Power distribution network fault analysis feedback system based on artificial intelligence

Abstract

The invention discloses a power distribution network fault analysis feedback system based on artificial intelligence, and belongs to the field of state monitoring. Comprising a data management and control module, a tree crown management framework of a topological structure is established, and a plurality of traceable detection lines are formed; the data acquisition module is used for acquiring feedback data according to the coronal relationship; the data analysis module creates a warning analysis order according to the feedback data; the task execution module is used for tracing according to a tree crown management architecture according to the generated warning analysis order, judging whether a fault exists or not based on a preset threshold value, and finding a plurality of fault lines related to the fault if the fault exists; and the network connection module is used for establishing network data transmission between the detection end and the terminal based on the created ground signal transfer stations. A tree crown management framework is formed; after any point of fault occurs, the fault is fed back in sequence through the group-level branches, so that the fault of the intelligent electric control equipment or the fault of the power supply line can be quickly judged, and the identification degree and the processing efficiency of the power grid fault are improved.

Description

Technical Field

[0001] This invention belongs to the field of electrical equipment condition monitoring, and particularly relates to an artificial intelligence-based power distribution network fault analysis and feedback system. Background Technology

[0002] With the progress of urbanization and the expansion of production and life, the power grid system is now very large. Power supply is affected by high temperature, electricity consumption, aging and failure of electrical control components, which can lead to distribution network failures. Once a failure occurs, it is necessary to isolate the distribution network failure in a timely, effective and accurate manner and restore power supply to the non-faulty sections to ensure a safe and stable power supply.

[0003] To address these issues, the smart grid was developed, aiming to provide more intelligent analysis and location of power supply faults and faulty segments, thereby facilitating the daily maintenance and emergency response of the power grid.

[0004] Patent application CN 117686841 A, published on March 12, 2024, discloses a "Distribution Network Fault Analysis Method and Device". The method includes: receiving alarm signals from various feeder terminals in the distribution network, wherein the alarm signals include at least one of the following: protection activation signals and undervoltage signals; when the alarm signals meet preset fault analysis activation conditions, determining a feeder diagram reflecting the distribution network topology; if the alarm signals include protection activation signals, determining that a feeder fault exists in the distribution network, and locating the fault based on the feeder diagram and the protection activation signals; if the alarm signals do not include protection activation signals, determining whether a substation fault exists or not based on the number of feeder terminals sending undervoltage signals.

[0005] The invention patent with authorization announcement date of July 9, 2019 and authorization announcement number CN 106841909 B discloses a "multi-source information fusion distribution network fault analysis based on directional hierarchical fuzzy Petri net", including the following steps: Step 1: Extract the action information, timing information, and related electrical quantity information of protection devices and circuit breakers from the SCADA system and WAMS system. Establish a row vector based on the protection direction and use the electrical quantity information of protection devices and circuit breakers for information error correction; Step 2: Obtain a fault area boundary circuit breaker from the corrected circuit breaker displacement information, and use a depth-first search method along its de-energized side to quickly determine the fault area. If there is only one element in the fault area, directly output that element as the fault element.

[0006] Patent application CN 120728854 A, published on September 30, 2025, discloses an "Internet of Things-based Intelligent Distribution Network Fault Analysis and Maintenance System." This system includes a multi-source data acquisition and preprocessing module, a dynamic digital twin modeling module, an edge-cloud fusion analysis module, an instantaneous fault topology location module, a self-healing control module, and an intelligent maintenance scheduling module. This invention extracts fault characteristics through edge-side AI inference and combines it with cloud-based big data analysis to achieve global anomaly detection. It adapts to anomaly patterns under different operating conditions through a dynamic balancing coefficient. Real-time edge-side inference quickly generates fault type labels through weighted fusion feature vectors. The global anomaly scoring mechanism significantly improves fault classification accuracy. Simultaneously, by injecting a dynamic digital twin model to update node states and probability density functions, combined with topology weights, fault influence functions, and spatial attenuation coefficients, it quantifies the distribution characteristics of faults in spatial coordinates, accurately identifying high-incidence fault areas.

