Inspection robot and power inspection method and system based on the same

Abstract

The application provides a kind of inspection robot and power inspection method and system based on inspection robot, the method comprises: image data acquisition step, using the vision equipment on the inspection robot to collect the image data of target inspection power equipment and input into the target detection model in the inspection robot;Image data analysis step, utilize target detection model based on the input image data output equipment operating condition information and corresponding confidence;Based on confidence rejection result step, in the case where confidence is not less than preset threshold, equipment operating condition information and confidence are stored to preset position;In the case where confidence is less than preset threshold, first execute correction operation, then re-execute image data acquisition step, image data analysis step and based on confidence rejection result step;When the number of times of correction operation execution reaches preset maximum correction number, stop executing correction operation.The application can realize efficient, automatic inspection of power equipment or system.

Description

Technical Field

[0001] This invention relates to the fields of artificial intelligence and power line inspection technology, and in particular to an inspection robot and a power line inspection method and system based on the inspection robot. Background Technology

[0002] In the inspection of power equipment (such as substations), traditional manual inspections are labor-intensive, time-consuming, and require a high level of skill from workers and are subject to favorable weather conditions. Current technologies utilize power inspection robots to replace human inspectors in these scenarios, performing inspections according to pre-set tasks. This approach is both convenient and reliable, significantly reducing labor costs.

[0003] However, existing power inspection robots only perform inspections according to preset tasks, and cannot make further judgments when encountering emergencies, so they can only call for human intervention.

[0004] Therefore, how to provide an inspection robot with certain intelligent processing capabilities and a corresponding inspection method is a technical problem that urgently needs to be solved. Summary of the Invention

[0005] In view of this, embodiments of the present invention provide an inspection robot and a power inspection method and system based on the inspection robot, so as to eliminate or improve one or more defects existing in the prior art.

[0006] One aspect of the present invention provides a power line inspection method based on an inspection robot, the method comprising the following steps:

[0007] The image data acquisition step involves using a vision device integrated on the inspection robot to acquire image data of the target power equipment being inspected, and then inputting the image data into the pre-trained target detection model built into the inspection robot.

[0008] The image data analysis step involves using a target detection model to perform calculations and analyses based on the input image data and outputting equipment operation status information and the corresponding confidence level of the equipment operation status information.

[0009] Based on the confidence level selection step, if the confidence level is not lower than the preset threshold, the equipment operation status information and the corresponding confidence level are stored in the preset location, and the current power inspection of the target power equipment is marked as completed; if the confidence level is lower than the preset threshold, the correction operation for adjusting the image data acquisition conditions of the inspection robot is executed first, and then the image data acquisition step, image data analysis step, and confidence level selection step are executed again.

[0010] When the number of correction operations reaches the preset maximum number of corrections, the correction operations will stop, and the current power inspection of the target power equipment will be marked as complete.

[0011] In some embodiments of the present invention, when the number of times the correction operation is executed reaches a preset maximum number of corrections, the correction operation is stopped, and the current power inspection of the target power equipment is marked as completed. This includes: when the number of times the correction operation is executed reaches a preset maximum number of corrections, the correction operation is stopped, and the image data of the current target power equipment collected during the current power inspection is transmitted to a cloud platform; receiving equipment operation status information of the current target power equipment obtained by manual evaluation based on the transmitted image data from the cloud platform; and marking the current power inspection of the current target power equipment as completed. Alternatively, when the number of times the correction operation is executed reaches a preset maximum number of corrections, the correction operation is stopped, and power equipment whose confidence level of the image data collected during the current power inspection is always lower than a preset threshold is recorded; after the current power inspection of all power equipment to be inspected is completed, the image data of power equipment whose confidence level is still lower than the preset threshold after reaching the maximum number of corrections is summarized and transmitted to a centralized processing device.

[0012] In some embodiments of the present invention, the target detection model is trained using a preset number of normal operation images and various equipment fault images of the target inspection power equipment as a training set.

