Remote control transmission, transformation and distribution unmanned aerial vehicle inspection system

Abstract

The invention relates to the technical field of transmission, transformation and distribution equipment inspection, and discloses a remote control transmission, transformation and distribution unmanned aerial vehicle inspection system, which comprises a task planning module, a data acquisition module, an obstacle avoidance decision module, a data processing module and a fault prediction module, by integrating the task planning module, the data acquisition module, the obstacle avoidance decision-making module, the data processing module and the fault prediction module, efficient and accurate routing inspection of the transmission, transformation and distribution equipment is realized, the task planning module adopts a weighted task planning algorithm, and according to equipment distribution and routing inspection requirements, the routing inspection efficiency is improved. According to the method, the optimal flight path of the unmanned aerial vehicle is automatically planned, the inspection area and precision are set in combination with the equipment type, the inspection efficiency and accuracy are improved, the data acquisition module carries various sensors, visible light images, infrared thermal imaging, ultraviolet imaging, laser radar, sound and electromagnetic multi-mode data of equipment can be comprehensively acquired, and the inspection efficiency and accuracy are improved. And an information basis is provided for subsequent data processing and analysis.

Description

Technical Field

[0001] The present invention relates to the technical field of transmission and distribution equipment inspection, and specifically to a remote-controlled transmission and distribution equipment unmanned aerial vehicle inspection system. Background Art

[0002] With the rapid development of the power system and the continuous expansion of the power grid, the inspection work of transmission and distribution equipment has become increasingly heavy and complicated. The traditional inspection method mainly relies on manual labor, which is not only inefficient, but also has great safety hazards when facing complex terrain, bad weather or high-risk areas. In addition, manual inspection is easily affected by human factors, such as the experience and sense of responsibility of the inspection personnel, which makes it difficult to ensure the accuracy and consistency of the inspection results. Therefore, the development of an efficient, safe and accurate automated inspection technology has become an urgent problem to be solved in the field of power system operation and maintenance. In recent years, the rapid development of drone technology has provided new ideas and methods for the automated inspection of transmission and distribution equipment. The drone inspection system has gradually become a research hotspot in the field of power system inspection due to its high flexibility, strong adaptability and wide coverage.

[0003] Traditional inspection technology for transmission and distribution equipment has the following main deficiencies: First, manual inspection methods are inefficient and cannot meet the needs of rapid inspection of large-scale power grids, especially when facing inspection tasks in large areas or remote areas. Manual inspections are time-consuming and costly. Second, manual inspections pose safety risks, especially in complex terrain, severe weather or high-risk areas, where the personal safety of inspectors is difficult to guarantee. In addition, the accuracy and consistency of manual inspections are greatly affected by human factors, and different inspectors may have different judgments on the same equipment, resulting in reduced credibility of inspection results. Finally, traditional inspection methods lack real-time and predictive capabilities, making it difficult to provide early warning and timely handling of equipment failures, thereby increasing the risks of power system operation.

[0004] Therefore, developing a remote-controlled transmission and distribution drone inspection system will significantly improve the efficiency and safety of power system operation and maintenance, reduce inspection costs and risks, and provide strong guarantees for the stable operation of the power system. Summary of the Invention

[0005] The purpose of the present invention is to make up for the shortcomings of the existing technology and provide a remote-controlled transmission and distribution drone inspection system. The system integrates multiple modules of data acquisition, data fusion modeling, equipment health decision-making, power supply optimization decision-making and emergency linkage processing. By real-time monitoring and analysis of various parameters of the underground power supply system, it realizes accurate assessment of equipment health, intelligent optimization of the power supply system and rapid response to emergency situations. The application of the present invention greatly improves the stability and reliability of the underground power supply system in mines, reduces operation and maintenance costs and failure risks, and provides strong technical support for safe production and efficient operation of mineral mining.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a remote-controlled transmission and distribution UAV inspection system, the system comprising: a task planning module, a data acquisition module, an obstacle avoidance decision module, a data processing module and a fault prediction module;

[0007] The mission planning module: The operator inputs the distribution of transmission and distribution equipment and inspection requirements through the remote control terminal. The system uses a weighted mission planning algorithm to preliminarily plan the flight path of the UAV, and at the same time sets the inspection area and accuracy based on the equipment type;