[0007] In the prior art, including the technical solutions disclosed in the aforementioned patents, the use of network topology as a basis for smart grid fault analysis is a common application. Its basic principle is relatively simple. It mainly involves acquiring the status of existing smart meters, smart switches, and other common networkable smart electrical control devices to determine whether the line at the location of the smart electrical control device is faulty. However, there are still significant problems in identifying whether the device is faulty and whether the circuit fault is due to a fault in the smart electrical control device or a fault in the power supply line. Therefore, manual troubleshooting is still required to determine whether the fault is due to a fault in the smart electrical control device or a fault in the power supply line. If it is a fault in the power supply line, a large number of personnel need to be dispatched for repair, which is one of the reasons for the delay in handling power supply accidents. Summary of the Invention

[0008] The purpose of this invention is to provide an artificial intelligence-based power distribution network fault analysis and feedback system. Using the power grid topology as an index, it forms a tree-like management architecture with power plants as starting points, transformers as subordinate branches, and distribution boxes as branch points. When a fault occurs at any point, feedback is sequentially fed back through group-level branches, enabling rapid determination of whether the fault is in intelligent electrical control equipment or a power supply line. This improves the accuracy of power grid fault identification, allowing for the assignment of appropriate personnel based on the identified hazard level, thus quickly eliminating faulty lines and significantly improving the efficiency of circuit fault handling, thereby ensuring greater power supply for production and daily life.

[0009] The technical solution of this invention is: to provide an artificial intelligence-based distribution network fault analysis and feedback system, characterized by comprising:

[0010] The data management module establishes a tree-canopy management architecture based on the historical circuit layout to form multiple traceable detection lines;

[0011] The data acquisition module acquires feedback data based on the tree canopy management architecture and according to the canopy relationship;

[0012] The data analysis module creates an alert analysis order based on the obtained feedback data;

[0013] The task execution module traces the generated alert analysis order according to the tree canopy management architecture and determines whether a fault exists based on a predetermined threshold. If a fault exists, it finds several faulty lines related to it.

[0014] The network connectivity module establishes a mesh data transmission network between the detection end and the terminal, based on multiple ground signal relay stations.

[0015] Specifically, the canopy management architecture is generated based on the number of canopy height grid units determined by the circuit layout, and a branch f with origin relationship T is created;

[0016] Taking the power plant as the starting point O, the transformer at the branch point of the power plant is T. 1 ......T n Then T 1 The subordinate branch is T 1+1 ......T 1+n ;

[0017] Then the branches of the cross relation T include f1...f n And generate branches with corresponding numbers for f based on the number of T, where:

[0018] The number of f is the air switch QF and smart meter Φ connected to the first branch of T.

[0019] Specifically, the feedback data includes the switch tripping signal and protection action signal of QF, as well as the reading changes of Φ within a predetermined period;

[0020] The data acquisition module includes a data processing unit that processes data sequentially along the same tier of the tree canopy management architecture in 1-second increments. 1 ......T n Sequentially obtain f1...f of the same ladder level n Feedback data, the first feedback data in production;

[0021] The data analysis module includes a data verification unit, which acquires the first feedback data and verifies the f1...f values ​​at the same tier. n Compare the QF state with the changes in the Φ reading:

[0022] If the reading of Φ in f1 changes quantitatively, and the switch tripping signal of QF is open, then it is determined to be normal operation;

[0023] If f n If the reading of Φ changes quantitatively and the switch tripping signal of QF is closed, then it is considered a QF fault.

[0024] If f n If the reading of Φ shows no change and the switch tripping signal of QF is open, then abnormal data is identified.

[0025] Furthermore, if the data analysis module determines that the first feedback data is abnormal, it will proactively generate a data dispatch plan and feed it back to the data acquisition module, waiting for a data response.

[0026] The data acquisition module also includes a data verification unit. If the data analysis module generates the alert analysis order, it creates data test cases based on the alert analysis order, actively acquiring the next tier's T in 1-second increments. 1+1 ......T 1+n subordinates at the same level f 1+1 ...f 1+n Feedback data, producing second feedback data;

[0027] The data verification unit re-verifies the second feedback data; if f 1+1 ...f 1+n If there is only one or more abnormal data, the nth dispatch plan will be generated. After the nth feedback data of the nth dispatch plan has no abnormal data, a report will be generated.

[0028] Furthermore, the generated report includes data from T. n The lower level of the system contains branch projects of the constant data processing results, corresponding geographical information of the branch project, and historical maintenance logs.

