Intelligent power distribution system line fault positioning and self-healing method

Abstract

The invention discloses an intelligent power distribution system line fault positioning and self-healing method which is realized based on a multi-module intelligent power distribution system. A1, the data acquisition module acquires multiple parameters and preprocesses the parameters; a2, a fault positioning module positions a fault point with the error smaller than or equal to 50 m and classifies fault types; a3, the self-healing control module is switched off within 30 ms and reconstructs the power distribution network within 30 seconds; and the A4 cooperative control module realizes distributed cooperative control and optimizes a self-healing strategy. Through the technologies of multi-sensor fusion, edge calculation, deep learning and the like, the problem that the self-healing success rate of the 10kV distribution network line in the harbor district is low is solved. According to the method, the self-healing success rate can be improved, the fault positioning and recovery time can be shortened, the power failure loss can be reduced, the operation and maintenance cost can be reduced, and the reliability and intelligent level of a power distribution system can be improved.

Description

Technical Field

[0001] The present invention relates to the field of smart grid and power system automation, and in particular to a method for locating and self-healing line faults in a smart power distribution system. Background Art

[0002] In modern society, the stability and reliability of power supply are crucial for ensuring social production and people's livelihoods. With the continuous advancement of science and technology, the construction of smart grids has become a key development direction for the power industry. Distribution network self-healing technology, as a key means of improving power supply reliability, has attracted widespread attention. Against this backdrop, State Grid Zhengzhou Airport Power Supply Company faced the challenge of improving the success rate of self-healing faults on its 10kV self-healing lines in the port area. This challenge also formed the core technical background content of this research.

[0003] Distribution network self-healing technology is a crucial component of smart grids and plays an irreplaceable role in improving power supply reliability. Traditional distribution network faults often require manual troubleshooting and repair, which not only consumes significant time and labor costs but also prolongs power outages, impacting users' electricity needs. Self-healing lines, on the other hand, can automatically detect, isolate faults, and restore power, significantly shortening troubleshooting time and minimizing the scope of outages. For example, if a short circuit occurs on a line in the distribution network, the self-healing line can quickly locate the fault point, automatically isolate the faulty area, and simultaneously restore power to non-faulty areas, making the outage virtually imperceptible to most users. This is crucial for ensuring the continuity of production, especially for businesses with stringent requirements for power supply stability, such as electronics manufacturers and financial institutions, as well as for ensuring the normal lives of residents.

[0004] State Grid Zhengzhou Airport Power Supply Company deployed 10kV self-healing lines in the port area to improve power supply reliability. However, actual operational data shows that the overall self-healing success rate of the port area's 10kV distribution lines was only 35.1% in the first half of 2023, far below the normal 90% success rate. The success rates for July to December 2023 were 34.62% in July, 28.57% in August, 22.22% in September, 47.62% in October, 29.17% in November, and 52% in December, showing significant monthly fluctuations and remaining at a relatively low level overall. This situation indicates that the implementation of the self-healing function of the port area's 10kV distribution lines is seriously flawed and requires urgent resolution.

[0005] However, existing technologies have limitations in addressing the low success rate of self-healing on the 10kV distribution network lines in the port area. While the quality of secondary equipment operation and maintenance is recognized as a key issue, the port company's distribution network secondary equipment, built earlier as a pilot area, is heavily outdated. The functionality and hardware performance of this equipment no longer meet the requirements of modern distribution network self-healing. Upgrading existing equipment, while initially cost-effective, offers limited performance improvements, and functional expansion is limited by the hardware and architecture of the original equipment, making it ineffective in addressing the issue. While directly replacing the distribution network with newer equipment can fundamentally address the issue of outdated equipment, it faces challenges such as high costs and prolonged power outages. Analyzing logical errors in the master station system requires in-depth optimization of the system's algorithms and data processing processes, which requires significant manpower and time for research and testing. Replanning and renovating the distribution network structure, addressing issues with inappropriate network architecture, is a massive undertaking involving extensive line laying and equipment replacement, resulting in high costs and significant implementation challenges. Although measures such as regular inspections and environmental improvements can be taken to address quality issues with distribution network switchgear and equipment operating environment, it is difficult to completely avoid equipment failures.