[0008] The data acquisition module: The drone flies along the planned path, and the sensors on board collect device status parameters in real time to form vectors. The collected data is initially encoded and compressed, and then transmitted to the remote control terminal via wireless communication;

[0009] The obstacle avoidance decision module: During flight, the obstacle avoidance decision module integrates data from the lidar and visual sensors to perceive the location and distance of obstacles, and simultaneously inputs the information into the reinforcement learning model. The model calculates obstacle avoidance instructions based on the drone's flight status and obstacle information. After the drone executes the instructions, it gives the model a reward or penalty signal based on the obstacle avoidance results, thereby reversely optimizing the model parameters;

[0010] The data processing module: The remote control terminal transmits the received multimodal data to the data processing module, preprocesses and standardizes the collected data, extracts its features, fuses the multimodal data feature vectors using a multimodal data fusion algorithm, and inputs them into a pre-trained classification model to evaluate the device status;

[0011] Fault prediction module: Based on the information output by the data processing module and historical inspection data, the memory network model is input, the weights are calculated through the temporal attention mechanism, the weighted hidden state is obtained, and the fully connected layer is input to calculate the equipment failure probability. When the failure probability exceeds the threshold, a fault warning information is sent.

[0012] Furthermore, the mission planning module uses the weighted mission planning algorithm to preliminarily plan the flight path of the UAV, and the comprehensive influencing factor matrix of the inspection mission is set as ,in is the number of device nodes, is the influencing factor dimension, Indicates the The device node is controlled by The effect value of the influencing factors, the introduction of intelligent preference vector , which is used to reflect the relative importance of different influencing factors in path planning. The objective function of path planning is: ,in, To plan the route, is a node To Node The spatial distance, is the dimension index.

[0013] Furthermore, the inspection area and accuracy are determined in the task planning module:

[0014] Inspection area determination:

[0015] For large equipment, a circular or rectangular area with a large radius centered on the equipment is designated as the inspection area;

[0016] For transmission lines, according to the line direction, with the line as the axis, a strip inspection area is extended to both sides. The extension distance is determined according to the line voltage level and the complexity of the surrounding environment.

[0017] For small devices, if they are densely distributed, the entire area is used as the inspection area; if they are dispersed, the areas where adjacent devices are located are merged and set as the inspection area based on the actual layout of the equipment;

[0018] Inspection accuracy determination:

[0019] For equipment with complex structures and many key components, a higher inspection accuracy is set, requiring the drone to maintain a lower flight altitude and slower flight speed during inspections, and the resolution of the captured images must be higher;

[0020] For equipment with relatively simple appearance and single function, the inspection accuracy is reduced, the drone is allowed to fly at a relatively high altitude and speed, and the image resolution requirement is also reduced accordingly;

[0021] For equipment that requires key monitoring of its internal status, regular precision inspections are conducted on the equipment's appearance, and sensors are used for high-precision temperature detection.

[0022] Furthermore, the sensors carried by the drone in the data acquisition module are: visible light camera, infrared thermal imager, ultraviolet imager, lidar, sound sensor, electromagnetic sensor and gas sensor.

[0023] Furthermore, the device status parameters collected by the data collection module are:

[0024] Visible light image-related parameters: device appearance image, including information on the shape, color, texture, and surface integrity of the device as a whole and its components;

[0025] Infrared thermal imaging related parameters: equipment surface temperature distribution data, equipment overall temperature field, hot spot location and temperature value and temperature difference of each component;

[0026] UV imaging related parameters: corona discharge related parameters, including corona discharge location, discharge intensity and discharge morphology;

[0027] LiDAR related parameters: 3D spatial parameters of the device and its surrounding environment, including the device's dimensions, spatial position coordinates, distance to surrounding obstacles, and terrain information;

[0028] Sound sensor related parameters: audio signal of the device operating sound, including characteristic parameters of the sound frequency, intensity, and waveform;

[0029] Electromagnetic sensor related parameters: parameters of the intensity, frequency component, and phase of the electromagnetic signals around the device.