[0029] Specifically, the task execution module includes data from multiple alert analysis orders executed under the current multiple data analysis modules, which are used as manually verified data and stored using timestamps;

[0030] And analyze the generated report results of the last warning analysis order follow-up:

[0031] If a QF fault is identified, the next level f n+n If operating normally, then the branch point f is considered to be... n damage;

[0032] If it is determined to be QF abnormal data, the next level fn+n If operating normally, then the branch point f is considered to be... n Manual exclusion is required, and a list of branch point fn coordinate information bound to manually verified data needs to be generated;

[0033] If it is determined to be QF abnormal data, and the next lower level f n+n If the number of data points identified as abnormal exceeds a threshold, the line is considered faulty, and the line is marked with coordinates to generate a list of faulty lines. The threshold is 3.

[0034] Specifically, the task execution module includes:

[0035] The historical log evaluation storage unit processes and arranges the received generated report results in timestamp order.

[0036] Risk assessment unit, if any T n or / and several levels of f in the corresponding branches n+n If the number of downtime days is less than three days, then generate the corresponding f. n+n The branch line fault investigation plan is then reported back to the central office.

[0037] Furthermore, the aforementioned ground signal relay stations are signal receiving stations created based on each branch, and at least three virtual machines connected to the terminal network are created within each signal receiving station, wherein the network connection is based on 5G, 4G, and 3G.

[0038] The AI-based power distribution network fault analysis and feedback system described in this invention forms a tree-like management architecture, starting from the power plant, with transformers as subordinate branches, and subordinate distribution boxes as branches. When a fault occurs at any point, feedback is sequentially fed back through group-level branches to determine whether the fault lies with intelligent electrical control equipment or the power supply line. This improves the accuracy of power grid fault identification and allows for the assignment of appropriate personnel based on the identified hazard level, enabling rapid troubleshooting of faulty lines and significantly improving the efficiency of circuit fault handling. This, in turn, ensures greater power supply for production and daily life.

[0039] The AI-based distribution network fault analysis and feedback system described in this invention improves the timeliness, accuracy, and completeness of equipment fault impact range analysis, enhances information interaction efficiency and collaborative processing efficiency, and raises the level of intelligence in distribution network fault services by applying integrated intelligent analysis and interactive business in the field of distribution network fault information.

[0040] Compared with the prior art, the advantages of the present invention are:

[0041] 1. The technical solution of the present invention uses the power grid topology as an index, and forms a tree-like management architecture by taking power plants as the starting point, transformers as subordinate branch points, and subordinate distribution boxes as branch points.

[0042] 2. The technical solution of this invention adopts a tree canopy management architecture; thus, when any fault occurs, feedback is sequentially fed back through group-level branches, which can quickly determine whether it is a fault of intelligent electrical control equipment or a fault of power supply line, thereby improving the identification of power grid faults. Based on the identified danger situation, the corresponding number of people can be assigned to quickly eliminate the faulty line, greatly improving the efficiency of circuit fault handling, and thus ensuring a greater guarantee of electricity for production and daily life.

[0043] 3. The technical solution of this invention uses the power grid topology as an index to conduct reliability index analysis on faults of main transformers, lines, and users, focusing on whether they cause a single power source or whether there is a power outage; for faults of lines and overhead lines, it conducts cable information analysis in the case of mixed cables; and in the case of power outages caused by equipment failures, it conducts a comprehensive analysis of the outage range focusing on the outage section, the number of outage users, and the outage area. This comprehensively improves the timeliness, accuracy, and completeness of the analysis of the impact range of equipment failures, provides data support for dispatching to report fault information and handle faults, thereby improving information interaction efficiency and collaborative processing efficiency, and enhancing the intelligence level of distribution network fault services. Attached Figure Description

[0044] Figure 1 This is a schematic diagram of the modules provided in an embodiment of the present invention;

[0045] Figure 2 A logical diagram of the tree crown management architecture for the topology provided in this embodiment of the invention. Detailed Implementation

[0046] The present invention will be further described below.

[0047] like Figures 1 to 2 As shown, the present invention provides an artificial intelligence-based distribution network fault analysis and feedback system, comprising:

[0048] The data management module establishes a tree-canopy management architecture based on the historical circuit layout to form multiple traceable detection lines;

[0049] The data acquisition module is based on the canopy management architecture and obtains feedback data according to the canopy relationship;

[0050] The data analysis module creates alert analysis orders based on the obtained feedback data.

