tree to locate the target risk devices. This cloud and edge collaboration reduces the delay in data transmission and significantly improves the response speed of fault detection. At the same time, the step-by-step detection mechanism of faults from the root node to the sub-node ensures that the fault propagation path can be tracked in time, thereby improving the detection efficiency. Through accurate fault location, false alarms are reduced and the effectiveness of the detection process is improved.

[0074] Based on the same inventive concept as the electrical fault detection method in the aforementioned embodiment, Figure 2 As shown, an embodiment of the present application provides an electrical fault detection system, the system comprising:

[0075] An association analysis module 10, the association analysis module 10 is used to perform a multi-dimensional association analysis of the equipment in the target production environment, and to construct multiple dynamic priority detection trees based on the analysis results; a communication link construction module 20, the communication link construction module 20 is used to construct a communication link between multiple root node devices in the multiple dynamic priority detection trees and a fault detection cloud, and to construct multiple edge processing units in the target production environment according to the multiple dynamic priority detection trees; a first device positioning module 30, the first device positioning module 30 is used to locate the first root node device among the multiple root node devices by analyzing the first electrical fault after the fault detection cloud identifies the first electrical fault; an electrical fault detection module 40, the electrical fault detection module 40 is used to activate a first edge processing unit according to the first root node device, the first edge processing unit uses the first dynamic priority detection tree as a fault detection constraint, performs progressive electrical fault detection on the target production environment, and locates the target risk device.

[0076] Furthermore, the system further includes a sample fault detection model building module to perform the following operation steps:

[0077] Collect true fault data of the target production environment, and aggregate the data collection results according to the device type to obtain N true fault data sets of N types of sample production equipment; construct and obtain N sample fault detection models based on the N true fault data sets; associate and store the N types of sample production equipment and the N sample fault detection models to obtain an electrical fault detection library; after obtaining multiple sample fault detection models from the electrical fault detection library according to the device types of the multiple root node devices, complete the construction of the fault detection cloud by connecting the multiple sample fault detection models in parallel; obtain multiple groups of sample fault detection models from the electrical fault detection library according to the device types that constitute the nodes of the multiple dynamic priority detection trees; after establishing the activation relationship of the multiple groups of sample fault detection models according to the multiple dynamic priority detection trees, store the multiple groups of sample fault detection models to the multiple edge processing units.

[0078] Furthermore, the system further includes a dynamic priority detection tree acquisition module to perform the following operation steps:

[0079] Perform multi-dimensional equipment association analysis on the target production environment, and build multiple association detection trees based on the analysis results; use the historical operation and maintenance logs of the target production environment to calculate the node priorities of the multiple association detection trees to obtain multiple priority topologies; pre-allocate detection computing power for the multiple association detection trees according to the multiple priority topologies to obtain the multiple dynamic priority detection trees.

[0080] Furthermore, the system further includes an associated detection tree acquisition module to perform the following operation steps:

[0081] Aggregate equipment association features of the target production environment to obtain a multi-dimensional association type; use fault conduction as the out-degree connection direction, adopt the multi-dimensional association type to perform equipment fault conduction analysis in the target production environment, and complete the construction of multiple groups of equipment association trees; preset an association scale, and use the association scale as a fusion constraint to perform cross-dimensional cross-fusion on the multiple groups of equipment association trees to obtain the multiple association detection trees.

[0082] Furthermore, the system further includes a priority topology acquisition module to perform the following operation steps:

[0083] According to the device composition of the first associated detection tree, multiple fault risk parameters of multiple node constituent devices are obtained from the historical operation and maintenance log call, wherein the device composition of the first associated detection tree includes the first root node device and multiple node constituent devices; the multiple fault risk parameters are used to evaluate the multiple node constituent devices to obtain multiple device risk values; according to the positional relationship of the multiple node constituent devices in the first associated detection tree, the multiple device risk values ​​are fitted to the first associated detection tree; cross-level value accumulation calculation is performed on the first associated detection tree to obtain a first priority topology; and by analogy, node priority calculation is performed on the multiple associated detection trees to obtain the multiple priority topologies.