[0030] Furthermore, the selection of the obstacle avoidance instruction in the data processing module defines the state space vector of the drone as and the action space vector is , according to the value function To evaluate the pros and cons of different actions a, the value function calculation formula is: , where the value function To evaluate the status Next action The degree of quality, is the parameter set of the reinforcement learning model, including the neural network weights and bias terms, It is the state space vector of the UAV, which contains the flight status information of the UAV’s position, speed, and attitude. is the action space vector, representing the obstacle avoidance action that the drone can perform, yes The reward value at that moment, Is the discount factor, ranging from [0, 1], used to measure the importance of future rewards. is the state at the next moment, measured in real time by the sensor, Compare different actions for possible actions at the next moment Calculated value function value , choose so that The action with the largest value As the optimal obstacle avoidance instruction.

[0031] Furthermore, the data processing module adopts a multimodal data fusion algorithm to fuse the multimodal data feature vector, assuming that the visible light image data feature vector is , the infrared thermal imaging data feature vector is , the UV imaging data feature vector is , the fused feature vector The calculation formula is: ,in is the adjacency matrix, Represents the original adjacency relationship, describing the connection relationship between multimodal data features, is the identity matrix, represents the dimension of the matrix, is the degree matrix, whose diagonal elements , represents the adjacency matrix Middle The connectivity of the nodes, It is a feature matrix, which contains the feature information extracted from each multimodal data. is the learnable weight matrix, is the activation function.

[0032] Furthermore, the model of the device status in the data processing module is constructed using the following calculation formula: ,in is the fused feature vector, which serves as the input data for evaluating the device status. It is a weight vector with the same dimension as the feature vector, which is used to adjust the influence of different features on the evaluation results. is a bias term, and the judgment rule is: when When the device is in normal state; when When the device is in abnormal state; , the device is judged to be in a fault state.

[0033] Furthermore, the construction of the memory network model in the fault prediction module assumes that the input time series data is , the hidden state is , represents the total number of time steps of the time series data, and the attention weight is , which is used to measure the importance of input data and hidden state in calculating weighted hidden state at time t. The calculation formula is: ,in The learnable parameter vector, is a learnable weight matrix used to adjust the hidden state Perform a linear transformation, is a learnable weight matrix for input Perform a linear transformation, represents the input data at time t, is the hidden state at time t, It is The hidden state value of time steps, It is The input value of the time step, is the sum index, the weighted hidden state The calculation formula is: , the calculation formula of the fault prediction probability output by the fully connected layer is: , Used to map the weighted hidden state to the fault prediction probability space, It is used to map the weighted hidden state to the fault prediction probability space.

[0034] Furthermore, the fault probability threshold in the fault prediction module Determination, if the equipment has Key status parameters , first calculate the value range of each parameter under normal conditions , , It is The mean value of the parameters under normal conditions, Is a coefficient, set according to actual conditions, For the The confidence coefficient of each parameter is used to adjust the standard deviation multiple of the normal range. No. The standard deviation of a parameter under normal conditions reflects the degree of parameter fluctuation. When the equipment status parameter deviates from the normal range, a risk score is assigned according to the degree of deviation. , if the parameter The degree beyond the normal range is , which can be set , the fault probability threshold is obtained by weighted summation , the formula is: , The weight of each parameter is determined according to the importance of the parameter to the equipment failure. Greater than the calculated failure probability threshold When the alarm is triggered immediately.

[0035] Compared with the existing technology, this remote-controlled transmission and distribution drone inspection system has the following beneficial effects:

[0036] 1. The present invention realizes efficient and accurate inspection of transmission and distribution equipment by integrating a task planning module, a data acquisition module, an obstacle avoidance decision module, a data processing module and a fault prediction module. Among them, the task planning module adopts a weighted task planning algorithm, which can automatically plan the optimal drone flight path according to the equipment distribution and inspection requirements, and set the inspection area and accuracy based on the equipment type, thereby improving the inspection efficiency and accuracy. The data acquisition module is equipped with a variety of sensors, which can comprehensively collect the equipment's visible light images, infrared thermal imaging, ultraviolet imaging, lidar, sound and electromagnetic multimodal data, providing an information basis for subsequent data processing and analysis, which not only improves the comprehensiveness and accuracy of the inspection, but also enables the system to adapt to various complex environments and equipment types.