[0051] The task execution module traces the generated alert analysis orders according to the tree canopy management architecture and determines whether a fault exists based on a predetermined threshold. If a fault exists, it finds several faulty lines related to it.

[0052] The network connectivity module establishes a mesh data transmission network between the detection end and the terminal, based on multiple ground signal relay stations.

[0053] Specifically, the canopy management architecture is generated based on the circuit layout to determine the number of canopy height grid cells and to create a branch f with an origin-point relationship T.

[0054] Taking the power plant as the starting point O, the transformer at the branch point of the power plant is T. 1 ......T n Then T 1 The subordinate branch is T 1+1 ......T 1+n ;

[0055] Then the branches of the cross relation T include f1...f n And generate branches with corresponding numbers for f based on the number of T, where:

[0056] The number of f is the air switch QF and smart meter Φ connected to the first branch of T.

[0057] Feedback data includes the switch tripping signal and protection action signal of QF, as well as the reading changes of Φ within a predetermined period;

[0058] The data acquisition module includes a data processing unit, which processes data sequentially along the same tier of the tree canopy management architecture in 1-second increments. 1 ......T n Obtain f of the same level in sequence 1. ....f n Feedback data, the first feedback data in production;

[0059] The data analysis module includes a data verification unit, which acquires the first feedback data and verifies the f1...f values ​​at the same tier. n Compare the QF state with the changes in the Φ reading:

[0060] If the reading of Φ in f1 changes quantitatively, and the switch tripping signal of QF is open, then it is determined to be normal operation;

[0061] If f n If the reading of Φ changes quantitatively and the switch tripping signal of QF is closed, then it is considered a QF fault.

[0062] If f nIf the reading of Φ shows no change and the switch tripping signal of QF is open, then it is considered that there is abnormal data.

[0063] Furthermore, if the data analysis module determines that the first feedback data is abnormal, it will proactively generate a data dispatch plan and send it back to the data acquisition module, waiting for the data response.

[0064] The data acquisition module also includes a data verification unit. If the data analysis module generates a warning analysis order, it creates data test cases based on the warning analysis order and actively acquires the next tier's T data in 1-second increments. 1+1 ......T 1+n subordinates at the same level f 1+1 ...f 1+n The feedback data is used to produce the second feedback data.

[0065] The data verification unit re-verifies the second feedback data. If f 1+1 ...f 1+n If there is only one or more abnormal data, the nth dispatch plan will be generated. After the nth feedback data of the nth dispatch plan has no abnormal data, a report will be generated.

[0066] It should be noted that the generated report includes data from T. n The lower-level components include branch projects with regular data processing results, corresponding geographical information for each branch project, and historical maintenance logs.

[0067] Specifically, the task execution module includes processing data from multiple alert analysis orders executed under the current multiple data analysis modules, using this data for manual verification, and storing it with timestamps; it also analyzes the generated report results following the last alert analysis order.

[0068] If a QF fault is identified, the next level f n+n If operating normally, then the branch point f is considered to be... n damage;

[0069] If it is determined to be QF abnormal data, the next level f n+n If operating normally, then the branch point f is considered to be... n Manual exclusion is required, and a list of branch point fn coordinate information bound to manually verified data needs to be generated;

[0070] If it is determined to be QF abnormal data, and the next lower level f n+n If the number of data points identified as abnormal exceeds the threshold, the line is considered faulty, and the line is marked with coordinates to generate a list of faulty lines. The threshold is 3.

[0071] Furthermore, the task execution module includes:

[0072] The historical log evaluation storage unit processes and arranges the received generated report results in timestamp order.

[0073] Risk assessment unit, if any T n or / and several levels of f in the corresponding branches n+n If the number of downtime days is less than three days, then generate the corresponding f. n+n The branch line fault investigation plan is then reported back to the central office.

[0074] Multiple ground signal relay stations are signal receiving stations created based on each branch, and at least three virtual machines (VMs) connected to the terminal network are created within each signal receiving station. The network connection is based on 5G, 4G, and 3G. That is, the three VMs are respectively 5G, 4G, and 3G. However, when the terminal executes the data acquisition module, it processes the feedback data provided by the above three VMs in real time.

