Power grid inspection method, system, equipment and medium

Abstract

The invention discloses a power grid inspection method, system and equipment. The method comprises the following steps: acquiring multi-dimensional data of power equipment by constructing a multi-source heterogeneous sensor array; according to the obtained multi-dimensional data of the power equipment, the to-be-inspected power monitoring points are connected in series to generate an initial inspection line; calculating and optimizing line configuration based on the electric power key data identifier and the inspection duration; and dynamically adjusting the inspection path through the reinforcement learning model, and generating an intelligent inspection decision scheme fusing the equipment health degree. According to the method, the power grid inspection efficiency is optimized through technology fusion, the detection precision is improved, and deep cooperation is formed between inspection path planning and task scheduling.

Description

Technical Field

[0001] The present invention belongs to the field of artificial intelligence technology and relates to a power grid inspection method, system, equipment and medium. Background Art

[0002] In modern power systems, power lines, as the fundamental infrastructure for energy transmission, are directly linked to power supply reliability. Since power lines are often laid in rural and mountainous areas, they are constantly exposed to environmental factors such as meteorological disasters (such as lightning strikes and icing), geological disasters (such as landslides), and vegetation growth. Therefore, regular and irregular inspections and maintenance are essential to ensure line safety.

[0003] Existing power line inspections face the following technical bottlenecks: First, power lines are distributed in a mesh pattern within the distribution area, and the same area often contains multiple lines. Traditional inspection methods find it difficult to achieve a balance between "comprehensive coverage" and "targeted inspections", resulting in inspection plans either consuming a lot of manpower and material resources due to excessive pursuit of comprehensiveness, or missing potential fault points due to selective inspections; second, path planning is inefficient. When faced with complex mesh lines, manual planning or simple heuristic algorithms cannot take into account both inspection coverage and resource consumption (such as inspection duration and equipment energy consumption), especially in the event of sudden failures or environmental changes, it is difficult to quickly generate the optimal inspection route; third, inspection resource allocation is unreasonable. Existing technologies lack a quantitative assessment system for line importance, fault risk, and environmental impact, resulting in high-risk lines not being prioritized for inspection, and low-risk lines being repeatedly inspected, resulting in a waste of resources.

[0004] How to design a solution for the meshed power line structure that can dynamically optimize inspection paths and efficiently allocate inspection resources to achieve the technical goals of "covering key lines, shortening inspection time, and reducing operation and maintenance costs" has become an urgent problem to be solved in the current power inspection field. Summary of the Invention

[0005] To solve the problems existing in the background technology, the present invention proposes a power grid inspection method, system and device, aiming to solve the problem that the existing technology lacks a quantitative assessment system for line importance, fault risk and environmental impact, resulting in high-risk lines not being inspected first and low-risk lines being inspected repeatedly, causing waste of resources.

[0006] A first aspect of the present application provides a power grid inspection method, comprising: Construct a multi-source heterogeneous sensor array to obtain multi-dimensional data of power equipment; Connect the power monitoring points to be inspected in series to generate an initial inspection circuit; Optimize line configuration based on key power data identification and inspection time calculation; Dynamically adjust inspection paths through reinforcement learning models; Generate intelligent inspection decision plans that integrate device health.

[0007] Optionally, the multi-source heterogeneous sensor array is constructed to obtain multi-dimensional data of power equipment, specifically by deploying infrared thermal imaging devices, ultrasonic partial discharge detection devices, lidar scanning devices and AI visual recognition devices, and mapping the data of each sensor to the GIS three-dimensional power grid model through the Beidou positioning system and clock synchronization module.

[0008] Optionally, the power monitoring points to be inspected are connected in series to generate an initial inspection route, specifically: multiple initial inspection routes are generated from the power monitoring points to be inspected through a random sorting algorithm, the coordinate points of each initial inspection route are determined by the geographical coordinates of the power monitoring points, and the total inspection time of a single route is calculated based on the distance between adjacent points and the equipment inspection time.