[0037] 2. The present invention integrates the data of lidar and visual sensors through the obstacle avoidance decision module to calculate obstacle avoidance instructions in real time, so that the UAV can autonomously avoid obstacles in complex environments, improving the safety and reliability of inspections. The fault prediction module is based on the memory network model and combines the temporal attention mechanism to monitor the equipment status in real time and predict faults. When the fault probability exceeds the preset threshold, the system can automatically send fault warning information, realizing early detection and timely processing of equipment faults, which not only improves the operation and maintenance efficiency of the power system, but also reduces the impact of faults on the power system, providing strong support for the safe and stable operation of the power system.

[0038] Other advantages, objects and features of the present invention will be described in part in the following description and, in part, will be apparent to those skilled in the art based on an examination of the following or may be learned from the practice of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can also derive other drawings based on these drawings without inventive effort.

[0040] Figure 1 This is a flow chart of a remote-controlled transmission and distribution drone inspection system.

[0041] Figure 2 This is a module architecture association diagram of a remote-controlled transmission and distribution drone inspection system. DETAILED DESCRIPTION

[0042] In order to further illustrate the technical means and effects adopted by the present invention to achieve the predetermined purpose of the invention, the specific implementation methods, structures, features and effects of the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments.

[0043] Example 1: Inspection of power transmission lines in mountainous areas.

[0044] There is a complex transmission line in a certain mountainous area. Due to the rugged terrain and dense vegetation, the inspection of the transmission line poses a great challenge. Relevant operators use remote control terminals to accurately input the distribution details and specific inspection requirements of the transmission line into the task planning module of the remote-controlled transmission and distribution drone inspection system.

[0045] The mission planning module starts to operate and uses the weighted mission planning algorithm to make preliminary plans for the flight path of the UAV. Here, the comprehensive influencing factor matrix of the inspection mission is defined as ,in is the number of device nodes, is the influencing factor dimension, Indicates the The device node is controlled by The effect value of the influencing factors, while introducing the intelligent preference vector ,Through the objective function of path planning, we fully consider various influencing factors in mountainous environments, such as terrain and vegetation distribution, and plan a reasonable flight path for the UAV. The objective function of path planning is: ,in, To plan the route, is a node To Node The spatial distance, is the dimension index, such as Figure 1 As shown in the figure, when determining the inspection area, the strip inspection area is demarcated according to the direction of the transmission line, with the line as the axis and a certain distance extended to both sides. In view of the complex mountain environment and the high voltage level of the line, the extension distance is correspondingly increased. For the inspection accuracy, a higher inspection accuracy is set for the tower part with complex structure on the transmission line, requiring the drone to maintain a lower flight altitude and a slower flight speed during the inspection, and the resolution of the captured image must meet a higher standard; for the relatively simple part of the line, the drone is allowed to fly at a relatively high altitude and speed, and the image resolution requirement is correspondingly reduced.

[0046] The data acquisition module is activated, and the drone, equipped with visible light cameras, infrared thermal imagers, ultraviolet imagers, lidar, sound sensors, and electromagnetic sensors, flies strictly according to the planned path. During flight, each sensor performs its own function. The visible light camera is responsible for collecting images of the device's appearance, obtaining visible light image-related parameters including the shape, color, texture, and surface integrity of the device as a whole and its components. The infrared thermal imager collects surface temperature distribution data, such as the overall device temperature field, hot spot locations and temperature values, and parameters of temperature differences between components. The ultraviolet imager collects corona discharge-related parameters, including the location, intensity, and morphology of corona discharge. The lidar collects three-dimensional spatial parameters of the device and its surrounding environment, including the device's external dimensions, spatial position coordinates, distance from surrounding obstacles, and information on mountainous terrain. The sound sensor collects audio signals from the device's operating sound, obtaining parameters including the sound frequency, intensity, and waveform. The electromagnetic sensor collects parameters of the intensity, frequency content, and phase of the electromagnetic signals around the device. The collected data is immediately encoded and compressed, and then quickly transmitted to the remote control terminal via wireless communication.

