Inspection service management system

Abstract

An inspection service management system, comprising an edge management device, a remote monitoring platform, a wireless communication base station and an edge computing device; the remote monitoring platform and the edge computing device are separately and electrically connected to the edge management device by means of the wireless communication base station, and the edge computing device is mounted within an unmanned aerial vehicle; when an inspection enabling instruction sent by the remote monitoring platform is received, the edge management device generates a path planning result of a target area on the basis of the inspection enabling instruction; the edge computing device controls an inspection operation of the unmanned aerial vehicle on the basis of the path planning result, so as to generate initial inspection results of the target area; the edge management device also performs consistency verification on the initial inspection results so as to obtain a final inspection result; and the remote monitoring platform performs fault processing on the target area on the basis of the final inspection result. In the present application, the inspection service management system implements synchronous automated fault identification and verification of image data returned by the unmanned aerial vehicle during a power inspection process, thereby improving the accuracy of device fault identification and the timeliness of fault processing.

Description

An inspection service management system

[0001] The present application claims priority to the Chinese patent application No. 202410901399.4, filed on July 5, 2024, and entitled "An inspection service management system", the content of which is incorporated herein by reference in its entirety. TECHNICAL FIELD

[0002] The present application relates to the technical field of power inspection, in particular to an inspection service management system. BACKGROUND

[0003] In the current era of rapid development of unmanned aerial vehicle technology, the power system is facing unprecedented opportunities for technological innovation. Among them, due to the need for a large number of manual operations in traditional transmission line inspection work, it leads to low efficiency, and there are safety hazards and high operating costs. In order to solve these challenges, the power system has begun to actively explore and introduce unmanned aerial vehicle technology for transmission line inspection and detection.

[0004] The unmanned aerial vehicle can carry high-resolution visible light cameras, high-precision infrared thermal imagers, and three-dimensional laser radar scanning equipment and other advanced detection equipment, greatly enriching the transmission line inspection means, not only effectively improving the work efficiency of power inspection, but also significantly reducing the operating cost and pressure of transmission line. However, the current unmanned aerial vehicle inspection method uses a data backhaul processing method, which requires maintenance personnel to manually process after the unmanned aerial vehicle data is backhauled, resulting in poor accuracy of equipment fault identification and timeliness of fault processing.

[0005] SUMMARY

[0006] The present application aims to at least solve one of the above technical defects, in particular the technical defect that in the prior art, maintenance personnel need to manually process after the unmanned aerial vehicle data is backhauled, resulting in poor accuracy of equipment fault identification and timeliness of fault processing.

[0007] The present application provides an inspection service management system, the system comprising an edge management device, a remote monitoring platform, a wireless communication base station and an edge computing device;

[0008] The remote monitoring platform and the edge computing device are respectively electrically connected with the edge management device through the wireless communication base station; the edge computing device is mounted in an unmanned aerial vehicle;

[0009] The edge management device is configured to generate a path planning result of a target area based on an inspection start instruction received from the remote monitoring platform;

[0010] The edge computing device is configured to control the inspection operation of the UAV based on the path planning result to generate an initial inspection result of the target area.

[0011] The edge management device is further configured to perform consistency verification on the initial inspection result to obtain a final inspection result.

[0012] The remote monitoring platform is configured to perform fault processing on the target area according to the final inspection result.

[0013] Optionally, the inspection business management system further comprises a camera.

[0014] The camera is mounted on the UAV and electrically connected to the edge computing device, and is configured to perform image shooting on the power equipment in the target area.

[0015] Optionally, the edge management device comprises an instruction analysis module, a path planning module and a result generation module.

[0016] The instruction analysis module is configured to analyze the inspection start instruction to obtain a target area and an inspection type corresponding to the inspection start instruction.

[0017] The path planning module is configured to perform path planning on the target area based on the inspection type to generate an inspection path of at least one UAV and an inspection target corresponding to the inspection path, wherein the inspection target comprises at least one inspection type.

[0018] The result generation module is configured to determine a lightweight fault identification model corresponding to each inspection type, and generate a path planning result corresponding to each UAV according to the inspection path, the inspection target and the lightweight fault identification model of each UAV.

[0019] Optionally, the process in which the path planning module performs path planning on the target area based on the inspection type comprises:

[0020] The path planning module divides the target area into a plurality of sub-areas, determines device data in each sub-area, and performs multi-objective optimization on a preset number of UAVs based on each device data, with the shortest inspection path and the most power equipment corresponding to the inspection target as the target.

[0021] Optionally, the path planning result comprises an inspection path, an inspection target and at least one lightweight fault identification model; and the edge computing device comprises an instruction control module, a type identification module and a fault identification module.

[0022] The instruction control module is configured to control the UAV to fly according to the inspection path after receiving the path planning result sent by the edge management device, and control the camera to take pictures of the power equipment when the UAV flies to the power equipment corresponding to the inspection target, so as to obtain the equipment image.

[0023] The type identification module is configured to identify the type of the equipment image, so as to obtain the inspection type corresponding to the equipment image.

[0024] The fault identification module is configured to determine a light fault identification model corresponding to the inspection type, and identify the fault of the equipment image by using the light fault identification model, so as to obtain the fault identification result of the power equipment.

[0025] The instruction control module is further configured to generate a flight instruction corresponding to the fault identification result, and control the flight path of the UAV by using the flight instruction.