[0084] Furthermore, the system further includes a fault risk parameter acquisition module to perform the following operation steps:

[0085] In the first association detection tree, a first hierarchical path node of a first node constituent device is obtained by tracing back along the node connection relationship, wherein the first hierarchical path node includes M node constituent devices, and the first node constituent device is any one of the multiple node constituent devices; after enumerating the first node constituent device and the M node constituent devices in combination to obtain M groups of device combinations, the M groups of device combinations are used to traverse the historical operation and maintenance log to obtain M associated fault frequencies; weighted calculation of the M associated fault frequencies is performed according to the hierarchical relationship between the first node constituent device and the M node constituent devices to obtain a first associated fault coefficient; a first device failure frequency, a first fault loss mean, and a first fault level mode of the first node constituent device are obtained from the historical operation and maintenance log, and the first device failure frequency, the first fault loss mean, the first fault level mode, and the first associated failure coefficient constitute a first fault risk parameter; and so on, the multiple fault risk parameters are obtained from the historical operation and maintenance log according to the device composition of the first association detection tree.

[0086] Furthermore, the system further includes a plurality of equipment risk value acquisition modules to perform the following operation steps:

[0087] A device value evaluation function is pre-constructed, and the device value evaluation function is as follows:

[0088]

[0089] Among them, I is the equipment risk value, F is the equipment failure frequency, L is the mean failure loss, K is the scaling factor, M is the mode of the failure level, and C is the associated failure coefficient; the multiple failure risk parameters are substituted into the equipment value evaluation function to obtain the multiple equipment risk values.

[0090] Through the above detailed description of an electrical fault detection method in this specification, those skilled in the art can clearly understand an electrical fault detection system in this embodiment. Since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.

[0091] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An electrical fault detection method, characterized in that: The method comprises: Conduct multi-dimensional correlation analysis of equipment in the target production environment and build multiple dynamic priority detection trees based on the analysis results; Constructing communication links between multiple root node devices in the multiple dynamic priority detection trees and a fault detection cloud, and constructing multiple edge processing units in the target production environment according to the multiple dynamic priority detection trees; When the fault detection cloud identifies a first electrical fault, locating the first root node device among the multiple root node devices by analyzing the first electrical fault; The first edge processing unit is activated according to the first root node device. The first edge processing unit uses the first dynamic priority detection tree as a fault detection constraint to perform progressive electrical fault detection on the target production environment and locate the target risk device.

2. An electrical fault detection method as claimed in claim 1, characterized in that: Building communication links between multiple root node devices in the multiple dynamic priority detection trees and the fault detection cloud, and building multiple edge processing units in the target production environment according to the multiple dynamic priority detection trees, the method comprising: Collect real fault data of the target production environment, aggregate the data collection results according to the equipment type, and obtain N real fault data sets of N types of sample production equipment; Constructing and obtaining N sample fault detection models based on the N true fault data sets; The N types of sample production equipment and the N sample fault detection models are stored in association to obtain an electrical fault detection library; After obtaining multiple sample fault detection models from the electrical fault detection library according to the device types of the multiple root node devices, the construction of the fault detection cloud is completed by connecting the multiple sample fault detection models in parallel; According to the device types constituted by the nodes of the multiple dynamic priority detection trees, multiple groups of sample fault detection models are obtained by calling from the electrical fault detection library; After establishing activation relationships among the multiple groups of sample fault detection models according to the multiple dynamic priority detection trees, the multiple groups of sample fault detection models are stored in the multiple edge processing units.

3. An electrical fault detection method as claimed in claim 1, characterized in that: Performing a multi-dimensional correlation analysis of equipment in a target production environment and constructing multiple dynamic priority detection trees based on the analysis results, the method comprising: Conduct multi-dimensional correlation analysis of equipment in the target production environment and build multiple correlation detection trees based on the analysis results; Using the historical operation and maintenance log of the target production environment, performing node priority calculation on the multiple associated detection trees to obtain multiple priority topologies; Pre-allocate detection computing power for the multiple associated detection trees according to the multiple priority topologies to obtain the multiple dynamic priority detection trees.