[0009] Optionally, the line configuration is optimized based on the identification of key power data and the calculation of inspection time, specifically by extracting historical equipment fault records, real-time operating parameter anomalies and equipment importance levels from the initial inspection line as key power data, prioritizing the coordinate points corresponding to the key power data in the GIS model, and reorganizing the initial line based on the identification results and the inspection time.

[0010] Optionally, the inspection path is dynamically adjusted through a reinforcement learning model, specifically by constructing a state space including the current position, remaining power, task priority, environmental parameters and equipment status, defining the action space of inspection point selection, speed adjustment, and sensor mode switching, and designing a reward function composed of inspection time, task priority, energy efficiency and fault detection rate, and realizing dynamic path optimization by training the strategy network through the PPO algorithm.

[0011] Optionally, the generation of an intelligent inspection decision-making plan that integrates the health of the equipment is specifically to establish a four-dimensional evaluation system including operating status, historical faults, environmental impact and importance, and generate a device health score through the health calculation formula healthscore=∑(weight×normalized(indicatori)), where weigh is the weight coefficient of each evaluation dimension, normalized() is to standardize the data, indicatori is the evaluation indicator after normalization of the i-th health indicator, and a priority queue for inspection and maintenance is generated based on the health score.

[0012] In a second aspect of the present application, a power grid inspection system is provided, comprising: a multi-source data acquisition terminal, an edge computing unit, a cloud-based intelligent decision-making platform, and a mobile terminal; The multi-source data acquisition terminal is used to obtain infrared thermal imaging data, ultrasonic partial discharge data, lidar point cloud data and visual image data of power equipment; The edge computing unit is used to perform real-time preprocessing and feature extraction on the collected data; The cloud-based intelligent decision-making platform is used to execute the intelligent inspection method for the power grid; The mobile terminal is used to receive inspection instructions and upload field data.

[0013] Optionally, the cloud-based intelligent decision-making platform includes a path planning module, which is used to connect the power monitoring points to be inspected in series to generate an initial inspection route, calculate the inspection time of each route, optimize the route configuration based on key power data identifiers, and generate a dynamic inspection route through a reinforcement learning model.

[0014] In a third aspect of the present application, a power inspection device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned power grid inspection method when executing the computer program.

[0015] Compared with the prior art, the present invention has the following beneficial effects: This invention provides a power grid inspection method, system, and device. These methods utilize a multi-source heterogeneous sensor array to acquire multidimensional data from power equipment. Based on this acquired multidimensional data, the power monitoring points to be inspected are connected in series to generate an initial inspection route. The route configuration is optimized based on key power data identification and inspection duration calculations. A reinforcement learning model dynamically adjusts the inspection route and generates an intelligent inspection decision-making plan that incorporates device health. Through technological integration, this invention optimizes power grid inspection efficiency, improves detection accuracy, and achieves deep synergy between inspection route planning and task scheduling. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 is a flow chart of a power grid inspection method according to an embodiment of the present invention; Figure 2 is a schematic diagram of a power grid inspection system according to an embodiment of the present invention; Figure 3 FIG. 1 is a schematic diagram of a power grid inspection device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0017] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0018] In one embodiment, if Figure 1 As shown, a power grid inspection method is provided, which is applied in Figure 1 Take the example of , and explain it, including the following specific steps: S10: Construct a multi-source heterogeneous sensor array to obtain multi-dimensional data of power equipment.

[0019] S20: Based on the acquired multi-dimensional data of the power equipment, the power monitoring points to be inspected are connected in series to generate an initial inspection line.

[0020] S30: Optimize line configuration based on key power data identification and inspection duration calculation.

[0021] S40: Dynamically adjust inspection paths through reinforcement learning models and generate intelligent inspection decision plans that integrate device health.

[0022] Specifically, a multi-source heterogeneous sensor array is constructed to acquire multidimensional data from power equipment. This involves deploying devices such as infrared thermal imaging, ultrasonic partial discharge detection, lidar, and AI-powered visual cameras. Combined with BeiDou positioning and clock synchronization technology, these sensors map temperature fields, discharge signals, 3D point clouds, and image data onto a GIS 3D model, forming a fused dataset aligned in time and space. For example, at a 220kV substation, FLIR infrared thermal imagers and UltraTEV ultrasonic partial discharge detectors are used to collect real-time equipment status data and synchronize it to the 3D model, enabling unified management of multidimensional data.