[0006] With the continuous development of smart grid technology, technologies such as distribution automation systems (DAS), supervisory control and data acquisition (SCADA), geographic information systems (GIS), and demand-side management (DSM) are increasingly being applied in distribution networks. These developments offer new approaches and methods for addressing the low self-healing success rate of 10kV distribution lines in the port area. For example, advanced data analysis techniques and artificial intelligence algorithms can further optimize the master station system's analysis and judgment logic, improving the accuracy of fault diagnosis. IoT technology enables real-time monitoring and remote control of distribution network equipment, enabling timely detection and resolution of equipment faults. Big data technology enables analysis of large amounts of operational data, uncovering potential issues and patterns, and providing a basis for equipment maintenance and upgrades. However, the application of these new technologies also faces challenges, such as technical compatibility and data security, which need to be addressed in practical applications.

[0007] In summary, State Grid Zhengzhou Airport Power Supply Company faces numerous challenges in improving the success rate of self-healing faults on its 10kV self-healing lines in the port area, and existing technical approaches have certain limitations. Against the backdrop of continuous technological advancements in the industry, finding a practical solution to improve the success rate of self-healing distribution network lines by integrating new technologies is a critical and pressing research topic. Summary of the Invention

[0008] The technical problem to be solved by the present invention is a method for locating and self-healing line faults in an intelligent power distribution system. The method is implemented based on a multi-module intelligent power distribution system and solves a series of technical problems in locating and self-healing line faults in a power distribution system.

[0009] In order to solve the above technical problems, the technical solutions adopted by the present invention are as follows:

[0010] A method for locating and self-healing line faults in an intelligent power distribution system is described. The method is implemented based on a multi-module intelligent power distribution system. The system includes an A1 data acquisition module, an A2 fault location module, an A3 self-healing control module, and an A4 collaborative control module. The specific details are as follows:

[0011] A1 data acquisition module: This module collects voltage, current, zero-sequence components, and environmental parameters in real time through intelligent terminals deployed on distribution lines. It uses algorithms to eliminate noise and pre-processes data using edge computing.

[0012] A2 fault location module: The master station system receives pre-processed data and locates the fault point with an error of ≤50 meters based on the newly developed positioning submodule. It also uses the deep learning submodule to classify the fault type.

[0013] A3 self-healing control module: Generates the optimal isolation strategy based on a new algorithm, controls the intelligent switch to open within 30ms, and reconstructs the distribution network through this algorithm, with a recovery time of ≤30 seconds;

[0014] A4 collaborative control module: realizes distributed collaborative control of the master station and terminals through a multi-agent system, and uses reinforcement learning to dynamically optimize the self-healing strategy.

[0015] As a preferred technical solution of the present invention, the A1 data acquisition module includes:

[0016] A1.1 multi-sensor fusion unit, integrating Rogowski coil, fiber Bragg grating sensor and temperature and humidity transmitter;

[0017] A1.2 edge computing unit, using a multi-core processor to perform local data preprocessing;

[0018] The algorithm described in A1.3 specifically: automatically optimizes the accuracy of the prediction model through real-time analysis of environmental changes. Unlike traditional intelligent distribution system line fault location and self-healing methods that rely on fixed rules, it continuously compares the differences between the predicted results and the actual observed data and dynamically adjusts the model error range accordingly: when the difference increases, the error range is expanded to accommodate uncertainty; when the difference decreases, the range is narrowed to improve accuracy. This process is achieved through a continuous cycle of "prediction-comparison-adjustment", ensuring that the model always adapts to actual changes and maintains stable performance in complex dynamic environments.

[0019] As a preferred technical solution of the present invention, the A2 fault location module includes:

[0020] The positioning submodule described in A2.1: When a line fault occurs, intelligent devices at both ends immediately detect the sudden change in current or voltage and perform high-precision time synchronization via a dedicated optical fiber link. The devices automatically freeze the timestamp upon signal detection, reducing data processing delays. Simultaneously, they automatically simulate faults daily to calibrate processing time to maintain accurate recorded time. The system dynamically adjusts the actual signal propagation speed based on the time difference between the fault signal's arrival at both ends, combined with a fiber propagation speed benchmark and factors including, but not limited to, conductor temperature and aging, to calculate the approximate location of the fault. The positioning results are superimposed on a power map and further verified using optical fiber detection technology, ensuring an error of no more than half a meter. If the device processing delay exceeds a set threshold, the system switches to pure optical fiber positioning mode, ensuring rapid identification of the fault point even in extreme situations.