[0026] Optionally, the instruction control module is further configured to generate a flight instruction corresponding to the fault identification result, and control the flight path of the UAV by using the flight instruction, and the process includes:

[0027] When the fault identification result is normal, the instruction control module generates a continue flight instruction, and controls the UAV to continue flying according to the inspection path by using the continue flight instruction.

[0028] When the fault identification result is abnormal, the instruction control module generates a return flight instruction, and controls the UAV to fly to the position of the power equipment by using the return flight instruction, and generates an initial inspection result corresponding to the power equipment, and sends the initial inspection result to the edge management system.

[0029] Optionally, the instruction control module generates an initial inspection result corresponding to the power equipment, and the process includes:

[0030] After the instruction control module determines the fault position of the power equipment according to the fault identification result, the instruction control module controls the UAV to hover at the fault position, controls the camera to take pictures of the power equipment according to a preset shooting strategy, and obtains a target image, and obtains the inspection type, the fault identification result and the target image of the power equipment, so as to form an initial inspection result.

[0031] Optionally, the initial inspection result includes the inspection type, the fault identification result and the target image; and the edge management device further includes an image identification module and a result comparison module.

[0032] The image recognition module model is configured to determine a target fault recognition model corresponding to the inspection type after receiving the initial inspection result sent by the edge computing module, and perform fault recognition on the fault recognition result by using the target fault recognition model to obtain a target recognition result; wherein the target fault recognition model is a light fault recognition model before being lightened.

[0033] The result comparison module is configured to compare the target recognition result with the fault recognition result, and generate a final inspection result according to a comparison result, and send the final inspection result to the remote monitoring platform.

[0034] Optionally, the result comparison module is configured to generate the final inspection result according to the comparison result, and the process includes:

[0035] When the comparison result is consistent, the result comparison module obtains a fault position corresponding to the target image, and generates a final inspection result according to the fault position and the target recognition result;

[0036] When the comparison result is inconsistent, the result comparison module obtains the initial inspection result and the target recognition result to form a final inspection result.

[0037] Optionally, the remote monitoring platform is configured to perform fault processing on the target area according to the final inspection result, and the process includes:

[0038] The remote monitoring platform determines a fault type corresponding to the final inspection result, and when the fault type is a specified type, issues a fault processing instruction to an intelligent operation and maintenance device corresponding to the fault type, so that the intelligent operation and maintenance device goes to the target area to perform fault processing; and when the fault type is a non-specified type, pushes the final inspection result to an operation and maintenance personnel of the target area.

[0039] From the above technical solutions, it can be seen that the embodiments of the present application have the following advantages:

[0040] The application provides a kind of inspection service management system, the system includes edge management device, remote monitoring platform, wireless communication base station and edge computing device, and remote monitoring platform and edge computing device are electrically connected with edge management device respectively by wireless communication base station, and edge computing device is hung in unmanned aerial vehicle.Edge management device can generate the path planning result of target area based on the inspection start instruction sent by remote monitoring platform when receiving the inspection start instruction, so that edge computing device can control the inspection operation of unmanned aerial vehicle based on the path planning result, to generate the initial inspection result of target area, by completing the operation of fault identification on unmanned aerial vehicle, the synchronism of power inspection and fault identification can be guaranteed, and the timeliness of equipment fault processing is greatly improved;Then edge management device can verify the consistency of initial inspection result, obtain final inspection result, to improve the accuracy of equipment fault identification, and finally remote monitoring platform can handle the fault of target area according to final inspection result.The application realizes automatic fault identification and verification of image data returned by unmanned aerial vehicle in power inspection process through inspection service management system, so as to improve the accuracy of equipment fault identification and the timeliness of fault handling. BRIEF DESCRIPTION OF DRAWINGS

[0041] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0042] Fig. 1 is a structural schematic diagram of an inspection service management system provided by an embodiment of the present application;

[0043] Fig. 2 is a structural schematic diagram of an edge management device provided by an embodiment of the present application;

[0044] Fig. 3 is a structural schematic diagram of an edge computing device provided by an embodiment of the present application;

[0045] Fig. 4 is an interaction schematic diagram of an image identification process provided by an embodiment of the present application. DETAILED DESCRIPTION

[0046] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0047] The unmanned aerial vehicle can carry high-resolution visible light cameras, high-precision infrared thermal imagers, three-dimensional laser radar scanning equipment and other advanced detection equipment, greatly enriching the inspection means of the power transmission line, effectively improving the work efficiency of power inspection, and significantly reducing the operation and maintenance cost and pressure of the power transmission line. However, the data backhaul processing method used in the current unmanned aerial vehicle inspection method needs maintenance personnel to manually process after the unmanned aerial vehicle data backhaul, resulting in poor accuracy of equipment fault identification and timeliness of fault processing.

[0048] Based on this, the technical scheme is proposed as follows, please refer to the following text:

[0049] In one embodiment, as shown in Figure 1, Figure 1 is a structure schematic diagram of a kind of inspection business management system provided by the embodiment of the application;The application improves a kind of inspection business management system, can include edge management device, remote monitoring platform, wireless communication base station and edge computing device.