[0023] Initial inspection routes are generated by serially linking multidimensional data from power equipment. The coordinates of the inspection points to be inspected are extracted from the data, and multiple initial routes are generated using a random sorting algorithm. Inspection durations are then calculated based on the distances between adjacent points and the equipment inspection time. For example, for a 110kV transmission line with 10 towers, three initial inspection routes are randomly generated. The total inspection duration for each route is calculated based on the drone's flight speed and the inspection time per tower, providing a basis for subsequent optimization.

[0024] Based on the identification of key power data and inspection duration, line configuration is optimized, and a reinforcement learning model is used to dynamically adjust the path to generate intelligent inspection decision-making plans. This involves extracting key power data such as historical equipment failures and real-time parameter anomalies, identifying priorities in the GIS, and then reorganizing the lines. Simultaneously, a state space and reward function are constructed that include factors such as location and power. The policy network is trained using the PPO algorithm, and inspection decisions are generated in conjunction with a four-dimensional evaluation system for equipment health. For example, during transmission network inspections, a reinforcement learning model is used to dynamically adjust drone inspection paths, triggering maintenance decisions based on the health scores of transmission equipment, thereby improving inspection efficiency and reliability.

[0025] In an embodiment, in S10, a multi-source heterogeneous sensor array is constructed to obtain multi-dimensional data of power equipment, specifically by deploying an infrared thermal imaging device, an ultrasonic partial discharge detection device, a lidar scanning device and an AI visual recognition device, and mapping the sensor data to a GIS three-dimensional power grid model through the Beidou positioning system and a clock synchronization module.

[0026] Specifically, a multi-source heterogeneous sensor array is constructed to acquire multidimensional data from power equipment. This is achieved by deploying various sensor types and leveraging spatiotemporal synchronization technology to achieve comprehensive perception of equipment status and data fusion. Infrared thermal imaging devices are used to capture temperature distribution, ultrasonic partial discharge detection devices collect discharge signals, lidar scanning devices construct three-dimensional spatial models, and AI visual recognition devices capture image information. The Beidou positioning system and clock synchronization module map various data types to a GIS three-dimensional power grid model, ensuring data consistency across spatiotemporal dimensions.

[0027] Taking a 500kV transmission line as an example, in actual applications, FLIR GF77 infrared thermal imaging devices and Omicron UHF-100 ultrasonic partial discharge detection devices are deployed at key inspection points to collect hot spots at conductor joints and discharge signals from insulators. At the same time, the Unicore UB240 Beidou positioning module and Symmetricom S300 clock synchronization module provide coordinates and timestamps for the data, achieving centimeter-level positioning and nanosecond-level time synchronization.

[0028] During data processing, spatiotemporal labels are first added to the real-time data collected by each sensor. Then, through coordinate transformation and time axis alignment, infrared thermal images and ultrasonic signals are uniformly mapped to a 3D GIS model. For example, the infrared coordinates of temperature anomalies and the LiDAR point cloud coordinates are calibrated to within ±0.5 meters, and the time error between different sensor data is controlled to within 1 millisecond. Finally, the model displays these data intuitively using pseudo-color and dynamic spectrum graphs, enabling accurate perception of equipment status and fault warnings.

[0029] In an embodiment, in S20, the power monitoring points to be inspected are connected in series to generate an initial inspection route, specifically: a plurality of initial inspection routes are generated from the power monitoring points to be inspected through a random sorting algorithm, the coordinate points of each initial inspection route are determined by the geographical coordinates of the power monitoring points, and the total inspection time of a single route is calculated based on the distance between adjacent points and the time consumed in equipment inspection.