[0013] In some embodiments of the present invention, after receiving the equipment operation status information of the current target power equipment under inspection, which is obtained by manual evaluation based on the transmitted image data from the cloud platform, the method further includes: the vision device using the equipment operation status information of the current target power equipment under inspection, which is manually evaluated by the cloud platform, as a label, and using the image data collected during this power inspection of the current target power equipment as a training set to train and update the target detection model.

[0014] In some embodiments of the present invention, after obtaining the equipment operation status information of the current target power equipment under inspection based on the transmitted image data through manual evaluation, the method further includes: the cloud platform using the manually evaluated equipment operation status information within a preset historical time period as a label, and using the image data of the power equipment corresponding to the equipment operation status information as a training set to train and update the target detection model.

[0015] In some embodiments of the present invention, the device operation status information includes device normal and device fault type. The method further includes: when the confidence level is not lower than a preset threshold and the device operation status information indicates the device fault type, sending device operation status information containing the device fault type to the cloud platform, so that the cloud platform generates an early warning based on the received device operation status information indicating the device fault type.

[0016] In some embodiments of the present invention, the step of transmitting the image data of the current target power equipment collected during the current power inspection to the cloud platform includes: uploading the image data of the current target power equipment collected in the most recent power inspection to the cloud platform; or uploading the image data of the current target power equipment collected at all historical moments during the current power inspection to the cloud platform; or uploading the image data with the highest confidence level collected at historical moments during the current power inspection to the cloud platform.

[0017] In some embodiments of the present invention, the method further includes: a positioning module integrated on the inspection robot sending location information to a cloud platform in real time, for visualizing the inspection path, current location and power inspection progress of the inspection robot on the cloud platform; the types of correction operations include: adjusting the height, angle, distance position and focus state of the vision device integrated on the inspection robot, supplementing the light source and cleaning the lens of the vision device.

[0018] Another aspect of the present invention provides an inspection robot, the inspection robot comprising:

[0019] The vision device is used to collect image data of the target power equipment for inspection and input the image data into the pre-trained target detection model built into the computing module of the inspection robot.

[0020] The drive module is used to drive the inspection robot to the inspection position;

[0021] The computation module contains a pre-trained target detection model. The target detection model is used to perform calculations and analyses based on the input image data and output device operation status information and the corresponding confidence level of the device operation status information.

[0022] The confidence level selection module is used to store the equipment operation status information and confidence level in a preset location when the confidence level is not lower than a preset threshold, and to mark the current power inspection of the target power equipment as completed. When the confidence level is lower than the preset threshold, it first performs a correction operation to adjust the image data acquisition conditions of the inspection robot, and then re-executes the image data acquisition step, image data analysis step, and confidence level selection result step. When the number of correction operations reaches the preset maximum number of corrections, it stops executing the correction operation and marks the current power inspection of the target power equipment as completed.

[0023] The present invention discloses a power inspection system based on an inspection robot. The power inspection system includes: an inspection robot as described in the above embodiments; and a cloud platform for receiving image data collected by the inspection robot during power inspection, obtaining equipment operation status information of the power equipment based on the manually evaluated image data collected during power inspection, and transmitting the equipment operation status information of the power equipment back to the corresponding inspection robot.

[0024] The inspection robot and the power inspection method and system based on the inspection robot provided by this invention can collect image data of the target power equipment to be inspected based on the pre-trained target detection model built into the inspection robot and the vision device on the inspection robot. It can also intelligently obtain the equipment operating status and related confidence level based on the calculation and analysis of the image data. If the confidence level is insufficient, it can correct the image data acquisition conditions to ensure the high reliability of the data. This enables efficient and automated inspection of power equipment / systems, effectively saving manpower, material resources and financial resources and avoiding the operational risks that may exist in harsh environments.

[0025] Additional advantages, objects, and features of the invention will be set forth in part in the description which follows, and will also become apparent in part to those skilled in the art upon studying the description, or may be learned by practice of the invention. The objects and other advantages of the invention can be realized and obtained by means of the structures specifically pointed out in the description and drawings.