[0050] The remote monitoring platform and the edge computing device are electrically connected with the edge management device through the wireless communication base station respectively;The edge computing device is hung in unmanned aerial vehicle;

[0051] The edge management device is used to generate the path planning result of target area based on the inspection start instruction when receiving the inspection start instruction sent by remote monitoring platform;

[0052] The edge computing device is used to control the inspection operation of the unmanned aerial vehicle based on the path planning result, to generate the initial inspection result of the target area;

[0053] The edge management device is further used to carry out consistency verification on the initial inspection result, and obtain final inspection result;

[0054] The remote monitoring platform is used to carry out fault processing on the target area according to the final inspection result.

[0055] In the embodiment, as shown in Figure 1, the inspection business management system is composed of edge management device, remote monitoring platform, wireless communication base station and edge computing device. Among them, the wireless communication base station as the core equipment of wireless communication network, can be electrically connected with the edge management device respectively by remote monitoring platform and edge computing device, so that users can interact with edge computing device through edge management device on remote monitoring platform, and, the edge computing device can be hung in unmanned aerial vehicle, so as to control the flight direction of unmanned aerial vehicle. In general, the application constitutes a complete power inspection system through these devices.

[0056] Specifically, the remote monitoring platform serves as an interactive bridge between the user and the inspection business management system, which can generate corresponding instructions according to the relevant operations of the user and send them to the edge management device to trigger the UAV to conduct power inspection in the target area. In detail, when the edge management device receives the inspection start instruction sent by the remote monitoring platform, it can generate a path planning result of the UAV in the target area based on the inspection start instruction, and then send the path planning result to the edge computing device mounted on the UAV, so that the edge computing device can control the UAV to fly on the planned path according to the path planning result. In the process of flight inspection of the UAV, the edge computing device can also synchronously identify the fault of the power equipment, and generate the initial inspection result when the power equipment has a fault and send it to the edge management device. At this time, the edge management device can further verify the initial inspection result, and generate the final inspection result according to the verification result and return it to the remote monitoring platform. Finally, the remote monitoring platform can select the corresponding fault handling measures according to the final inspection result to handle the fault of the power equipment in the target area.

[0057] It can be understood that the edge management device, the edge computing device and the remote monitoring platform in the present application are all provided with corresponding data processing strategies. Therefore, in the process of UAV inspection, the edge computing device can acquire the corresponding device image in real time for fault identification, so as to improve the efficiency of device fault identification. If the edge computing device identifies that the power equipment has a fault, the edge management device can further acquire the high-quality device image of the power equipment to confirm the fault condition of the power equipment again, so as to avoid the false detection of the edge computing device, thereby improving the accuracy of device fault identification. When the edge management device determines that the power equipment has a fault, the remote monitoring platform can also issue relevant instructions to dispatch corresponding intelligent operation and maintenance equipment or operation and maintenance personnel to the target area to handle the fault of the power equipment, so as to improve the timeliness of device fault handling.

[0058] Further, the power transmission line responsible for inspection management by the inspection business management system often spans a large geographical area, covering mountains, plains, rivers and other terrains. In addition, the power equipment in the power transmission line can include towers, ground wires, insulators, fittings, auxiliary facilities, grounding devices and other types. Based on this, when the inspection business management system conducts power inspection in the target area, it can trigger multiple UAVs to conduct synchronous inspection to improve the inspection efficiency. Each UAV corresponds to an edge computing device, and the processes of each edge computing device in processing device data are independent of each other.

[0059] Further, the unmanned aerial vehicle for power inspection in cooperation with the inspection business management system can include various types, such as multi-rotor unmanned aerial vehicles, fixed-wing unmanned aerial vehicles, and unmanned unmanned aerial vehicles, and the business management system can plan inspection paths and inspection targets of different unmanned aerial vehicles according to the characteristics of each type. For example, multi-rotor unmanned aerial vehicles are small in size and good in control performance, and are suitable for operation in narrow or complex environments; fixed-wing unmanned aerial vehicles are fast in flight speed and wide in coverage, and are suitable for large-area and long-distance inspection operation; and unmanned unmanned aerial vehicles are strong in load capacity and good in stability, and are suitable for mounting heavy equipment or long-time hovering operation.

[0060] In the above embodiment, the system includes an edge management device, a remote monitoring platform, a wireless communication base station, and an edge computing device, and the remote monitoring platform and the edge computing device are electrically connected with the edge management device through the wireless communication base station respectively, and the edge computing device is mounted in the unmanned aerial vehicle. Among them, when the edge management device receives the inspection starting instruction sent by the remote monitoring platform, the path planning result of the target area can be generated based on the inspection starting instruction, so that the edge computing device can control the inspection operation of the unmanned aerial vehicle based on the path planning result to generate the initial inspection result of the target area. Through the operation of completing fault identification on the unmanned aerial vehicle, the synchronism of power inspection and fault identification can be ensured, and the timeliness of equipment fault processing is greatly improved; then the edge management device can verify the consistency of the initial inspection result to obtain the final inspection result, so as to improve the accuracy of equipment fault identification, and finally the remote monitoring platform can perform fault processing on the target area according to the final inspection result. Through the inspection business management system, the application realizes automatic fault identification and verification of image data returned by the unmanned aerial vehicle in the power inspection process, thereby improving the accuracy of equipment fault identification and the timeliness of fault processing.

[0061] In one embodiment, the inspection business management system can further include a camera.

[0062] The camera is mounted on the unmanned aerial vehicle and electrically connected with the edge computing device, and is used for image shooting of the power equipment in the target area.