[0047] The obstacle avoidance decision module plays a key role. During the flight of the drone, it integrates the data of the lidar and visual sensors to accurately perceive the location and distance of obstacles, and simultaneously inputs this information into the reinforcement learning model to define the state space vector of the drone as and the action space vector is , the obstacle avoidance instruction is calculated based on the value function, and the value function calculation formula is: , where the value function To evaluate the status Next action The degree of quality, are model parameters, It is the state space vector of the UAV, which contains the flight status information of the UAV’s position, speed, and attitude. is the action space vector, representing the obstacle avoidance action that the drone can perform, yes The reward value at that moment, Is the discount factor, ranging from [0, 1], used to measure the importance of future rewards. is the state at the next moment, measured in real time by the sensor, Compare different actions for possible actions at the next moment Calculated value function value , choose so that The action with the largest value As the optimal obstacle avoidance instruction, after the drone executes the instruction, it gives the model a reward or penalty signal based on the obstacle avoidance result, and then reversely optimizes the model parameters to ensure the safe flight of the drone in complex mountainous environments.

[0048] After receiving the multimodal data from the remote control terminal, the data processing module begins to process the data in depth. The feature vector of the visible light image data is , the infrared thermal imaging data feature vector is , the UV imaging data feature vector is , the multimodal data fusion algorithm is used to fuse the multimodal data feature vectors, and the fused feature vectors The calculation formula is: ,in is the adjacency matrix, Represents the original adjacency relationship, describing the connection relationship between multimodal data features, Is the identity matrix, used to ensure the stability of the adjacency matrix and the accuracy of the calculation, represents the dimension of the matrix, Is a degree matrix, whose diagonal elements are the degrees of the corresponding nodes in the adjacency matrix, which is used to normalize the adjacency matrix. It is a feature matrix, which contains the feature information extracted from each multimodal data. is the learnable weight matrix, is an activation function, which is input into the pre-trained classification model to evaluate the status of the transmission line equipment. The formula is: ,in is the fused feature vector, which serves as the input data for evaluating the device status. It is a weight vector with the same dimension as the feature vector, which is used to adjust the influence of different features on the evaluation results. is the bias term, when When the device is in normal state; when When the device is in abnormal state; , the device is judged to be in a fault state.

[0049] The fault prediction module is based on the output information of the data processing module and a large amount of historical inspection data. These data are input into the memory network model. Assume that the input time series data is , the hidden state is , the attention weight is calculated through the temporal attention mechanism, and the calculation formula is: ,in The learnable parameter vector, is a learnable weight matrix used to adjust the hidden state Perform a linear transformation, is a learnable weight matrix for input Perform a linear transformation, represents the input data at time t, is the hidden state at time t, It is The hidden state value of time steps, It is The input value of the time step, is the sum index, the weighted hidden state The calculation formula is: , output the fault prediction probability through the fully connected layer: , Used to map the weighted hidden state to the fault prediction probability space, It is used to map the weighted hidden state to the fault prediction probability space. Assuming that the transmission line equipment has n key state parameters, the value range of each parameter under normal conditions is calculated first. When the equipment state parameter deviates from the normal range, a risk score is assigned according to the degree of deviation. The fault probability threshold is obtained by weighted summation. , failure probability threshold The calculation formula is: , is the weight of each parameter, when the calculated equipment failure probability Greater than the calculated failure probability threshold When a fault occurs, a fault warning message is immediately sent so that the operation and maintenance personnel can take timely measures to ensure the stable operation of the transmission line.

[0050] In summary, the drone inspection system effectively solves the complex scenarios of power transmission line inspections in mountainous areas. The mission planning module fully considers the mountainous environmental factors to plan routes and areas. The data acquisition module collects multiple types of data during flight. The obstacle avoidance decision-making module ensures the safe flight of the drone. The data processing module integrates data to evaluate the line status. The fault prediction module warns of faults in advance. With the help of various formulas in the system, the entire process of intelligent inspection of transmission lines from data collection to fault prediction is realized, which improves the inspection efficiency and accuracy, reduces the operation and maintenance costs of transmission lines in mountainous areas, and ensures the reliability of power transmission.

[0051] Example 2: Inspection of large substations.

[0052] In a large substation with numerous internal devices and a complex layout, operators use remote control terminals to input the distribution information of the substation's transmission and distribution equipment and inspection requirements into the task planning module of the remotely controlled transmission and distribution drone inspection system.