[0026] Those skilled in the art will understand that the objectives and advantages achievable with the present invention are not limited to the specific ones described above, and that the above and other objectives achievable with the present invention will become clearer from the following detailed description. Attached Figure Description

[0027] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, are not intended to limit the scope of the invention. In the drawings:

[0028] Figure 1 This is a flowchart of a power inspection method based on an inspection robot in one embodiment of the present invention.

[0029] Figure 2 This is a schematic diagram of a power equipment scenario - a substation.

[0030] Figure 3 This is a schematic diagram of an inspection robot in a substation scenario according to one embodiment of the present invention. Detailed Implementation

[0031] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the embodiments and accompanying drawings. Here, the illustrative embodiments and descriptions of this invention are used to explain the invention, but are not intended to limit the invention.

[0032] It should also be noted that, in order to avoid obscuring the invention with unnecessary details, only the structures and / or processing steps closely related to the solution according to the invention are shown in the accompanying drawings, while other details that are not closely related to the invention are omitted.

[0033] It should be emphasized that the term "including / comprises" as used herein refers to the presence of a feature, element, step, or component, but does not exclude the presence or addition of one or more other features, elements, steps, or components.

[0034] It should also be noted that, unless otherwise specified, the term "connection" in this article can refer not only to a direct connection, but also to an indirect connection involving an intermediary.

[0035] In the following description, embodiments of the invention will be illustrated with reference to the accompanying drawings. In the drawings, the same reference numerals represent the same or similar parts, or the same or similar steps.

[0036] Figure 2 This is a schematic diagram of a power equipment scenario - a substation. Substations require various types of equipment, including transformers, switches, small transformers, reactive power devices, and other auxiliary equipment such as wave traps, insulators, high-voltage bushings, conductors, grounding devices, secondary equipment, and high-voltage DC equipment.

[0037] To address the problems existing in current power inspection methods using inspection robots, this invention provides an inspection robot and a power inspection method and system based on the inspection robot. This method uses a pre-trained target detection model to intelligently analyze the operating status of power equipment.

[0038] Figure 1 This is a flowchart of a power line inspection method based on an inspection robot according to an embodiment of the present invention. The method includes the following steps:

[0039] Step S110: Image data acquisition step, using the vision device integrated on the inspection robot to acquire image data of the target power equipment to be inspected, and inputting the image data into the pre-trained target detection model built into the inspection robot.

[0040] The target detection model is trained using a preset number of normal operation images and various equipment fault images of the target inspection power equipment as a training set. The model uses manually labeled images corresponding to normal and fault types of equipment as labels, including equipment fault types such as fracture, fire, explosion and short circuit.

[0041] Step S120: Image data analysis step, using the target detection model to perform calculations and analysis based on the input image data and outputting equipment operation status information and the confidence level corresponding to the equipment operation status information.

[0042] The robot's built-in target detection model outputs confidence scores for different equipment operating conditions, classifying them into normal and various identifiable conditions (or malfunctions). The classification confidence scores are mapped to [0,1]. Assuming a threshold of 0.6, resampling and recognition will be triggered when the confidence scores for each category are all below 0.6.

[0043] Step S130: Based on the confidence level selection step, if the confidence level is not lower than the preset threshold, store the equipment operation status information and the corresponding confidence level of the equipment operation status information in the preset location, and mark the completion of this power inspection of the current target power equipment; if the confidence level is lower than the preset threshold, first perform the correction operation to adjust the image data acquisition conditions of the inspection robot, and then re-execute the image data acquisition step, the image data analysis step, and the confidence level selection step.

[0044] The types of correction operations include: adjusting the height, angle, distance, and focus of the vision device integrated into the inspection robot; supplementing the light source; and cleaning the vision device lens. The correction operations are not limited to these; their purpose is to obtain better quality images by correcting the state of the images acquired by the vision device, thereby improving confidence levels.