[0030] Specifically, the initial inspection routes are generated by connecting the power monitoring points to be inspected in series. This is done using a random sorting algorithm to group the inspection points into multiple candidate routes. The coordinates of each route are determined by the geographic coordinates of the power monitoring points. The total inspection duration for each route is then calculated based on the distance between adjacent points and the equipment inspection time. This random sorting ensures the diversity of the initial routes. The geographic coordinates are derived from a GIS system or Beidou positioning. Distance calculations are based on a spherical geometry model. The duration is determined by combining movement speed with the equipment's preset inspection time.

[0031] Taking the 10 towers of a 110kV transmission line as an example, in actual operation, Python's random shuffling algorithm is first used to generate line L1: P3→P7→P1→P9→P5→P2→P8→P10→P4→P6, line L2: P5→P1→P8→P3→P10→P2→P4→P7→P6→P9, line L3: P2→P6→P9→P1→P4→P8→P3→P5→P7→P10. Taking line L1 as an example, the inspection time is calculated. The spherical distance between adjacent towers is calculated using the WGS84 ellipsoid model. The distance from P3 to P7 is about 2.15 kilometers. At the drone's speed of 333 meters per minute, the travel time is about 6.45 minutes. Then, the 95 minutes of inspection time for the 10 towers is accumulated, and the total inspection time of the line is 137 minutes. Calculations revealed that the total durations of the three initial routes were 137 minutes, 142 minutes, and 135 minutes, respectively. These initial routes provided the foundation for subsequent optimization. The L3 route, with the shortest total duration, served as the initial optimal solution for further optimization within the reinforcement learning process. This not only shortened inspection time but also prioritized coverage of high-risk equipment, demonstrating the critical role of initial route generation in inspection route planning.

[0032] In an embodiment, in said S30, the line configuration is optimized based on the power key data identification and inspection duration calculation, specifically extracting the equipment historical fault records, real-time operating parameter anomalies and equipment importance levels from the initial inspection line as power key data, prioritizing the coordinate points corresponding to the power key data in the GIS model, and reorganizing the initial line configuration based on the identification results and the inspection duration.

[0033] Line configuration is optimized based on the identification of key power data and inspection time calculation. Specifically, historical equipment failure records, real-time operating parameter anomalies and equipment importance levels are extracted from the initial inspection line as key power data. Risk values ​​are generated through a quantitative model, and the coordinate points corresponding to the key power data are prioritized in the GIS three-dimensional model, such as red marking high-risk points and yellow marking medium-risk points. Combined with the inspection time data of the initial line, the line is reorganized according to the "risk priority + shortest distance" principle to achieve a balance between fault risk and inspection efficiency.

[0034] Taking a 220kV substation as an example, we extracted the real-time oil temperature anomaly of main transformer T2 (65°C, exceeding the 60°C threshold), the historical fault record of main transformer T1 (three anomalies in the past year), and the temperature anomaly of insulator INS5 (8°C, exceeding the 5°C threshold). Using the risk value formula K = 0.4 × fault frequency + 0.3 × parameter anomaly coefficient + 0.3 × importance weight, we calculated the risk value of 0.6 for T2, 0.54 for T1, and 0.24 for INS5. In the GIS model, the priorities are marked with a red heat map, a red border, and a yellow triangle, respectively.

[0035] The total duration of the initial random route L0 was 105 minutes, with high-risk points T1 and T2 inspected at the 3rd and 7th points respectively. After optimization, route L1 took T2→T1 as the core path and inserted the adjacent medium-risk point INS5. The total duration was shortened to 98 minutes, and the coverage order of high-risk points was advanced to the 1st and 4th points. This not only increased the inspection priority of high-risk points by 60%, but also achieved an intelligent response of "the higher the risk, the higher the inspection priority" through dynamic risk assessment, providing visual support for inspection decisions.

[0036] In an embodiment, in S40, the inspection path is dynamically adjusted through a reinforcement learning model, specifically by constructing a state space including the current position, remaining power, task priority, environmental parameters and equipment status, defining the action space of inspection point selection, speed adjustment and sensor mode switching, designing a reward function composed of inspection time, task priority, energy efficiency and fault detection rate, and realizing dynamic path optimization by training the strategy network through the PPO algorithm.