[0053] In the mission planning module, the system uses the weighted mission planning algorithm to preliminarily plan the flight path of the UAV and defines the comprehensive influencing factor matrix of the inspection mission as , while introducing intelligent preference vector , the flight path of the UAV in the substation is planned through the objective function of path planning. The objective function of path planning is: At the same time, due to the complex structure of large equipment in the substation, a circular area with a larger radius is delineated with the equipment as the center as the inspection area; and a higher inspection accuracy is set, requiring the drone to maintain a lower flight altitude and a slower flight speed during inspection, and the resolution of the captured image must meet higher standards.

[0054] In the data acquisition module, the drone is equipped with a visible light camera, an infrared thermal imager, an ultraviolet imager, a lidar, an acoustic sensor, and an electromagnetic sensor, and flies according to a planned path. During the flight, the visible light camera collects images of the device's appearance, including information on the shape, color, texture, and surface integrity of the device as a whole and its components; the infrared thermal imager collects data on the device's surface temperature distribution, such as the device's overall temperature field, hotspot locations and temperature values, and temperature differences between components; the ultraviolet imager collects parameters related to corona discharge, including the location, intensity, and form of corona discharge; the lidar collects three-dimensional spatial parameters of the device and its surrounding environment, such as the device's dimensions, spatial position coordinates, distance from surrounding obstacles, and terrain information; the acoustic sensor collects audio signals of the device's operating sound, including characteristic parameters of the sound's frequency, intensity, and waveform; the electromagnetic sensor collects parameters of the intensity, frequency component, and phase of electromagnetic signals around the device. The collected data is preliminarily encoded and compressed, and transmitted to the remote control terminal via wireless communication, such as Figure 2 shown.

[0055] In the obstacle avoidance decision module, the flying drone uses the data from the lidar and visual sensors to perceive the position and distance of obstacles, and simultaneously inputs the information into the reinforcement learning model to define the state space vector of the drone as and the action space vector is , according to the value function To evaluate the pros and cons of different actions a, the value function calculation formula is: , compare different actions Calculated value function value , choose so that The action with the largest value As the optimal obstacle avoidance instruction, after the drone executes the instruction, it gives the model a reward or penalty signal based on the obstacle avoidance result, and then reversely optimizes the model parameters.

[0056] In the data processing module, the remote control terminal transmits the received multimodal data to this module, preprocesses and standardizes the collected data, and extracts its features. The multimodal data fusion algorithm is used to fuse the multimodal data feature vector. Suppose the visible light image data feature vector is , the infrared thermal imaging data feature vector is , the UV imaging data feature vector is , the multimodal data fusion algorithm is used to fuse the multimodal data feature vectors, and the fused feature vectors The calculation formula is: , then preprocess and standardize the fused data, extract its features, and input it into the pre-trained classification model to evaluate the equipment status. The formula is: ,when When the device is in normal state; when When the device is in abnormal state; , the device is judged to be in a fault state.

[0057] In the fault prediction module, based on the output information of the data processing module and a large amount of historical inspection data, the memory network model is input. The input time series data is set as , calculate the attention weight through the temporal attention mechanism , get the weighted hidden state , input the fully connected layer and calculate the probability of equipment failure , the equipment is Key state parameters are first calculated. The value range of each parameter under normal conditions is calculated. When the equipment state parameter deviates from the normal range, a risk score is assigned according to the degree of deviation. The fault probability threshold is obtained by weighted summation. , when the failure prediction probability Greater than the failure probability threshold When the fault occurs, a fault warning message is sent.

[0058] To sum up, in the large-scale substation inspection scenario, the remote-controlled transmission and distribution drone inspection system demonstrates powerful functions. The task planning module can plan a reasonable path based on equipment distribution and demand, determine the inspection area and accuracy, the data acquisition module comprehensively collects equipment status parameters through a variety of sensors, the obstacle avoidance decision module ensures flight safety, the data processing module integrates and analyzes data to evaluate equipment status, and the fault prediction module predicts faults based on historical data and current information. The modules work closely together and use formulas for precise calculations to achieve efficient monitoring of the equipment status of large substations, timely discover potential problems, and provide strong support for the stable operation of substations.