[0045] Step S140: When the number of times the correction operation is executed reaches the preset maximum number of corrections, stop executing the correction operation and mark the current power inspection of the target power equipment as completed.

[0046] Step S140 above can be as follows: when the number of times the correction operation is executed reaches a preset maximum number of corrections, the correction operation is stopped, and the image data of the current target power equipment collected during this power inspection is transmitted to the cloud platform. The system receives equipment operation status information of the current target power equipment obtained from the cloud platform based on the transmitted image data and through manual evaluation, and marks the completion of this power inspection of the current target power equipment. Alternatively, step S140 can also be as follows: when the number of times the correction operation is executed reaches a preset maximum number of corrections, the correction operation is stopped, and power equipment whose confidence level in the image data collected during this power inspection is consistently lower than a preset threshold is recorded. After the power inspection of all power equipment to be inspected is completed, the image data of power equipment whose confidence level remains lower than the preset threshold after reaching the maximum number of corrections is summarized and transmitted to a centralized processing device. This centralized processing device can be a platform with computing and processing capabilities that can establish a connection with the inspection robot and transmit data. It can be located in the cloud or as an inspection information receiving device located locally at a power equipment such as a substation. This invention is not limited to this; the intended purpose of this limitation is to express that when the number of times the correction operation is executed reaches a preset maximum number of corrections, the correction operation is stopped to avoid falling into an endless loop. After stopping the correction operation, cases where the confidence level is below a threshold can also be directly ignored, and the corresponding equipment operating status information can be recorded. Furthermore, in one embodiment of this invention, if the equipment operating status information corresponding to multiple image data is inconsistent after multiple correction operations and re-acquiring image data of the power equipment, the equipment operating status information corresponding to the case with the highest confidence level is selected and retained.

[0047] In one embodiment of the present invention, step S140 includes: when the number of times the correction operation is executed reaches a preset maximum number of corrections, stopping the execution of the correction operation, transmitting the image data of the current target power equipment collected in this power inspection to the cloud platform, receiving the equipment operation status information of the current target power equipment obtained by manual evaluation based on the transmitted image data from the cloud platform, and marking the completion of this power inspection of the current target power equipment.

[0048] Optionally, in the above embodiments, after receiving the equipment operation status information of the current target power equipment being inspected, obtained through manual evaluation of transmitted image data from the cloud platform, the method further includes: the vision device using the received equipment operation status information of the current target power equipment being inspected, obtained through manual evaluation from the cloud platform, as a label, and using the image data collected during this power inspection of the current target power equipment as a training set, to train and update the target detection model. The purpose of this step is to update the robot's local target detection model based on the relevant data labeled after manual evaluation. Optionally, in another embodiment of the present invention, after the inspection robot receives the equipment operation status information of the current target power equipment being inspected, obtained through manual evaluation of transmitted image data from the cloud platform, the method further includes a data augmentation step. The inspection robot receives the manually labeled content, performs data augmentation to form a training set, and updates the local target detection model. The power inspection robot receives the manually labeled content (which is labeled equipment operation status information and 100% confidence level), performs data augmentation on the manually labeled content to form a training set, and updates the local target detection model.

[0049] Optionally, in the above embodiments, after obtaining the equipment operation status information of the current target inspected power equipment based on the transmitted image data through manual evaluation, the method further includes: the cloud platform using the manually evaluated equipment operation status information within a preset historical time period as labels, and the image data of the power equipment corresponding to the equipment operation status information as a training set, to train and update the target detection model. The purpose of this step is to retrain the model in the cloud based on the relevant data labeled after manual evaluation, and then update the model locally. This can be done periodically or on demand.

[0050] In one embodiment of the present invention, the equipment operation status information includes both normal equipment status and equipment fault type. When the confidence level is not lower than a preset threshold and the equipment operation status information indicates a equipment fault type, the method further includes: sending the equipment operation status information containing the equipment fault type to a cloud platform, so that the cloud platform generates an early warning based on the received equipment operation status information indicating the equipment fault type. Based on this method, the cloud platform can issue an early warning to relevant technical personnel for handling, avoiding personal injury or property damage. The fault types include fracture, fire, explosion, and short circuit, but the present invention is not limited to these.