[0063] In this embodiment, the inspection business management system can further include a camera mounted on the unmanned aerial vehicle and electrically connected with the edge computing device. Based on this, when the unmanned aerial vehicle flies to the position of the power equipment, the edge computing device can drive the camera to shoot the power equipment to obtain the basic data for equipment fault identification of the inspection business management system.

[0064] Specifically, the camera in the present application can adopt a high-definition camera, an infrared thermal imager, a long-focus camera or a multi-spectral camera, which is not limited here. Among them, the image effect obtained by each type of camera is different, and the applicable inspection area is also different, so the present application can select the type of camera based on the specific inspection task and the geographical environment of the inspection area, which is not limited here.

[0065] For example, the high-definition camera can capture clear power line and equipment images, so it is suitable for general power inspection tasks, such as checking towers, transmission lines, line corridors, etc.; the infrared thermal imager can detect temperature anomalies of power equipment to find potential fault points, such as poor contact, overload, etc., so it is suitable for night or bad weather conditions, and tasks that require detection of equipment temperature anomalies; the long-focus camera can clearly capture the details of distant power lines and equipment, so it is suitable for tasks that require long-distance observation or detailed inspection of power lines and power equipment; and the multi-spectral camera can obtain multi-spectral images of power lines and power equipment, including visible light, infrared, ultraviolet, etc. Therefore, it is suitable for power inspection tasks in complex terrain and variable environmental conditions.

[0066] In one embodiment, as shown in Figure 2, Figure 2 is a structural schematic diagram of an edge management device provided by an embodiment of the present application; the edge management device can include an instruction analysis module, a path planning module and a result generation module.

[0067] The instruction analysis module is configured to analyze the inspection start instruction to obtain a target area and an inspection type corresponding to the inspection start instruction;

[0068] The path planning module is configured to plan a path for the target area based on the inspection type, generate an inspection path of at least one of the drones and an inspection target corresponding to the inspection path; wherein the inspection target includes at least one inspection type;

[0069] The result generation module is configured to determine a lightweight fault identification model corresponding to each inspection type, and generate a path planning result corresponding to each of the drones according to the inspection path, the inspection target and the lightweight fault identification model of each of the drones.

[0070] In this embodiment, as shown in FIG. 2, the edge management device can also be divided into multiple functional modules, such as an instruction analysis module, a path planning module, and a result generation module, each of which performs a corresponding function. For example, the instruction analysis module can analyze the target area and at least one inspection type contained in the target area carried in the inspection start instruction. The path planning module can plan the inspection path and inspection target of each drone in the target area. The result generation module can determine the lightweight fault identification model corresponding to each inspection type, and then generate the path planning result corresponding to each drone.

[0071] The inspection types of the target area can be divided according to common problems of power equipment faults, such as small part inspection type, connector type, foreign object hanging line type, and plant climbing line type, etc. The small part inspection type can be non-standard installation of pins, pin disengagement, pin disappearance, nut disappearance, bolt and nut rust, etc. The connector type can be the connection of metal connectors, bolts and screws provided on the tower line, etc. The foreign object hanging line type can be plastic hanging on the line, animal carcasses or plants hanging on the line, etc. The plant climbing line type can be that the plant growth around the tower exceeds the warning line, etc.

[0072] It should be noted that the inspection target of the drone can include at least one inspection type. For example, if the inspection target of the drone is to inspect the pin, the inspection type it includes can be the small part inspection type, and the path planning result corresponding to the drone carries the lightweight fault identification model corresponding to the small part inspection type. If the inspection target of the drone is to inspect the bolt, the inspection type it includes can include the small part inspection type and the connector type, and the path planning result corresponding to the drone carries the lightweight fault identification model corresponding to the small part inspection type and the connector type.

[0073] It can be understood that the lightweight fault identification model in this application refers to a model that performs fault identification on the input device image of the corresponding inspection type and obtains a fault identification result. When training the lightweight fault identification model, the sample device images of different devices in the same inspection type can be used as training samples, and each training sample can be labeled with a sample label, i.e., the corresponding fault identification result. When all the training samples are labeled, the training samples with sample labels can be input into the preset initial training model for forward propagation to train the model, and the model can be parameter-optimized using a preset target loss function during the backward propagation of the model. When the model meets certain training conditions or parameter convergence conditions, such as when the number of iterations reaches a set value, it is considered that the training is complete. At this time, the trained model can be lightened to obtain the final lightweight fault identification model.

[0074] Further, the application can set different initial training models for the fault features of each inspection type. For example, the Faster R-CNN algorithm has the characteristics of high precision and strong stability, and can identify some metal types with connectors, such as the connector types in the tower line, so the application can use the Faster R-CNN algorithm to train the initial training model of the connector type. YOLOR3 or 5 or 7 has the characteristics of fast speed and strong real-time, and can identify small connecting metal types, such as small part inspection types, so the application can use YOLOR3 or 5 or 7 channels to train the initial training model of the small part inspection type. In addition, the CBAM attention mechanism can be introduced into YOLOR3 or 5 or 7 channels to enhance the expressiveness of the backbone network feature information and improve the fault recognition accuracy.

[0075] In one embodiment, the path planning module for path planning of the target area based on the inspection type can include:

[0076] The path planning module divides the target area into multiple sub-areas, obtains device data in each sub-area, and optimizes a preset number of drones based on the device data.