[0021] The deep learning submodule described in A2.2 is implemented based on the characteristics of current and voltage signals with different characteristics generated by distribution line faults. First, the signal is scanned layer by layer through multiple layers of "filters" to capture local abnormal fluctuations, including but not limited to sudden spikes and depressions. Then, the model records the temporal variation of these fluctuations and analyzes whether the abnormal signal is continuous or short-lived. Finally, all information is integrated to quickly determine the type of fault.

[0022] As a preferred technical solution of the present invention, the A3 self-healing control module includes:

[0023] A3.1 Intelligent switch control submodule: uses permanent magnet mechanism circuit breaker, opening and closing time ≤ 30ms;

[0024] A3.2 Network Reconfiguration Submodule: Optimizing the power supply efficiency of a distribution network is essentially a process of finding the optimal path. The algorithm is a tool for determining this optimal path. It first randomly generates multiple possible network configurations, including the on / off states of switches and the locations of distributed generation (DG) connections. Each solution is then scored based on pre-defined objectives, including but not limited to reducing losses and improving power supply reliability. High-scoring solutions are retained, while low-scoring solutions are eliminated. The advantages of high-quality solutions are then combined and some details are randomly adjusted to generate new candidate solutions. Finally, through continuous iterative optimization, the optimal network configuration is found.

[0025] A3.3 self-healing verification submodule verifies the isolation effect through the differential protection principle.

[0026] As a preferred technical solution of the present invention, the A4 collaborative control module includes:

[0027] A4.1 Multi-agent system submodule: Each regional agent negotiates a power restoration plan through a contract network protocol;

[0028] A4.2 Reinforcement Learning Optimization Submodule: This module dynamically updates self-healing strategies based on a fault case database. Specifically, the system automatically analyzes historical cases and finds that "isolating the faulty branch line first, then restoring the main line" can shorten recovery time by at least 50%. Through reinforcement learning, the system increases the reward value of the "branch line priority isolation" strategy and applies it to subsequent faults.

[0029] A4.3 Digital twin verification submodule: Evaluates the feasibility of the self-healing solution through real-time simulation. Specifically: The digital twin verification submodule builds a virtual model that is completely consistent with the actual power grid by replicating the operating status of the real distribution network in real time, including but not limited to equipment parameters, distributed power output and user load; the system simulates fault scenarios in a virtual environment, including but not limited to short circuits and line breaks, and conducts a preview verification of the self-healing solution generated by the master station: voltage stability, equipment load conditions and power supply recovery time are evaluated through power flow calculations; the simulation results automatically optimize the solution, and intuitively display the fault point, power flow and solution comparison through a three-dimensional visualization interface, helping operation and maintenance personnel to make quick decisions, avoid the trial and error risks of the real power grid, and significantly improve the reliability and efficiency of the self-healing solution.

[0030] The beneficial effects of adopting the above technical solutions are as follows: in terms of fault location, the A2 fault location module can accurately locate the fault point with an error of ≤50 meters, and can also quickly classify the fault type so that maintenance personnel can quickly determine the fault location and nature, significantly shortening the fault investigation time and improving maintenance efficiency. In terms of self-healing control, the A3 self-healing control module can control the intelligent switch to open within 30ms when a fault occurs, and complete the distribution network reconstruction and power restoration within 30 seconds, greatly reducing the duration of power outages, ensuring the continuity of users' electricity use, and reducing the impact of power outages on production and life. The collaborative control module optimizes the self-healing strategy and evaluates the feasibility of the solution through multi-agent systems, reinforcement learning and digital twin technologies, improving decision-making accuracy and solution reliability, avoiding trial and error in actual power grid operations, saving manpower and material costs, and comprehensively improving the intelligence level and operational stability of the distribution system. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 : Fault handling flow chart;

[0032] Figure 2 :Installation Figure 1 ;

[0033] Figure 3 :Installation Figure 2 ;

[0034] Figure 4 : Statistics of the self-healing success rate of replaced terminal lines;

[0035] Figure 5 : Comparison chart of the success rate of self-healing actions;

[0036] Figure 6 : A statistical chart showing the reasons for self-healing failure. DETAILED DESCRIPTION

[0037] The principles of the present disclosure will now be described with reference to several exemplary embodiments shown in the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that these embodiments are only described to facilitate those skilled in the art to better understand and implement the present disclosure, and are not intended to limit the scope of the present disclosure in any way.