[0051] In one embodiment of the present invention, when the image data collected by the inspection robot after reaching the maximum number of corrections is still lower than a preset threshold, the image data of the current target power equipment collected in this power inspection is transmitted to the cloud platform. Specifically, this includes: uploading the image data of the current target power equipment collected in the most recent power inspection to the cloud platform; or uploading the image data of the current target power equipment collected at all historical moments of this power inspection to the cloud platform; or uploading the image data with the highest confidence level collected at historical moments of this power inspection to the cloud platform.

[0052] In another embodiment of the present invention, the method further includes: a positioning module integrated on the inspection robot sending location information to the cloud platform in real time, for visually generating the inspection path, current location, and power inspection progress of the inspection robot on the cloud platform. Based on this method, global visual status monitoring of power equipment, including substations, can be achieved. The inspection path can be pre-set manually or generated by the inspection robot after identification and modeling within the inspected area.

[0053] In one embodiment of the present invention, the inspection path is used to number the electrical equipment to be inspected. The inspection robot periodically inspects each numbered electrical equipment within a preset working range, and displays the inspection progress and results in real time on a cloud platform. Optionally, the inspection progress and results are distinguished using preset symbols, colors, and text.

[0054] In a specific embodiment of the present invention, the inspection process of the power inspection robot in a substation scenario is as follows:

[0055] (1) The power inspection robot performs inspections along a predetermined route and uses the built-in target detection network to identify the operating status of the equipment.

[0056] (2) When a fault is detected, the power inspection robot will alert the staff.

[0057] (3) When the confidence level of identification is lower than the threshold, the power inspection robot changes the shooting angle and adjusts the camera focal length to magnify the target and re-identify it.

[0058] (4) If the confidence level is still below the threshold, the power inspection robot will move to a position closer to the target and re-identify it;

[0059] (5) If the confidence level is still below the threshold, the power inspection robot will upload the identified content to the cloud and notify the staff to make a manual judgment.

[0060] (6) After the manual judgment is completed, the relevant data will be labeled and the labeled data will be sent to all power inspection robots to update the local target detection model of the power inspection robots.

[0061] Corresponding to the above method, the present invention also provides an inspection robot, which includes:

[0062] (1) A vision device for acquiring image data of the target power equipment to be inspected and inputting the image data into the pre-trained target detection model built into the computing module of the inspection robot;

[0063] (2) Drive module, used to drive the inspection robot to the inspection position;

[0064] (3) A computation module, wherein a pre-trained target detection model is built into the computation module. The target detection model is used to perform calculation and analysis based on the input image data and output the device operation status information and the confidence level corresponding to the device operation status information.

[0065] (4) A confidence-based rounding module is used to store the equipment operation status information and confidence level in a preset location when the confidence level is not lower than a preset threshold, and mark the current power inspection of the target power equipment as completed; when the confidence level is lower than the preset threshold, it is used to first perform a correction operation to adjust the image data acquisition conditions of the inspection robot, and then re-execute the image data acquisition step, the image data analysis step, and the confidence-based rounding result step; when the number of correction operations reaches the preset maximum number of corrections, it is used to stop the correction operation and mark the current power inspection of the target power equipment as completed.

[0066] Figure 3 This is a schematic diagram of an inspection robot in a substation scenario according to one embodiment of the present invention. A robot is an automated machine, but unlike humans or other living beings, it possesses intelligent capabilities similar to humans, such as perception, planning, movement, and coordination. It is a highly flexible automated machine. Furthermore, a robot should be able to mimic the functions of certain human organs (primarily motor functions), have an independent control system, and be a multi-purpose automated operating device capable of changing its work procedures and being programmed. Industrial robots, in particular, can replace humans in performing monotonous, frequent, and repetitive long-term tasks, or in dangerous or harsh environments.