[0075] 1. Technical route of the present invention:

[0076] Real-time fault alarm information, including fault data from main transformers, lines, and users, is acquired from the D5000 system. The collected data undergoes preprocessing, including data cleaning, formatting, and outlier handling, to ensure data quality and accuracy. A network model of the power grid is constructed using grid topology data, including nodes (substations, users, etc.) and edges (transmission lines). The grid topology is analyzed to identify key nodes and lines, as well as their interdependencies. Reliability index analysis, such as Mean Time Between Failures (MTBF) and Mean Time to Repair (MTBR), is conducted for single-source faults and power outages. These indicators are used to assess the grid's reliability level and identify potential weaknesses. For line and overhead line faults, cable information in mixed cable configurations is analyzed, including cable type, length, and aging. The distribution and operational status of different cable types in the grid, as well as their response and impact during faults, are assessed. In cases of power outages caused by equipment failures, the affected circuit segment, number of affected users, and affected transformer area are analyzed to determine the outage scope and impact. Using the grid topology model and real-time data, the impact of faults on grid operation is simulated, and the outage area and duration are predicted. Based on different interaction methods and needs, construct fault information reporting templates, including various formats such as text, charts, and graphs. Templates should include key fault information, impact scope, and possible recovery time to facilitate quick understanding and response by dispatchers. Provide data support to assist dispatchers in fault information reporting and fault handling decisions. Utilize fault handling knowledge graphs and artificial intelligence technologies to assist in generating fault handling operation strategies.

[0077] 2. Real-time monitoring and alarm information processing:

[0078] 2.1 Data Acquisition:

[0079] The D5000 system collects real-time power grid operation data, enabling timely detection and reporting of various fault alarms. The system employs efficient data processing technology to rapidly analyze and classify alarm information, ensuring accuracy and timeliness. Simultaneously, the system uses power grid topology as an index to combine alarm information with the power grid structure, providing a foundation for subsequent analysis.

[0080] 2.2 Data Processing:

[0081] The collected data from the D5000 system is organized and categorized according to different power stations, operational conditions, and alarm classifications. Based on the power grid topology and geographical location information, the power station (e.g., power plant, substation) to which each data record belongs is identified and labeled. Data records from the same power station are aggregated to form a power station-level dataset. This facilitates subsequent fault analysis and performance evaluation by power station. Based on the operational information in the data records (e.g., circuit breaker tripping, protection device operation), different operational types are identified and classified. For each operational type, the sequence and logical relationships of the actions are analyzed to form an operational sequence. This helps in understanding the development process and causes of faults. Based on the characteristics and content of alarm information, it is divided into different alarm types (e.g., abnormal, displacement, limit violation, accident, etc.). Based on the severity and urgency of the alarm, an appropriate level (e.g., urgent, important, general, etc.) is assigned to each alarm. This helps dispatchers quickly identify and handle critical alarms. By combining power grid operating status and historical data, the cause of each alarm is analyzed, and an alarm analysis report is generated. This helps prevent similar alarms from recurring.

[0082] 2.3 Data Analysis:

[0083] 2.3.1 Line Fault Analysis:

[0084] By integrating distribution network topology data into a line ledger tree list, a power grid structure model is constructed, incorporating rich line ledger information such as line type, length, commissioning date, and historical maintenance records. Based on this, when a power grid fault occurs, a preliminary judgment can be quickly made based on fault information, such as fault location, type, and affected area. Through advanced algorithms and logical analysis, the system organizes and summarizes the range of lines directly or indirectly affected by the fault. This relies first on the accurate capture of fault information, including the action records of protection devices and real-time feedback from on-site inspection personnel. Subsequently, the system uses distribution network topology data to identify line segments, branch lines, and even the entire power supply area directly or indirectly connected to the fault point by traversing the line ledger tree list. During this process, the algorithm comprehensively considers the electrical connectivity of the lines, switch status (closed or open), load transfer strategies, and the coordination logic of protection devices to ensure accurate definition of the affected area. The system also automatically matches and counts the number of users affected by the fault based on user information recorded in the line ledger, such as user number, electricity usage type, and power supply line. This considers users directly connected to the faulty line, as well as those indirectly affected by load shifting due to the fault, ensuring the comprehensiveness and accuracy of the statistical results. Simultaneously, the system will categorize affected users based on their electricity usage characteristics (e.g., important users) and electricity demand (e.g., critical loads, non-critical loads), providing strong support for subsequent fault repair, load restoration, and user communication.