[0038] In the following description of the embodiments, specific details such as specific system structures and technologies are provided for illustration rather than limitation to facilitate a thorough understanding of the embodiments of the present application. However, it will be apparent to those skilled in the art that the present application may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, and methods are omitted to avoid obscuring the description of the present application with unnecessary details.

[0039] It should be understood that when used in this specification and the appended claims, the term "comprising" indicates the presence of described features, integers, steps, operations, components, but does not preclude the presence or addition of one or more other features, integers, steps, operations, components.

[0040] Example 1: Overview of overall steps

[0041] 1. System architecture and module functions

[0042] This method is based on a multi-module collaborative intelligent power distribution system, which consists of the following four core modules:

[0043] Data acquisition module: responsible for real-time collection of voltage, current, zero-sequence components and environmental parameters of the distribution line, and pre-processing of data through edge computing.

[0044] Fault location module: Receives pre-processed data, accurately locates the fault point (error ≤ 50 meters), and uses deep learning technology to classify the fault type.

[0045] Self-healing control module: Generates the optimal isolation strategy, controls the intelligent switch to quickly open (≤30ms), and reconstructs the distribution network to ensure recovery time ≤30 seconds.

[0046] Collaborative control module: Through multi-agent systems and reinforcement learning technology, it realizes distributed collaborative control of the master station and terminals, and dynamically optimizes the self-healing strategy.

[0047] 2. Specific implementation of data acquisition module

[0048] Multi-sensor fusion: Integrates Rogowski coils, fiber grating sensors, and temperature and humidity transmitters to achieve multi-dimensional data collection.

[0049] Edge computing: Use multi-core processors for local data preprocessing to improve data processing efficiency.

[0050] Intelligent algorithm: By analyzing environmental changes in real time, dynamically optimizing the prediction model, and adopting a closed-loop mechanism of "prediction, comparison, and adjustment", it ensures the stability and accuracy of the model in complex environments.

[0051] 3. Specific implementation of the fault location module

[0052] Positioning submodule: Utilizes the sudden change signals of current and voltage when a fault occurs, combined with the signal propagation time difference and propagation speed, to calculate the fault point location, and combines the result with the electronic map to achieve visual positioning of the fault point.

[0053] Deep learning submodule: Analyzes local abnormal fluctuations (such as spikes and sags) of current and voltage signals through a multi-layer neural network, and quickly determines the fault type based on the temporal variation patterns.

[0054] 4. Specific implementation of the self-healing control module

[0055] Intelligent switch control: permanent magnet mechanism circuit breaker is used to ensure the opening and closing time is ≤30ms.

[0056] Network Reconfiguration: Generate multiple network configuration plans through optimization algorithms, score and iteratively optimize based on preset goals (such as reducing loss and improving reliability), and ultimately determine the optimal configuration.

[0057] Self-healing verification: Use the differential protection principle to verify the isolation effect and ensure the reliability of the self-healing operation.

[0058] 5. Specific implementation of collaborative control module

[0059] Multi-agent system: Agents in each region negotiate power restoration plans through contract network protocols to achieve distributed collaborative control.

[0060] Reinforcement learning optimization: Dynamically updates self-healing strategies based on a historical fault case database. For example, if analysis reveals that "isolating the faulty branch line first, then restoring the main trunk line" can shorten recovery time by over 50%, the system will prioritize this strategy.

[0061] Digital twin verification: By building a virtual model consistent with the actual power grid, fault scenarios (such as short circuits and line breaks) are simulated to preview and verify self-healing solutions. The fault point, power flow, and solution comparison are displayed through a three-dimensional visualization interface to assist operation and maintenance personnel in decision-making and reduce the risk of trial and error.

[0062] 6. Method advantages

[0063] High-precision positioning: The fault point positioning error is ≤50 meters, significantly improving fault handling efficiency.

[0064] Fast self-healing: The intelligent switch tripping time is ≤30ms, and the distribution network recovery time is ≤30 seconds, minimizing the impact of power outages.

[0065] Intelligent optimization: Through reinforcement learning and digital twin technology, dynamic optimization of self-healing strategies is achieved to improve system reliability and efficiency.