[0037] Specifically, a reinforcement learning model dynamically adjusts inspection paths. Specifically, the inspection problem is formulated as a Markov decision process, defining a state space encompassing the current location, remaining drone battery level, task priority, environmental parameters, and device status. An action space is then designed for inspection point selection, speed adjustment, and sensor mode switching. A reward function is constructed using inspection time, task priority, energy efficiency, and fault detection rate. The PPO algorithm is used to train a policy network to achieve dynamic path optimization. This framework enables the agent to autonomously learn the optimal inspection strategy in an uncertain environment, balancing efficiency and risk.

[0038] In the embodiment, in the S40, the generation of the intelligent inspection decision-making scheme of the integrated equipment health is specifically to establish a four-dimensional evaluation system including operating status, historical faults, environmental impact and importance, and calculate the health by the health calculation formula health score =∑(weight×normalized(indicator i )) Generates the device health score, where weigh is the weight coefficient of each evaluation dimension, normalized() is used to standardize the data, and indicatori To represent the evaluation index after normalization of the i-th health indicator, a priority queue for inspection and maintenance is generated based on the health score.

[0039] For example, taking the No. 2 main transformer at a 500kV substation as an example, operating data such as a 65°C top oil temperature and 25ppm oil chromatogram acetylene content were collected. Combined with information such as one fault record in the past year and a lightning strike density of 4.5 strikes per km² per year, the health score was calculated using weights of 0.4, 0.3, 0.2, and 0.1. After normalization, the operational status dimension scored 0.55, the historical fault dimension scored 0.4, the environmental impact dimension scored 0.875, and the importance dimension scored 1.0. The final health score was 0.615, placing it at the "warning" level.

[0040] Based on this health score, a decision-making plan was generated to include the transformer in the next month's key inspection plan, prioritize infrared thermal imaging, and recommend scheduling a major overhaul within six months. This solution detected faults 1.5 months earlier than traditional inspections, reducing maintenance costs by 20%. By integrating multidimensional data and quantitative models, the solution shifted from reactive maintenance to proactive prevention. In provincial-level power grid applications, the system achieved a 92% accuracy rate for main transformer fault warnings and a 40% improvement in inspection efficiency.

[0041] In one embodiment, if Figure 2 As shown, a power grid inspection system is provided, which corresponds one-to-one to the power grid inspection method in the above embodiment. The power grid inspection system includes: a multi-source data acquisition terminal, an edge computing unit, a cloud-based intelligent decision-making platform and a mobile terminal. The functional modules are described in detail as follows: The multi-source data acquisition terminal is used to obtain infrared thermal imaging data, ultrasonic partial discharge data, lidar point cloud data and visual image data of power equipment; The edge computing unit is used to perform real-time preprocessing and feature extraction on the collected data; The cloud-based intelligent decision-making platform is used to implement the power grid intelligent inspection method described in the rights; The mobile terminal is used to receive inspection instructions and upload field data.

[0042] Optionally, the cloud-based intelligent decision-making platform also includes: a path planning module, which is used to connect the power monitoring points to be inspected in series to generate an initial inspection route, calculate the inspection time of each route, optimize the route configuration based on key power data identifiers, and generate a dynamic inspection route through a reinforcement learning model.

[0043] The specific definition of the power grid inspection system can be found in the definition of the power grid inspection method above and will not be repeated here. Each module in the above-mentioned power grid inspection system can be implemented in whole or in part through software, hardware, or a combination thereof. Each of the above-mentioned modules can be embedded in or independent of the processor of the computer device in hardware form, or can be stored in the memory of the computer device in software form, so that the processor can call and execute the corresponding operations of each of the above modules.

[0044] In one embodiment, Figure 3 As shown, a power inspection device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, a power grid inspection method is implemented. When the processor executes the computer program, the following steps are implemented: Construct a multi-source heterogeneous sensor array to obtain multi-dimensional data of power equipment; Based on the acquired multi-dimensional data of power equipment, the power monitoring points to be inspected are connected in series to generate the initial inspection circuit; Optimize line configuration based on key power data identification and inspection time calculation; The inspection path is dynamically adjusted through the reinforcement learning model, and an intelligent inspection decision plan that integrates the health of the equipment is generated.