[0077] In this embodiment, when the path planning module plans the path of the target area, it can first divide the target area into multiple sub-areas according to a preset size, and then obtain device data in each sub-area. The device data can include the geographic location and inspection type of each power device in the sub-area. Therefore, the path planning module can optimize a preset number of drones based on the device data, so that the inspection path of each drone is the shortest, and the power devices corresponding to the inspection targets in the inspection path are the most. This can greatly improve the power inspection efficiency and reduce the waste of drone resources.

[0078] Specifically, the path planning module can use an improved genetic algorithm to optimize a preset number of drones. The objective function used in this optimization process is to make the inspection path the shortest and the power devices corresponding to the inspection targets the most. The boundary conditions can include: for any inspection station, only one drone can start from the station, perform all inspection tasks, and then return to the station; a inspection shooting point can be visited at most once; the in-degree and out-degree of each inspection shooting point are equal; a certain number of inspection shooting points need to be set in each inspection task; a drone cannot fly directly from one station to another station; and the MTZ constraint method is used to prevent sub-loops.

[0079] More specifically, in the process of multi-objective optimization, the path planning module can determine the corresponding fitness function according to the objective function. In detail, the path planning module first initializes the population, sets the initial iteration value, sets the initial Q value table, such as 1 or the maximum number of iterations N; takes the initialized population as the current population, calculates the fitness value of the individuals in the current population according to the determined fitness function; then, according to the calculated fitness value and the current Q value table, the fitness state is evaluated, the initial iteration number is taken as the current iteration number, and the initial Q value table is taken as the current Q value. If the current iteration number is greater than 1, the reward value is calculated according to the set reward function, the action value is generated by means of the set action selection strategy according to the reward value and the state value. If the iteration number is greater than a specified threshold, the Q value is calculated according to Q-learning, the current Q value table is updated, the action value is regenerated based on the Q value and the set action selection strategy, and then the crossover and mutation probabilities are generated according to the regenerated action value. If the iteration number is less than or equal to the specified threshold, the Q value is calculated according to SARS A, the current Q value table is updated, and then the crossover and mutation probabilities are generated based on the Q value and the action value. If the current iteration number is equal to 1, the crossover and mutation probabilities are set according to the "roulette" method. After the crossover and mutation probabilities are determined, the path planning module performs two-point crossover operator, chromosome verification and correction mechanism, combination exchange mutation, and local optimization processing based on the greedy strategy to generate a new population. Finally, it can be determined whether the iteration number reaches N. If not, the new population is taken as the current population, and the step of calculating the fitness value of the individuals in the current population according to the determined fitness function is returned. If yes, the current population is output, the fitness value is calculated, and then the final optimization target is obtained.

[0080] In one embodiment, as shown in FIG. 3, FIG. 3 is a structural schematic diagram of an edge computing device provided by an embodiment of the present application; the edge computing device comprises an instruction control module, a type identification module, and a fault identification module; wherein the path planning result can comprise an inspection path, an inspection target, and at least one lightweight fault identification model.

[0081] The instruction control module is configured to control the UAV to fly according to the inspection path after receiving the path planning result sent by the edge management device, and control the camera to take pictures of the power equipment when the UAV flies to the power equipment corresponding to the inspection target, to obtain an equipment image.

[0082] The type identification module is configured to perform type identification on the equipment image to obtain an inspection type corresponding to the equipment image.

[0083] The fault identification module is configured to determine a lightweight fault identification model corresponding to the inspection type, and perform fault identification on the equipment image by using the lightweight fault identification model to obtain a fault identification result of the power equipment.

[0084] The instruction control module is further configured to generate flight instructions corresponding to the fault identification result, and control the flight path of the UAV through the flight instructions.

[0085] In this embodiment, the instruction control module can also be divided into multiple functional modules, such as an instruction control module, a type identification module, and a fault identification module. Each module performs a corresponding function. For example, the instruction control module is mainly responsible for controlling the camera to take pictures and controlling the UAV to fly. The type identification module is mainly responsible for identifying the type of the equipment image taken by the camera. The fault identification module is mainly responsible for identifying the fault of the equipment image based on the inspection type.

[0086] Specifically, after the instruction control module receives the path planning result sent by the edge management device, it can forward the lightweight fault identification model in the path planning result to the fault identification module, and generate flight instructions according to the inspection path and inspection target in the path planning result, and issue the flight instructions to the UAV, so that the UAV flies according to the inspection path, and hovers at the corresponding shooting point of the power equipment corresponding to the inspection target when flying to the position of the power equipment. Then, the instruction control module can issue a shooting instruction to the camera to control the camera to take pictures of the power equipment and obtain an equipment image. Finally, the instruction control module can forward the equipment image to the type identification module and the fault identification module, so that the type identification module can identify the inspection type of the equipment image, and the fault identification module can identify the equipment fault of the equipment image.

[0087] Further, the application can pre-store a trained type identification model in the type identification module, so that when subsequent image type identification is needed, the type identification module can directly call the type identification model to perform identification operation. Wherein, when training the type identification model, a corresponding neural network model can be selected as a preset initial training model for improvement and training, such as R-CNN (Region with CNN Feature), SSD (Single Shot MultiBox Detector), YOLO or similar structure neural network, etc. In the training process, a sample device image can be obtained first, and the sample device image is labeled with a corresponding sample label, which is the true inspection type of the sample device image. Thus, after inputting the sample device image into the preset initial training model, the predicted inspection type output by the initial training model can be obtained. Then, the application can take the target that the predicted inspection type approaches the true inspection type of the sample device image, and train the initial training model using a target loss function. When the initial training model meets the preset training condition, the trained initial training model is taken as the type identification model, so as to determine the type identification model.