[0066] Collaborative control: Multi-agent systems achieve distributed collaboration and enhance the flexibility and adaptability of the system.

[0067] This method achieves rapid location, intelligent classification and efficient self-healing of distribution system faults through the collaborative work of multiple modules, significantly improving the reliability and operation and maintenance efficiency of the distribution system.

[0068] Example 2: Fault Location System Based on Multi-Sensor Fusion and Dynamic Model Optimization

[0069] Background: In an industrial park in the port area, with the increasing number of businesses and expansion of production, the demand for reliable power supply is becoming increasingly stringent. However, the park's 10kV distribution network has been experiencing frequent failures, and fault location and resolution have been time-consuming, severely impacting normal production. By early 2024, the self-healing success rate of the park's 10kV distribution network was only 35.1%, significantly lower than the expected 90%. Consequently, the economic losses caused by power outages to businesses continued to increase.

[0070] Processing flow:

[0071] 1. Reference Figure 2 Data acquisition and preprocessing: To address this issue, 53 distribution automation terminals from Manufacturer A (including multi-sensor fusion units) were deployed. Integrated Rogowski coils (current measurement accuracy ±0.5%), fiber Bragg grating sensors (strain measurement accuracy ±1με), and temperature and humidity transmitters (accuracy ±0.5°C / ±3%RH) began collecting data in real time. The edge computing unit uses a multi-core processor and a "prediction-comparison-adjustment" algorithm to dynamically optimize the model's error range. During humid seasons, ambient humidity often exceeds 80%. At this time, the system automatically expands the zero-sequence component measurement error range. Through multiple field measurements, data noise has been reduced by 40%, ensuring data accuracy and reliability.

[0072] 2. Reference Figure 1, Fault location and classification: When a line fault occurs, the positioning submodule begins to play a role. The devices at both ends will accurately record the time of the voltage and current mutation with an accuracy of microseconds. The system calculates the location of the fault point based on the signal propagation speed (200m / μs) and the time difference. In one test, the fault point positioning error was successfully achieved ≤30 meters, and the average error in the actual test was 28 meters. The deep learning submodule conducts deep training on the current and voltage signals of 98 self-healing failure cases from July to December 2023, so that its fault type identification accuracy rate reaches 98.6%.

[0073] Reference Attachment Figure 4 Implementation results: After a period of operation, from July to December 2024, the branch line's self-healing success rate significantly increased from 35.1% to 91.67%. Fault location time was also significantly reduced from an average of 2.3 hours to 47 seconds. For example, in August 2024, a line fault occurred due to insulator breakdown. The system accurately located the fault point within 32 seconds (with an error of 19 meters) and accurately classified it as a permanent fault. This enabled maintenance personnel to quickly arrive at the scene to handle the problem, successfully avoiding power outages for 3,500 users and ensuring the normal production of enterprises in the park.

[0074] Example 3: Second-level self-healing control and network reconstruction system

[0075] Background: An electronics industrial park in the port area houses numerous electronics manufacturers with extremely high demands for power supply stability. The park is equipped with distributed photovoltaic systems to achieve comprehensive energy utilization. However, the previous distribution network's self-healing capabilities were insufficient in the event of failures, resulting in prolonged power outages. This not only impacted production but also wasted distributed photovoltaic energy. By early 2024, the park's distribution network's self-healing success rate was only 38%, with an average recovery time exceeding five minutes, resulting in significant economic losses for businesses.

[0076] Processing flow:

[0077] 1. Reference Figure 1 and 3 , Fault isolation and recovery: In the 10kV distribution network of the park, the intelligent switch control submodule uses a permanent magnet mechanism circuit breaker (opening and closing time ≤ 28ms). In September 2024, a feeder fault occurred. The system quickly disconnected the fault point within 28ms at the moment of detecting the fault, and used the differential protection principle to verify the isolation effect to ensure that the fault area was effectively isolated. At the same time, the network reconstruction submodule started working based on the improved genetic algorithm, randomly generating more than 2,000 candidate network configurations. With the goal of reducing losses and improving reliability, these configurations were iteratively optimized, and each round of optimization time was ≤

[0078] After multiple rounds of screening, the distributed photovoltaic system was finally connected to switch #5, successfully restoring power to 92% of the load within 30 seconds and reducing line losses by 12%.