[0085] 2.3.2 Lazy loading of ledger lines:

[0086] Based on the summary display of the ledger's route tree, due to the large amount of data, lazy loading was considered. Lazy loading, also known as deferred loading or on-demand loading, is a common technique for optimizing the performance of web pages or applications. It refers to loading only the necessary resources or data during the initial load, and then dynamically loading these resources or data when the user needs to view more content. In a tree control, lazy loading typically manifests as loading only the direct child nodes of the current node, and loading the child node data when the user expands a node.

[0087] The lazy loading method takes two parameters: one is the information of the current node (including its hierarchy data); the other is a method to re-render the child nodes under the current node (resolve), which takes an array and displays them according to the mapping relationship in props.

[0088] Note: The lazy loading method (:load) is executed once during the initial load, and then triggered again each time the arrow in front of a node is clicked to get its child nodes; it will not be triggered again even if the data loaded initially changes, nor will it be triggered again when the node is collapsed after clicking to expand its child nodes.

[0089] Since lazy loading retrieves data level by level, each level needs to use `resolve` to render its child nodes. If there is no content to display under a node, it will keep spinning. In this case, `resolve` needs to be set to return an empty array. This way, if it does not retrieve the content of the child nodes, it will display an empty array after spinning (and remove the downward arrow at the beginning), and will not keep spinning.

[0090] Real-world scenario: The ID of each layer can be retrieved via `node`, and an API can be called based on this ID to obtain data. Then, `resolve` is used to output the returned data, which consists of the child nodes associated with that ID.

[0091] 2.3.4 Secondary packaging:

[0092] Scenario: Since tree controls are used in many places and the data to be displayed is different, the tree control is encapsulated in another layer, and the tree diagram is displayed differently according to different parameters passed from external components.

[0093] The approach is to listen for changes in externally input data and re-render the tree to display different data. However, `:load` only loads the tree once initially and retrieves the node currently bound to it. If `loadTree` is called again when listening for data later, it cannot retrieve the node and resolve, leading to rendering failure. In this case, we can use `:data` to display the data. When there are nodes in the tree, `:load` can be triggered normally to perform lazy loading of child nodes.

[0094] 2.3.5 Data Display:

[0095] Scenario: When setting checkboxes on a lazy-loaded tree, the previously added lazy-loaded selected parts need to be displayed in the table editor.

[0096] Approach: Since lazy-loaded data is retrieved level by level, we can use the `default-expanded-keys` and `default-checked-keys` properties to set the nodes that need to be displayed during the tree rendering process. (Note the importance of using these two properties.)

[0097] Echo issue:

[0098] Scenario: When using lazy loading to display data, the following situation occurs when adding selected data: the first and second child nodes under a parent node are selected at the same time, and the default selected node array obtained during display only contains the IDs of these two nodes. However, the final lazy-loaded display data shows that all child nodes under this parent node are selected (for other similar situations).

[0099] Analysis: Since lazy-loaded trees are loaded asynchronously, when the tree is determining whether a child node is selected, the selected child node may cause its parent node to be calculated and determined as selected due to the association.

[0100] Solution: If the parent node's checkbox is not needed, or the parent node has no checkbox and only the child node has it, or if the parent-child node relationship is not required, the check-strictly property can be used to disconnect the connection between the parent and child nodes.

[0101] 2.4 Self-Switching Station Analysis:

[0102] Based on the collected data and power network topology model, fault detection and impact range analysis will be developed. It needs to be able to acquire D5000 monitoring data in real time to quickly locate the fault point and accurately analyze the impact range of the fault on the automatic disconnection station.

[0103] By implementing this system, real-time monitoring, rapid location, and impact range analysis of power line faults (V) can be achieved, facilitating automatic disconnection operations. Expected benefits include improved efficiency and accuracy in fault handling, ensured stable operation of the power supply system, and reduced economic losses caused by faults.

[0104] To evaluate system performance, a series of evaluation metrics can be set, such as fault detection accuracy, clearing time, and power restoration time, and the system can be evaluated and optimized regularly.

[0105] 2.5 Analysis of Circuit Stoppage Segments:

[0106] In response to power outages caused by equipment failures, a comprehensive analysis is conducted from multiple dimensions, including the affected circuit segment, the number of affected users, and the affected electrical area, to quickly and accurately determine the scope of the power outage and provide strong support for subsequent emergency repair and restoration work.