[0088] Illustratively, as shown in FIG. 4, FIG. 4 is an interaction schematic diagram of an image identification process provided by an embodiment of the application; wherein the edge computing device can include one or more light fault identification models of different inspection types and a type identification model, and the type identification model is connected with each light fault identification model. Thus, when the type identification model receives a device image captured by a camera, the inspection type of the device image can be identified, and then the device image is input into the light fault identification model corresponding to the inspection type to obtain the fault identification result output by the light fault identification model.

[0089] Further, if the identification result of the device image output by the type identification model includes multiple inspection types, the type identification module can also perform image cutting on the device image according to each inspection type, so that the sub-image obtained by cutting contains a single inspection type, and then each sub-image can be normalized and pre-processed, and each pre-processed sub-image is input into the corresponding light fault identification model for fault identification.

[0090] In one embodiment, the instruction control module is further configured to generate flight instructions corresponding to the fault identification result, and control the flight path of the UAV through the flight instructions, which can include:

[0091] When the fault identification result is normal, the instruction control module generates a continue flight instruction, and controls the UAV to continue flying according to the inspection path through the continue flight instruction;

[0092] When the fault identification result is abnormal, the instruction control module generates a return flight instruction, controls the UAV to fly to the location of the power equipment through the return flight instruction, generates an initial inspection result corresponding to the power equipment, and sends the initial inspection result to the edge management system.

[0093] In this embodiment, the fault identification result output by the lightweight fault identification model can be normal or abnormal. When the fault identification result is normal, it indicates that the corresponding power equipment is not faulty. When the fault identification result is abnormal, it indicates that the corresponding power equipment is faulty. Based on this, the instruction control module can generate corresponding flight instructions according to the actual situation of the fault identification result to control the flight path of the UAV.

[0094] Specifically, when the fault identification result is normal, the instruction control module can generate a continue flight instruction and issue the continue flight instruction to the UAV, so that the UAV can continue to fly according to the inspection path to the next power equipment to continue equipment fault identification. When the fault identification result is abnormal, the instruction control module can generate a return flight instruction and issue the return flight instruction to the UAV, so that the UAV can fly back to the location of the faulty power equipment for further processing, and then generate an initial inspection result corresponding to the power equipment and return it to the edge management system.

[0095] Of course, after the instruction control module returns the initial inspection result of the faulty power equipment to the edge management system, it can continue to generate a continue flight instruction and issue the continue flight instruction to the UAV, so that the UAV can continue to fly according to the inspection path to the next power equipment to continue equipment fault identification.

[0096] In one embodiment, the process of generating an initial inspection result corresponding to the power equipment by the instruction control module can include:

[0097] After the instruction control module determines the fault location of the power equipment according to the fault identification result, it controls the UAV to hover at the fault location and controls the camera to take pictures of the power equipment according to a preset shooting strategy to obtain a target image. In addition, the instruction control module obtains the inspection type, fault identification result, and target image of the power equipment to form an initial inspection result.

[0098] In this embodiment, after the UAV returns to the location of the faulty power equipment, the instruction control module can determine the specific fault location of the power equipment according to the identification result, then control the UAV to fly to the fault location and hover, and control the camera to take pictures of the power equipment according to the shooting strategy to obtain a target image. Then, the instruction control module can generate an initial inspection result based on the inspection type, fault identification result, and target image of the power equipment.

[0099] It is understandable that the preset shooting angle here refers to the shooting rules and parameters that the user predefines according to the type of power equipment, including but not limited to shooting angle and height, image resolution and quality, shooting sequence and lighting conditions, so as to improve the quality and reliability of the target image obtained.

[0100] For example, the command control module can ensure clear images of the fault area by defining the specific hovering height of the drone and the shooting angle relative to the power equipment; high-quality images can be obtained by setting parameters such as the camera's image resolution, frame rate, and exposure time; the timing and coverage of the target images can be improved by specifying the drone to shoot at multiple preset shooting angles and time intervals; and high-quality images can be obtained under different lighting conditions by adjusting the camera's light compensation, white balance, and other settings for different lighting conditions.

[0101] In one embodiment, the initial inspection results may include inspection type, fault identification results, and target image; the edge management device may also include an image recognition module and a result comparison module.

[0102] The image recognition module model is used to determine the target fault identification model corresponding to the inspection type after receiving the initial inspection result sent by the edge computing module, and to use the target fault identification model to perform fault identification on the fault identification result to obtain the target identification result; wherein, the target fault identification model is the lightweight fault identification model before lightweighting;

[0103] The result comparison module is used to compare the target identification result with the fault identification result, generate the final inspection result based on the comparison result, and send the final inspection result to the remote monitoring platform.

[0104] In this embodiment, the edge management device can also be divided into multiple functional modules, such as an image recognition module and a result comparison module. Each functional model performs a corresponding function. For example, the image recognition module model is mainly responsible for fault identification of target images of power equipment, while the result comparison module is mainly responsible for comparing the fault identification results sent by the edge computing device with the target identification results obtained by the image recognition module, and then generating the final inspection results to be sent to the remote monitoring platform.