4. An electrical fault detection method as claimed in claim 3, characterized in that: Performing a multi-dimensional correlation analysis of equipment in a target production environment and constructing multiple correlation detection trees based on the analysis results, the method comprising: Aggregating equipment-related features of the target production environment to obtain multi-dimensional association types; Taking fault conduction as the out-degree connection direction, the multi-dimensional association type is used to perform equipment fault conduction analysis in the target production environment, and the construction of multiple groups of equipment association trees is completed; A correlation scale is preset, and the correlation scale is used as a fusion constraint to perform cross-dimensional cross fusion on the multiple groups of device correlation trees to obtain the multiple correlation detection trees.

5. An electrical fault detection method as claimed in claim 3, characterized in that: Using the historical operation and maintenance log of the target production environment, performing node priority calculation on the multiple associated detection trees to obtain multiple priority topologies, the method includes: According to the device composition of the first association detection tree, multiple fault risk parameters of multiple node composition devices are obtained from the historical operation and maintenance log, wherein the device composition of the first association detection tree includes the first root node device and multiple node composition devices; Using the multiple fault risk parameters to evaluate the multiple node constituent devices, and obtaining multiple device risk values; According to the positional relationship of the multiple node-constituting devices in the first association detection tree, fitting the multiple device risk values ​​to the first association detection tree; Performing cross-level value accumulation calculation on the first association detection tree to obtain a first priority topology; Similarly, node priorities are calculated for the multiple associated detection trees to obtain the multiple priority topologies.

6. An electrical fault detection method as claimed in claim 5, characterized in that: According to the equipment composition of the first association detection tree, multiple fault risk parameters of multiple node composition equipment are obtained from the historical operation and maintenance log call, and the method includes: Tracing back along the node connection relationship in the first association detection tree to obtain a first-level path node of a first node constituent device, wherein the first-level path node includes M node constituent devices, and the first node constituent device is any one of the multiple node constituent devices; After enumerating the first node constituent device and M node constituent devices in combination to obtain M device combinations, the M device combinations are used to traverse the historical operation and maintenance logs to obtain M associated fault frequencies; Perform weighted calculation of the M associated fault frequencies according to the hierarchical relationship between the first node constituent device and the M node constituent devices to obtain a first associated fault coefficient; Obtaining, from the historical operation and maintenance log, a first device failure frequency, a first failure loss mean, and a first failure level mode of the first node constituent device, wherein the first device failure frequency, the first failure loss mean, the first failure level mode, and the first associated failure coefficient constitute a first failure risk parameter; By analogy, the multiple fault risk parameters are obtained by calling from the historical operation and maintenance log according to the equipment composition of the first association detection tree.

7. An electrical fault detection method as claimed in claim 6, characterized in that: The method comprises: using the multiple fault risk parameters to evaluate the multiple node constituent devices to obtain multiple device risk values, the method comprising: A device value evaluation function is pre-constructed, and the device value evaluation function is as follows: Among them, I is the equipment risk value, F is the equipment failure frequency, L is the mean failure loss, K is the scaling factor, M is the mode of the failure level, and C is the correlation failure coefficient; Substituting the multiple fault risk parameters into the equipment value evaluation function, the multiple equipment risk values ​​are obtained.

8. An electrical fault detection system, characterized in that: For implementing an electrical fault detection method according to any one of claims 1 to 7, the system comprises: An association analysis module, which is used to perform multi-dimensional association analysis of equipment in a target production environment and construct multiple dynamic priority detection trees based on the analysis results; A communication link construction module, the communication link construction module is used to construct communication links between multiple root node devices in the multiple dynamic priority detection trees and the fault detection cloud, and to construct multiple edge processing units in the target production environment according to the multiple dynamic priority detection trees; A first device positioning module, the first device positioning module is used to locate the first root node device among the multiple root node devices by analyzing the first electrical fault after the fault detection cloud identifies the first electrical fault; An electrical fault detection module, wherein the electrical fault detection module is used to activate a first edge processing unit according to the first root node device, wherein the first edge processing unit uses a first dynamic priority detection tree as a fault detection constraint, performs progressive electrical fault detection on the target production environment, and locates target risk equipment.

Images