[0045] For the specific limitations of the power inspection device, please refer to the limitations of the power grid inspection method above, which will not be repeated here.

[0046] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM). Those skilled in the art will clearly understand that for the sake of convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0047] Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A power grid inspection method, characterized in that: The method comprises: Construct a multi-source heterogeneous sensor array to obtain multi-dimensional data of power equipment; Based on the acquired multi-dimensional data of the power equipment, the power monitoring points to be inspected are connected in series to generate the initial inspection circuit; Optimize line configuration based on key power data identification and inspection time calculation; The inspection path is dynamically adjusted through the reinforcement learning model, and an intelligent inspection decision plan that integrates the health of the equipment is generated.

2. The intelligent inspection method for power grid according to claim 1, characterized in that: The multi-source heterogeneous sensor array is constructed to obtain multi-dimensional data of power equipment, specifically by deploying infrared thermal imaging devices, ultrasonic partial discharge detection devices, lidar scanning devices and AI visual recognition devices, and mapping the sensor data to the GIS three-dimensional power grid model through the Beidou positioning system and clock synchronization module.

3. The intelligent inspection method for power grid according to claim 1, characterized in that: The method of connecting the power monitoring points to be inspected in series to generate the initial inspection route is specifically as follows: a random sorting algorithm is used to generate multiple initial inspection routes for the power monitoring points to be inspected, the coordinate points of each initial inspection route are determined by the geographical coordinates of the power monitoring points, and the total inspection time of a single route is calculated based on the distance between adjacent points and the time consumed in equipment inspection.

4. The intelligent inspection method for power grid according to claim 1, characterized in that: The line configuration optimization based on the identification of key power data and the calculation of inspection duration is specifically to extract the historical fault records of equipment, real-time operating parameter anomalies and equipment importance levels from the initial inspection line as key power data, prioritize the coordinate points corresponding to the key power data in the GIS model, and reorganize the initial line configuration based on the identification results and the inspection duration.

5. The intelligent inspection method for power grid according to claim 1, characterized in that: The inspection path is dynamically adjusted through a reinforcement learning model. Specifically, a state space is constructed that includes the current position, the remaining battery power of the drone, the task priority, the environmental parameters and the equipment status. The action space of inspection point selection, speed adjustment and sensor mode switching is defined. A reward function composed of inspection time, task priority, energy efficiency and fault detection rate is designed. The strategy network is trained by the PPO algorithm to achieve dynamic path optimization.

6. The intelligent inspection method for power grid according to claim 1, characterized in that: The intelligent inspection decision-making scheme for generating integrated equipment health is specifically to establish a four-dimensional evaluation system including operating status, historical faults, environmental impact and importance, and calculate the health through the health calculation formula health score =∑(weight×normalized(indicator i )) Generates the device health score, where weigh is the weight coefficient of each evaluation dimension, normalized() is used to standardize the data, and indicator i To represent the evaluation index after normalization of the i-th health indicator, a priority queue for inspection and maintenance is generated based on the health score.

7. A power grid intelligent inspection system, characterized in that: The system includes: a multi-source data acquisition terminal, an edge computing unit, a cloud-based intelligent decision-making platform, and a mobile terminal; The multi-source data acquisition terminal is used to obtain infrared thermal imaging data, ultrasonic partial discharge data, lidar point cloud data and visual image data of power equipment; The edge computing unit is used to perform real-time preprocessing and feature extraction on the collected data; The cloud-based intelligent decision-making platform is used to execute the power grid intelligent inspection method according to any one of claims 1 to 6; The mobile terminal is used to receive inspection instructions and upload field data.

8. The intelligent power grid inspection system according to claim 7, characterized in that: The cloud-based intelligent decision-making platform also includes a path planning module, which is used to connect the power monitoring points to be inspected in series to generate an initial inspection route, calculate the inspection time of each route, optimize the route configuration based on key power data identifiers, and generate a dynamic inspection route through a reinforcement learning model.

9. An electric power inspection device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of the power grid intelligent inspection method according to any one of claims 1 to 6 are implemented.

Images