[0107] 2.5.1 Objective:

[0108] Precisely locate power outage sections: Through real-time monitoring and analysis, quickly identify the power line sections affected by the fault.

[0109] Counting the number of users affected by power outages: Accurately counting the number of users affected by power outages provides a basis for emergency response and resource allocation.

[0110] Identify affected distribution radio areas: Identify and mark all affected distribution radio areas to quickly locate repair points.

[0111] 2.5.2 Analysis of Power Outage Area:

[0112] Power outage impact analysis: Using topology models and fault location information, algorithms simulate the impact of faults on the power network to determine key information such as affected circuit sections, number of affected customers, and affected transformer areas.

[0113] Data logging: Records key information for each power outage event, including fault type, affected circuit segment, number of affected users, affected area, and repair time.

[0114] Data analysis: Regularly analyze power outage event data to identify frequently occurring fault areas and causes, providing a basis for preventative maintenance.

[0115] 2.5.3 Analysis of Users Affected by Power Outages:

[0116] When obtaining faulty lines based on D5000, it is necessary to analyze the impact on users in real time, based on the number of affected users, the number of affected transformer areas, and the number of important users.

[0117] By using D5000 data analysis to track the location of faulty lines, the system can quickly calculate the specific extent of the power outage, including the number of affected residential and commercial users, the number of distribution substations involved, and the number of users in critical facilities essential to the community's operation (such as key customers). This real-time analysis capability enables more effective allocation of repair resources, prioritizing power outages with the greatest impact on public safety and social welfare, thereby ensuring rapid power restoration and mitigating the inconvenience and losses caused by the fault.

[0118] 3. Visualized data summary and display:

[0119] The visualization is divided into 6 parts: alarm model statistics, power outage impact analysis, equipment failure statistics, site map, real-time fault list, and load impact.

[0120] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0121] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0122] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0123] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0124] In summary, the technical solution of this invention forms a tree-like management architecture by starting with the power plant, using transformers as subordinate branches, and sub-distribution boxes as branches. When a fault occurs at any point, feedback is sequentially relayed through group-level branches to determine whether the fault lies with the intelligent electrical control equipment or the power supply line. This improves the identification of power grid faults and allows for the assignment of appropriate personnel based on the identified hazard level, enabling rapid troubleshooting of faulty lines. This significantly improves the efficiency of handling circuit faults and ensures greater power supply for production and daily life.

[0125] The technical solution of this invention, through its application in the field of integrated intelligent analysis and interactive business of distribution network faults, can improve the timeliness, accuracy, and completeness of equipment fault impact range analysis, enhance information interaction efficiency and collaborative processing efficiency, and improve the intelligence level of distribution network fault services.

[0126] This invention can be widely used in the field of power distribution network operation scheduling management and fault monitoring.

Claims

1. An artificial intelligence-based distribution network fault analysis and feedback system, characterized in that: include: The data management module establishes a tree-canopy management architecture based on the historical circuit layout to form multiple traceable detection lines; The data acquisition module acquires feedback data based on the tree canopy management architecture and according to the canopy relationship; The data analysis module creates an alert analysis order based on the obtained feedback data; The task execution module traces the generated alert analysis order according to the tree canopy management architecture and determines whether a fault exists based on a predetermined threshold. If a fault exists, it finds several faulty lines related to it. The network connectivity module establishes a mesh data transmission network between the detection end and the terminal, based on multiple ground signal relay stations.

2. The artificial intelligence-based distribution network fault analysis and feedback system according to claim 1, characterized in that: The canopy management architecture is generated based on the number of canopy height grid cells determined by the circuit layout, and a branch f with origin relationship T is created. Taking the power plant as the starting point O, the transformer at the branch point of the power plant is T. 1 ......T n Then T 1 The subordinate branch is T 1 +1 ......T 1+n ; Then the branches of the cross relation T include f1...f n And generate branches with corresponding numbers for f based on the number of T, where: The number of f is the air switch QF and smart meter Φ connected to the first branch of T.