[0079] 2. Reference Figure 5 Collaborative Optimization and Verification: The reinforcement learning submodule conducted an in-depth analysis of 1,500 historical fault cases and found that the "branch line priority isolation" strategy effectively shortened recovery time. Consequently, the reward value for this strategy was increased by 40%. The digital twin verification submodule simulated the grid's operating status in real time and rehearsed three self-healing solutions. Power flow calculations were used to assess voltage stability, equipment load, and power restoration time, ultimately selecting the optimal path (voltage deviation ≤ 2%, equipment load factor ≤ 85%).

[0080] Implementation Results: Through a series of measures, the park's self-healing success rate reached 94.2% by 2024, and the average recovery time was reduced to 27 seconds, a 92% improvement compared to traditional manual operations. This not only reduced power outage losses by approximately 4.5 million yuan but also increased the distributed photovoltaic utilization rate by 18%, improving energy efficiency and ensuring stable production for park enterprises.

[0081] Example 4: Intelligent Operation and Maintenance System with Multi-Agent Collaboration and Reinforcement Learning

[0082] Background: A mixed commercial and residential area in the Hong Kong area encompasses a large area, encompassing nine power substations and a complex distribution network. In the past, cross-regional line faults were handled with inefficient coordination among the substations, resulting in lengthy troubleshooting times and numerous complaints from residents and businesses regarding power supply reliability. By early 2024, the area's self-healing success rate was only 33%, with a 35% false positive rate and high O&M costs.

[0083] Processing flow:

[0084] 1. Reference Figure 1 Distributed collaborative control: When a cross-regional line fault occurred in November 2024, the multi-agent system submodule coordinated nine regional agents through the contract network protocol. The master station quickly communicated with the three relevant power supply station agents. After two rounds of negotiation, the optimal isolation strategy (disconnecting switches #15 and #22) was determined within 480ms, achieving rapid and effective fault isolation.

[0085] 2. Reference Figure 6Strategy Optimization and Verification: The reinforcement learning submodule dynamically updates strategies based on over 5,000 historical cases, increasing the reward value of the "isolate first, restore later" strategy by 35%. When a branch line failed in December 2024, the system prioritized isolating the branch line, reducing restoration time by 22 seconds compared to traditional strategies. The digital twin verification submodule maps the grid status in real time. When heavy rain caused a line failure in June 2024, it performed power flow calculations and three-dimensional visualization verification on candidate solutions. System rehearsals showed that one solution could reduce the scope of power outages by 12%, successfully avoiding power outages for 2,300 customers in actual application.

[0086] Implementation Results: By 2024, the region's self-healing success rate had increased to 91.7%, and the fault misdiagnosis rate had dropped by 67%. Manual inspections were reduced by 40%, and the equipment failure rate dropped by 32%. This effectively reduced operation and maintenance costs, improved power supply reliability, and boosted electricity satisfaction among residents and businesses.

[0087] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. A method for locating and self-healing line faults in an intelligent power distribution system, characterized in that: The method is implemented based on a multi-module intelligent power distribution system, which includes an A1 data acquisition module, an A2 fault location module, an A3 self-healing control module, and an A4 collaborative control module, as follows: A1 data acquisition module: This module collects voltage, current, zero-sequence components, and environmental parameters in real time through intelligent terminals deployed on distribution lines. It uses algorithms to eliminate noise and pre-processes data using edge computing. A2 fault location module: The master station system receives pre-processed data and locates the fault point with an error of ≤50 meters based on the newly developed positioning submodule. It also uses the deep learning submodule to classify the fault type. A3 self-healing control module: Generates the optimal isolation strategy based on a new algorithm, controls the intelligent switch to open within 30ms, and reconstructs the distribution network through this algorithm, with a recovery time of ≤30 seconds; A4 collaborative control module: realizes distributed collaborative control of the master station and terminals through a multi-agent system, and uses reinforcement learning to dynamically optimize the self-healing strategy.