[0059] The above description is merely a preferred embodiment of the present invention and does not constitute any form of limitation to the present invention. Although the present invention has been disclosed as above in terms of a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can, without departing from the scope of the technical solution of the present invention, make some changes or modifications to equivalent embodiments using the technical contents disclosed above. However, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A remote-controlled transmission and distribution drone inspection system, characterized in that: The system includes: mission planning module, data acquisition module, obstacle avoidance decision module, data processing module and fault prediction module; The mission planning module: The operator inputs the distribution of transmission and distribution equipment and inspection requirements through the remote control terminal. The system uses a weighted mission planning algorithm to preliminarily plan the flight path of the UAV, and at the same time sets the inspection area and accuracy based on the equipment type; The data acquisition module: The drone flies along the planned path, and the sensors on board collect device status parameters in real time to form vectors. The collected data is initially encoded and compressed, and then transmitted to the remote control terminal via wireless communication; The obstacle avoidance decision module: During flight, the obstacle avoidance decision module integrates data from the lidar and visual sensors to perceive the location and distance of obstacles, and simultaneously inputs the information into the reinforcement learning model. The model calculates obstacle avoidance instructions based on the drone's flight status and obstacle information. After the drone executes the instructions, it gives the model a reward or penalty signal based on the obstacle avoidance results, thereby reversely optimizing the model parameters; The data processing module: The remote control terminal transmits the received multimodal data to the data processing module, preprocesses and standardizes the collected data, extracts its features, fuses the multimodal data feature vectors using a multimodal data fusion algorithm, and inputs them into a pre-trained classification model to evaluate the device status; Fault prediction module: Based on the information output by the data processing module and historical inspection data, the memory network model is input, the weights are calculated through the temporal attention mechanism, the weighted hidden state is obtained, and the fully connected layer is input to calculate the equipment failure probability. When the failure probability exceeds the threshold, a fault warning information is sent.

2. A remote-controlled transmission and distribution drone inspection system according to claim 1, characterized in that: The mission planning module uses the weighted mission planning algorithm to preliminarily plan the flight path of the UAV. The comprehensive influencing factor matrix of the inspection mission is set as ,in is the number of device nodes, is the influencing factor dimension, Indicates the The device node is controlled by The effect value of the influencing factors, the introduction of intelligent preference vector , which is used to reflect the relative importance of different influencing factors in path planning. The objective function of path planning is: ,in, To plan the route, is a node To Node The spatial distance, is the dimension index.

3. A remote-controlled transmission and distribution drone inspection system according to claim 1, characterized in that: Determination of inspection area and accuracy in the task planning module: Inspection area determination: For large equipment, a circular or rectangular area with a large radius centered on the equipment is designated as the inspection area; For transmission lines, according to the line direction, with the line as the axis, a strip inspection area is extended to both sides. The extension distance is determined according to the line voltage level and the complexity of the surrounding environment. For small equipment, if they are densely distributed, the entire area is used as the inspection area; If the equipment is dispersed, the areas where adjacent equipment are located will be merged and set as inspection areas based on the actual equipment layout; Inspection accuracy determination: For equipment with complex structures and many key components, a higher inspection accuracy is set, requiring the drone to maintain a lower flight altitude and slower flight speed during inspections, and the resolution of the captured images must be higher; For equipment with relatively simple appearance and single function, the inspection accuracy is reduced, the drone is allowed to fly at a relatively high altitude and speed, and the image resolution requirement is also reduced accordingly; For equipment that requires key monitoring of its internal status, regular precision inspections are conducted on the equipment's appearance, and sensors are used for high-precision temperature detection.

4. A remote-controlled transmission and distribution drone inspection system according to claim 1, characterized in that: The sensors carried by the drone in the data acquisition module are: visible light camera, infrared thermal imager, ultraviolet imager, lidar, sound sensor, electromagnetic sensor and gas sensor.