[0067] In one embodiment of the present invention, the inspection robot is also equipped with sensor devices such as humidity and temperature sensors, temperature sensors, infrared thermal imaging and noise sensors, so as to help determine whether there are abnormal temperatures, humidity or noise at the site of power equipment / system.

[0068] Corresponding to the above method, the present invention also provides a power inspection system based on an inspection robot. The power inspection system includes: an inspection robot as described in the above embodiments; a cloud platform for receiving image data collected by the inspection robot during power inspection, obtaining equipment operation status information of the power equipment based on the manually evaluated image data collected during power inspection, and transmitting the equipment operation status information of the power equipment back to the corresponding inspection robot.

[0069] The inspection robot and the power inspection method and system based on the inspection robot provided by this invention can collect image data of the target power equipment to be inspected based on the pre-trained target detection model built into the inspection robot and the vision device on the inspection robot. It can also intelligently obtain the equipment operating status and related confidence level based on the calculation and analysis of the image data. If the confidence level is insufficient, it can correct the image data acquisition conditions to ensure the high reliability of the data. This enables efficient and automated inspection of power equipment / systems, effectively saving manpower, material resources and financial resources and avoiding the operational risks that may exist in harsh environments.

[0070] Furthermore, the inspection robot and cloud platform provided by this method work together to continuously train, update and optimize the original target detection model based on the confidence mechanism during the inspection process. Each inspection robot can generate a personalized target inspection model for its own work scenario. The cloud platform can promptly notify professionals of abnormal situations to avoid personal injury and property loss.

[0071] Those skilled in the art will understand that the exemplary components, systems, and methods described in conjunction with the embodiments disclosed herein can be implemented in hardware, software, or a combination of both. Whether implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this invention. When implemented in hardware, it can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this invention are programs or code segments used to perform the desired tasks. The programs or code segments can be stored in a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried in a carrier wave.

[0072] It should be clarified that the present invention is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present invention is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of the present invention.

[0073] In this invention, features described and / or illustrated for one embodiment may be used in the same or similar manner in one or more other embodiments, and / or combined with or in place of features of other embodiments.

[0074] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, various modifications and variations of the embodiments of the present invention are possible. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A power inspection method based on a patrol robot, characterized by, The method comprises the following steps: An image data acquisition step, using a visual device integrated on the inspection robot to acquire image data of a target power inspection equipment, and inputting the image data into a pre-trained target detection model built-in the inspection robot; An image data analysis step, using the target detection model to perform calculation and analysis based on the input image data and output equipment operation status information and confidence corresponding to the equipment operation status information; A confidence-based result selection step, in the case that the confidence is not lower than a preset threshold, storing the equipment operation status information and the confidence corresponding to the equipment operation status information to a preset location, and marking that the current power inspection of the target power inspection equipment is completed; in the case that the confidence is lower than the preset threshold, first performing a correction operation for adjusting the image data acquisition condition of the inspection robot, and then re-performing the image data acquisition step, the image data analysis step and the confidence-based result selection step; When the number of times of performing the correction operation reaches a preset maximum correction number, stopping performing the correction operation, and marking that the current power inspection of the target power inspection equipment is completed; A positioning module integrated on the inspection robot sends position information to a cloud platform in real time, for visualizing generation of an inspection path, a current position and a power inspection progress of the inspection robot on the cloud platform, the inspection path being an inspection path artificially preset or modeled generated after identification by the inspection robot in a range to be inspected; The power inspection equipment to be inspected is numbered according to the inspection path, the inspection robot periodically inspects each numbered power inspection equipment in a preset working range, and the inspection progress and the inspection result are displayed in real time on the cloud platform, the inspection progress and the inspection result being distinguished by using preset symbols, colors and words; The step of stopping performing the correction operation and marking that the current power inspection of the target power inspection equipment is completed when the number of times of performing the correction operation reaches the preset maximum correction number comprises: when the number of times of performing the correction operation reaches the preset maximum correction number, stopping performing the correction operation, transmitting image data of the current target power inspection equipment collected in the current power inspection to the cloud platform, receiving equipment operation status information of the current target power inspection equipment based on the transmitted image data and obtained by artificial assessment from the cloud platform, and marking that the current power inspection of the target power inspection equipment is completed; or when the number of times of performing the correction operation reaches the preset maximum correction number, stopping performing the correction operation, and recording power inspection equipment whose confidence of the image data collected in the current power inspection is always lower than the preset threshold, and after the current power inspection of all the power inspection equipment to be inspected is completed, collecting image data of the power inspection equipment whose confidence is still lower than the preset threshold after the maximum correction number is reached, and transmitting the image data to a centralized processing device. The step of transmitting the image data of the current target power inspection equipment collected in the current power inspection to the cloud platform comprises: uploading the image data collected in the last power inspection of the current target power inspection equipment to the cloud platform; or uploading the image data collected at all historical time points in the current power inspection of the current target power inspection equipment to the cloud platform; or uploading the image data with the highest confidence level collected at historical time points in the current power inspection of the current target power inspection equipment to the cloud platform.