2. The method for locating and self-healing line faults in an intelligent power distribution system according to claim 1, characterized in that: The A1 data acquisition module includes: A1.1 multi-sensor fusion unit, integrating Rogowski coil, fiber Bragg grating sensor and temperature and humidity transmitter; A1.2 edge computing unit, using a multi-core processor to perform local data preprocessing; The algorithm described in A1.3 specifically optimizes the accuracy of the prediction model by analyzing environmental changes in real time. Unlike traditional intelligent distribution system line fault location and self-healing methods that rely on fixed rules, it continuously compares the differences between predictions and actual observed data and dynamically adjusts the model's error range accordingly. When the difference increases, the error range is expanded to accommodate uncertainty, and when the difference decreases, the range is narrowed to improve accuracy. This process is achieved through a continuous cycle of "prediction-comparison-adjustment", ensuring that the model always adapts to actual changes and maintains stable performance in complex and dynamic environments.

3. The method for locating and self-healing line faults in an intelligent power distribution system according to claim 1, characterized in that: The A2 fault location module includes: The positioning submodule described in A2.1: When a line fault occurs, intelligent devices at both ends immediately detect the sudden change in current or voltage and perform high-precision time synchronization via a dedicated optical fiber link. The devices automatically freeze the timestamp upon signal detection, reducing data processing delays. Simultaneously, they automatically simulate faults daily to calibrate processing time to maintain accurate recorded time. The system dynamically adjusts the actual signal propagation speed based on the time difference between the fault signal's arrival at both ends, combined with a fiber propagation speed benchmark and factors including, but not limited to, conductor temperature and aging, to calculate the approximate location of the fault. The positioning results are superimposed on a power map and further verified using optical fiber detection technology, ensuring an error of no more than half a meter. If the device processing delay exceeds a set threshold, the system switches to pure optical fiber positioning mode, enabling rapid identification of the fault point even in extreme cases. The deep learning submodule described in A2.2 is implemented based on the characteristics of current and voltage signals with different characteristics generated by distribution line faults. First, the signal is scanned layer by layer through multiple layers of "filters" to capture local abnormal fluctuations, including but not limited to sudden spikes and depressions. Then, the model records the temporal variation of these fluctuations and analyzes whether the abnormal signal is continuous or short-lived. Finally, all information is integrated to quickly determine the type of fault.

4. The method for locating and self-healing line faults in an intelligent power distribution system according to claim 1, characterized in that: The A3 self-healing control module includes: A3.1 Intelligent switch control submodule: uses permanent magnet mechanism circuit breaker, opening and closing time ≤ 30ms; A3.2 Network Reconfiguration Submodule: Optimizing the power supply efficiency of a distribution network is essentially a process of finding the optimal path. The algorithm is a tool for determining this optimal path. It first randomly generates multiple possible network configurations, including the on / off states of switches and the locations of distributed generation (DG) connections. Each solution is then scored based on pre-defined objectives, including but not limited to reducing losses and improving power supply reliability. High-scoring solutions are retained, while low-scoring solutions are eliminated. The advantages of high-quality solutions are then combined and some details are randomly adjusted to generate new candidate solutions. Finally, through continuous iterative optimization, the optimal network configuration is found. A3.3 self-healing verification submodule verifies the isolation effect through the differential protection principle.

5. The method for locating and self-healing line faults in an intelligent power distribution system according to claim 1, characterized in that: The A4 collaborative control module includes: A4.1 Multi-agent system submodule: Each regional agent negotiates a power restoration plan through a contract network protocol; A4.2 Reinforcement Learning Optimization Submodule: Dynamically updates self-healing strategies based on a fault case library. Specifically, the system automatically analyzes historical cases, identifies solutions that reduce actual recovery time, identifies and records the reasons for the time reduction, and ultimately generates new strategies. Through reinforcement learning, the system increases the reward value of the recorded recovery time reduction strategies and applies them to subsequent faults. A4.3 Digital twin verification submodule: Evaluates the feasibility of the self-healing solution through real-time simulation. Specifically: The digital twin verification submodule builds a virtual model that is completely consistent with the actual power grid by replicating the operating status of the real distribution network in real time, including but not limited to equipment parameters, distributed power output and user load; the system simulates fault scenarios in a virtual environment, including but not limited to short circuits and line breaks, and conducts a preview verification of the self-healing solution generated by the master station: voltage stability, equipment load conditions and power supply recovery time are evaluated through power flow calculations; the simulation results automatically optimize the solution, and intuitively display the fault point, power flow and solution comparison through a three-dimensional visualization interface, helping operation and maintenance personnel to make quick decisions, avoid the trial and error risks of the real power grid, and significantly improve the reliability and efficiency of the self-healing solution.

Images