5. The remote-controlled transmission and distribution drone inspection system according to claim 1 is characterized in that: The device status parameters collected in the data acquisition module are: Visible light image-related parameters: device appearance image, including information on the shape, color, texture, and surface integrity of the device as a whole and its components; Infrared thermal imaging related parameters: equipment surface temperature distribution data, equipment overall temperature field, hot spot location and temperature value and temperature difference of each component; UV imaging related parameters: corona discharge related parameters, including corona discharge location, discharge intensity and discharge morphology; LiDAR related parameters: 3D spatial parameters of the device and its surrounding environment, including the device's dimensions, spatial position coordinates, distance to surrounding obstacles, and terrain information; Sound sensor related parameters: audio signal of the device operating sound, including characteristic parameters of the sound frequency, intensity, and waveform; Electromagnetic sensor related parameters: parameters of the intensity, frequency component, and phase of the electromagnetic signals around the device.

6. A remote-controlled transmission and distribution drone inspection system according to claim 1, characterized in that: The selection of obstacle avoidance instructions in the data processing module defines the state space vector of the drone as and the action space vector is , according to the value function To evaluate the pros and cons of different actions a, the value function calculation formula is: , where the value function To evaluate the status Next action The degree of quality, is the parameter set of the reinforcement learning model, including the neural network weights and bias terms, It is the state space vector of the UAV, which contains the flight status information of the UAV’s position, speed, and attitude. is the action space vector, representing the obstacle avoidance action that the drone can perform, yes The reward value at that moment, is a discount factor, ranging from [0, 1], used to measure the importance of future rewards. is the state at the next moment, measured in real time by the sensor, Compare different actions for possible actions at the next moment Calculated value function value , choose so that The action with the largest value As the optimal obstacle avoidance instruction.

7. The remote-controlled transmission and distribution drone inspection system according to claim 1 is characterized in that: The data processing module adopts a multimodal data fusion algorithm to fuse the multimodal data feature vectors. Let the visible light image data feature vector be , the infrared thermal imaging data feature vector is , the UV imaging data feature vector is , the fused feature vector The calculation formula is: ,in is the adjacency matrix, Represents the original adjacency relationship, describing the connection relationship between multimodal data features, is the identity matrix, represents the dimension of the matrix, is the degree matrix, whose diagonal elements , represents the adjacency matrix Middle The connectivity of the nodes, It is a feature matrix, which contains the feature information extracted from each multimodal data. is the learnable weight matrix, is the activation function.

8. The remote-controlled transmission and distribution drone inspection system according to claim 1 is characterized in that: The model construction of the device status in the data processing module is calculated as follows: ,in is the fused feature vector, which serves as the input data for evaluating the device status. It is a weight vector with the same dimension as the feature vector, which is used to adjust the influence of different features on the evaluation results. is a bias term, and the judgment rule is: when When the device is in normal state; when When the device is in abnormal state; , the device is judged to be in a fault state.

9. The remote-controlled transmission and distribution drone inspection system according to claim 1 is characterized in that: The construction of the memory network model in the fault prediction module assumes that the input time series data is , the hidden state is , represents the total number of time steps of the time series data, and the attention weight is , which is used to measure the importance of input data and hidden state in calculating weighted hidden state at time t. The calculation formula is: ,in The learnable parameter vector, is a learnable weight matrix used to adjust the hidden state Perform a linear transformation, is a learnable weight matrix for input Perform a linear transformation, represents the input data at time t, is the hidden state at time t, It is The hidden state value of time steps, It is The input value of the time step, is the sum index, the weighted hidden state The calculation formula is: , the calculation formula of the fault prediction probability output by the fully connected layer is: , Used to map the weighted hidden state to the fault prediction probability space, It is used to map the weighted hidden state to the fault prediction probability space.

10. A remote-controlled transmission and distribution drone inspection system according to claim 9, characterized in that: The fault probability threshold in the fault prediction module Determination, if the equipment has Key status parameters , first calculate the value range of each parameter under normal conditions , , It is The mean value of the parameters under normal conditions, Is a coefficient, set according to actual conditions, For the The confidence coefficient of each parameter is used to adjust the standard deviation multiple of the normal range. No. The standard deviation of a parameter under normal conditions reflects the degree of parameter fluctuation. When the equipment status parameter deviates from the normal range, a risk score is assigned according to the degree of deviation. , if the parameter The degree beyond the normal range is , which can be set , the fault probability threshold is obtained by weighted summation , the formula is: , The weight of each parameter is determined according to the importance of the parameter to the equipment failure. Greater than the calculated failure probability threshold When the alarm is triggered immediately.

Images