2. The power inspection method of claim 1, wherein, The target detection model is trained using a preset number of normal operation images and a plurality of device failure images of the target power inspection equipment as a training set.

3. The electric power inspection method according to claim 1, characterized by, After receiving the device operation status information of the current target power inspection equipment based on the transmitted image data and manually evaluated from the cloud platform, the method further comprises: The visual device uses the received device operation status information of the current target power inspection equipment manually evaluated from the cloud platform as a label, uses the image data collected in the current power inspection of the current target power inspection equipment as a training set, trains and updates the target detection model.

4. The method of claim 1, wherein, After the device operation status information of the current target power inspection equipment based on the transmitted image data and manually evaluated, the method further comprises: The cloud platform uses the device operation status information manually evaluated within a preset historical time length as a label, uses the image data of the power equipment corresponding to the device operation status information as a training set, trains and updates the target detection model.

5. The method of claim 1, wherein, The device operation status information includes device normal and device failure types, and the method further comprises: When the confidence level is not lower than a preset threshold, and the device operation status information indicates a device failure type, the device operation status information indicating the device failure type is sent to the cloud platform, so that the cloud platform generates a warning based on the received device operation status information indicating the device failure type.

6. The method of claim 1, wherein, The types of the correction operation include: adjusting the height, angle, distance position and focusing state of the visual device integrated with the inspection robot, supplementing the light source and cleaning the visual device lens.

7. A patrol robot characterized by comprising: The inspection robot is used to implement the power inspection method of any one of claims 1-6, and the inspection robot comprises: a visual device configured to collect image data of the target power inspection equipment and input the image data into a pre-trained target detection model built in an operation module of the inspection robot; a driving module configured to drive the inspection robot to move to an inspection position; an operation module, wherein a pre-trained target detection model is built in the operation module, and the target detection model is configured to perform calculation and analysis based on the input image data and output device operation status information and a confidence level corresponding to the device operation status information; The confidence-based selection module is configured to store the device operation status information and the confidence to a preset location and mark that the current power inspection of the target power inspection device is completed when the confidence is not lower than a preset threshold; configured to perform a correction operation for adjusting the image data collection condition of the inspection robot first, and then re-perform the image data collection step, the image data analysis step and the confidence-based selection result step when the confidence is lower than the preset threshold; configured to stop performing the correction operation and mark that the current power inspection of the target power inspection device is completed when the number of times of performing the correction operation reaches a preset maximum correction number.

8. A power inspection system based on a patrol robot, characterized by, The power inspection system comprises: An inspection robot according to claim 7; A cloud platform configured to receive image data collected by the inspection robot during power inspection, and obtain device operation status information of a power device artificially evaluated on the image data collected during the power inspection, and transmit the device operation status information of the power device back to the corresponding inspection robot.

Images