3. The artificial intelligence-based distribution network fault analysis and feedback system according to claim 1, characterized in that: The feedback data includes the switch tripping signal and protection action signal of QF, as well as the reading changes of Φ within a predetermined period; The data acquisition module includes a data processing unit that processes data sequentially along the same tier of the tree canopy management architecture in 1-second increments. 1 ......T n Sequentially obtain f1...f of the same ladder level n Feedback data, the first feedback data in production; The data analysis module includes a data verification unit, which acquires the first feedback data and verifies the f1...f values ​​at the same tier. n Compare the QF state with the changes in the Φ reading: If the reading of Φ in f1 changes quantitatively, and the switch tripping signal of QF is open, then it is determined to be normal operation; If f n If the reading of Φ changes quantitatively and the switch tripping signal of QF is closed, then it is considered a QF fault. If f n If the reading of Φ shows no change and the switch tripping signal of QF is open, then abnormal data is identified.

4. The artificial intelligence-based distribution network fault analysis and feedback system according to claim 3, characterized in that: If the data analysis module determines that the first feedback data is abnormal, it will actively generate a data dispatch plan and feed it back to the data acquisition module, waiting for the data response. The data acquisition module also includes a data verification unit. If the data analysis module generates the alert analysis order, it creates data test cases based on the alert analysis order, actively acquiring the next tier's T in 1-second increments. 1 +1 ......T 1+n subordinates at the same level f 1+1 ...f 1+n Feedback data, producing second feedback data; The data verification unit re-verifies the second feedback data; if f 1+1 ...f 1+n If there is only one or more abnormal data, the nth dispatch plan will be generated. After the nth feedback data of the nth dispatch plan has no abnormal data, a report will be generated.

5. The artificial intelligence-based distribution network fault analysis and feedback system according to claim 4, characterized in that: The generated report includes data from T. n The lower level of the system contains branch projects of the constant data processing results, corresponding geographical information of the branch project, and historical maintenance logs.

6. The artificial intelligence-based distribution network fault analysis and feedback system according to claim 1, characterized in that: The task execution module includes data from multiple alert analysis orders executed under the current multiple data analysis modules, which are used as manually verified data and stored using timestamps; And analyze the generated report results of the last warning analysis order follow-up: If a QF fault is identified, the next level f n+n If operating normally, then the branch point f is considered to be... n damage; If it is determined to be QF abnormal data, the next level f n+n If operating normally, then the branch point f is considered to be... n Manual exclusion is required, and a list of branch point fn coordinate information bound to manually verified data needs to be generated; If it is determined to be QF abnormal data, and the next lower level f n+n If the number of data points identified as abnormal exceeds a threshold, the line is considered faulty, and the line is marked with coordinates to generate a list of faulty lines. The threshold is 3.

7. The artificial intelligence-based distribution network fault analysis and feedback system according to claim 1, characterized in that: The task execution module includes: The historical log evaluation storage unit processes and arranges the received generated report results in timestamp order. Risk assessment unit, if any T n or / and several levels of f in the corresponding branches n+n If the number of downtime days is less than three days, then generate the corresponding f. n+n The branch line fault investigation plan is then reported back to the central office.

8. The artificial intelligence-based distribution network fault analysis and feedback system according to claim 1, characterized in that: The aforementioned ground signal relay stations are signal receiving stations created based on each branch, and at least three virtual machines connected to the terminal network are created within each signal receiving station, wherein the network connection is based on 5G, 4G and 3G.

9. The artificial intelligence-based distribution network fault analysis and feedback system according to claim 1, characterized in that: The aforementioned AI-based power distribution network fault analysis and feedback system forms a tree-like management architecture, starting from the power plant, with transformers as subordinate branches, and subordinate distribution boxes as branches. When a fault occurs at any point, feedback is sequentially fed back through the group-level branches to determine whether the fault lies with intelligent electrical control equipment or the power supply line. This improves the accuracy of power grid fault identification and allows for the assignment of appropriate personnel based on the identified hazard level, enabling rapid troubleshooting of faulty lines and significantly improving the efficiency of circuit fault handling. This, in turn, ensures greater assurance of electricity supply for production and daily life.

10. The artificial intelligence-based distribution network fault analysis and feedback system according to claim 1, characterized in that: The aforementioned AI-based distribution network fault analysis and feedback system improves the timeliness, accuracy, and completeness of equipment fault impact range analysis, enhances information interaction efficiency and collaborative processing efficiency, and raises the level of intelligence in distribution network fault services by applying integrated intelligent analysis and interactive business in the field of distribution network fault information.

Images