[0105] It can be understood that the target fault identification model adopted in the image recognition module model is a light fault identification model before being lightened. Here, lightening refers to a technology for improving storage efficiency, loading speed and rendering efficiency by reducing data volume and file size while maintaining the basic appearance and structure of the model; compared with the target fault identification model, the light fault identification model will lack some model detail information, affecting the recognition accuracy of the model. Therefore, the edge management device can use the lightened target fault identification model to identify the high-quality target image of the power equipment with high precision, so as to improve the accuracy of fault identification.

[0106] In one embodiment, the result comparison module can be configured to generate a final inspection result according to the comparison result, and the process can include:

[0107] When the comparison result is consistent, the result comparison module obtains the fault position corresponding to the target image and generates a final inspection result according to the fault position and the target identification result;

[0108] When the comparison result is inconsistent, the result comparison module obtains the initial inspection result and the target identification result to form a final inspection result.

[0109] In this embodiment, if the comparison result generated by the result comparison module is consistent, it means that the fault identification result identified by the light fault identification model is correct, so the result comparison model can obtain the fault position corresponding to the target image and generate a final inspection result according to the fault position and the target identification result, so that the subsequent remote monitoring platform can dispatch the corresponding intelligent operation and maintenance equipment to the fault position for fault handling according to the final inspection result; if the generated comparison result is inconsistent, it means that the fault identification result identified by the light fault identification model may be incorrect, so the result comparison model can obtain the initial inspection result and the target identification result to form a final inspection result, so that the subsequent operation and maintenance personnel can manually judge the fault problem of the power equipment according to the final inspection result.

[0110] Further, after the operation and maintenance personnel manually judge the fault problem of the power equipment, the present application can also determine the model with incorrect identification result according to the judgment result, so as to retrain the model until its recognition accuracy meets the preset ending condition.

[0111] In one embodiment, the remote monitoring platform can be configured to perform fault handling on the target area according to the final inspection result, and the process can include:

[0112] The remote monitoring platform determines a fault type corresponding to the final inspection result, and when the fault type is a specified type, issues a fault processing instruction to an intelligent operation and maintenance device corresponding to the fault type, so that the intelligent operation and maintenance device goes to the target area for fault processing; and when the fault type is a non-specified type, pushes the final inspection result to an operation and maintenance personnel of the target area.

[0113] In this embodiment, after receiving the final inspection result sent by the result comparison module, the remote monitoring platform can first determine the fault type corresponding to the final inspection result. The fault type can include a specified type and a non-specified type. For different fault types, the remote monitoring platform can adopt different strategies for fault processing. For example, when the fault type is a specified type, the remote monitoring platform can issue a fault processing instruction to an intelligent operation and maintenance device corresponding to the fault type, so that the intelligent operation and maintenance device goes to the target area for fault processing; when the fault type is a non-specified type, the remote monitoring platform can push the final inspection result to an operation and maintenance personnel of the target area, so that the operation and maintenance personnel goes to the target area for fault processing.

[0114] It can be understood that the specified type refers to a fault in the power system that is common and easy to automatically identify and process. These faults usually have clear identification criteria, processing methods and steps, and do not require or require little human intervention, so the remote monitoring platform can pre-set the automatic processing flow of these fault types. The non-specified type refers to a fault in the power system that is rare or relatively complex. These faults can be caused by multiple reasons, and the corresponding automatic processing flow cannot be set or human intervention is required, so the remote monitoring platform needs to push the final inspection result to the operation and maintenance personnel of the target area to notify the operation and maintenance personnel to process the fault in time.

[0115] For example, if the fault type corresponding to the final inspection result is a tree branch shielding equipment, the remote monitoring platform can issue a fault processing instruction to a helicopter cutting unmanned aerial vehicle carrying a tree branch cutting mechanism to control the helicopter cutting unmanned aerial vehicle to go to the tree branch at the fault position in the target area for cutting and cleaning; if the fault type corresponding to the final inspection result is a foreign matter winding line, the remote monitoring platform can issue a fault processing instruction to an automatic cleaning laser to control the automatic cleaning laser to go to the fault position in the target area and start laser scanning or cutting the foreign matter winding on the line.

[0116] Finally, it should be noted that the terms "first", "second", and the like, herein do not denote any order, quantity, combination, or importance, but rather are used to distinguish one element from another, and are not intended to denote the presence of any such actual relationship or order. Moreover, the terms "include", "have", or any other variant thereof are intended to encompass non-exclusive inclusions, such that processes, methods, articles, or apparatuses that comprise a list of elements are not required to comprise only those elements in the list, but can include other elements not expressly listed, or also include elements inherent in such processes, methods, articles, or apparatuses. Without additional restrictions, an element preceded by "comprises... a" does not exclude the existence of additional identical elements in the process, method, article, or apparatus that comprises the stated element.

[0117] The various embodiments in the specification are described in progressive order with each embodiment building on one or more of the previous embodiments, however the order of the embodiments described is not intended to be construed as a requirement or limitation for these embodiments. Any one or more of the embodiments described with reference to a particular set of one or more other embodiments are optionally employable together with one or more other embodiments and / or in any appropriate combination.

[0118] The above description of disclosed embodiments is intended to enable those skilled in the art to make or use the application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein can be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A patrol service management system characterized by comprising: The system comprises an edge management device, a remote monitoring platform, a wireless communication base station and an edge computing device; The remote monitoring platform and the edge computing device are electrically connected with the edge management device through the wireless communication base station respectively; The edge computing device is mounted in a UAV; The edge management device is configured to generate a path planning result of a target area based on a patrol starting instruction received from the remote monitoring platform; The edge computing device is configured to control the patrol operation of the UAV based on the path planning result to generate an initial patrol result of the target area; The edge management device is further configured to perform consistency verification on the initial patrol result to obtain a final patrol result; The remote monitoring platform is configured to perform fault processing on the target area according to the final patrol result.

2. The patrol service management system according to claim 1, characterized by, The patrol service management system further comprises a camera; The camera is mounted on the UAV and electrically connected with the edge computing device, and is configured to perform image shooting on power equipment in the target area.

3. The patrol service management system according to claim 1, characterized by, The edge management device comprises an instruction analysis module, a path planning module and a result generation module; The instruction analysis module is configured to analyze the patrol starting instruction to obtain a target area and a patrol type corresponding to the patrol starting instruction; The path planning module is configured to perform path planning on the target area based on the patrol type to generate a patrol path of at least one UAV and a patrol target corresponding to the patrol path; wherein the patrol target comprises at least one patrol type; The result generation module is configured to determine a lightweight fault identification model corresponding to each patrol type, and generate a path planning result corresponding to each UAV according to the patrol path, the patrol target and the lightweight fault identification model of each UAV.

4. The patrol service management system according to claim 1, characterized by, The process in which the path planning module performs path planning on the target area based on the patrol type comprises: The path planning module divides the target area into a plurality of sub-areas, determines equipment data in each sub-area, and performs multi-objective optimization on a preset number of UAVs based on each equipment data, with the objectives of minimizing the patrol path and maximizing the power equipment corresponding to the patrol target.

5. The patrol service management system according to claim 2, characterized by, The path planning result comprises a patrol path, a patrol target and at least one lightweight fault identification model; the edge computing device comprises an instruction control module, a type identification module and a fault identification module; The instruction control module is configured to control the UAV to fly according to the patrol path after receiving the path planning result sent by the edge management device, and control the camera to shoot the power equipment corresponding to the patrol target when the UAV flies to the power equipment, to obtain an equipment image; The type identification module is configured to perform type identification on the equipment image to obtain a patrol type corresponding to the equipment image; The fault identification module is configured to determine a lightweight fault identification model corresponding to the patrol type, and perform fault identification on the equipment image by using the lightweight fault identification model to obtain a fault identification result of the power equipment; The instruction control module is further configured to generate flight instructions corresponding to the fault identification result, and control a flight path of the UAV through the flight instructions.

6. The patrol service management system according to claim 5, characterized by The process in which the instruction control module generates flight instructions corresponding to the fault identification result and controls a flight path of the UAV through the flight instructions comprises: When the fault identification result is normal, the instruction control module generates a continue-to-fly instruction, and controls the UAV to continue flying along the inspection path according to the continue-to-fly instruction; When the fault identification result is abnormal, the instruction control module generates a return-to-fly instruction, and controls the UAV to fly to the location of the power equipment according to the return-to-fly instruction, generates an initial inspection result corresponding to the power equipment, and sends the initial inspection result to the edge management system.

7. The patrol service management system according to claim 6, characterized by, The process in which the instruction control module generates an initial inspection result corresponding to the power equipment comprises: After determining the fault position of the power equipment according to the fault identification result, the instruction control module controls the UAV to hover at the fault position, controls the camera to take pictures of the power equipment according to a preset shooting strategy, and obtains a target image, and acquires the inspection type, the fault identification result and the target image of the power equipment to form an initial inspection result. The initial inspection result comprises the inspection type, the fault identification result and the target image; the edge management device further comprises an image recognition module and a result comparison module; 8. The patrol business management system according to claim 1, characterized by, The image recognition module is configured to, after receiving the initial inspection result sent by the edge computing module, determine a target fault identification model corresponding to the inspection type, and perform fault identification on the fault identification result by using the target fault identification model to obtain a target identification result; wherein the target fault identification model is a lightweight fault identification model before being lightened. The result comparison module is configured to compare the target identification result with the fault identification result, generate a final inspection result according to a comparison result, and send the final inspection result to the remote monitoring platform. The process in which the result comparison module generates a final inspection result according to a comparison result comprises:

9. The patrol service management system according to claim 8, characterized by, When the comparison result is consistent, the result comparison module acquires a fault position corresponding to the target image, and generates a final inspection result according to the fault position and the target identification result; When the comparison result is inconsistent, the result comparison module acquires the initial inspection result and the target identification result to form a final inspection result. The process in which the remote monitoring platform performs fault processing on the target area according to the final inspection result comprises:

10. The patrol business management system according to claim 1, characterized by, The remote monitoring platform determines a fault type corresponding to the final inspection result, and when the fault type is a specified type, issues a fault processing instruction to an intelligent operation and maintenance device corresponding to the fault type, so that the intelligent operation and maintenance device goes to the target area to perform fault processing; and when the fault type is a non-specified type, pushes the final inspection result to an operation and maintenance personnel of the target area. ​

Images