An overhaul method and system for power equipment

Abstract

The present invention discloses a maintenance method and system for electric power equipment, the method comprising identifying an electric power equipment maintenance image, determining type data of the current electric power equipment, obtaining corresponding necessary pre-operation data according to each maintenance behavior standard data determined by the type data, performing behavioral recognition on the maintenance action of the maintenance personnel according to the maintenance personnel image, judging the maintenance action according to the recognition result, the maintenance behavior standard data and the corresponding necessary pre-operation data, generating an abnormal warning signal when it is judged that the maintenance action does not meet the preset safety standard, identifying the maintenance personnel feedback information obtained during the current electric power equipment maintenance process, and correcting the abnormal warning signal according to the recognition result. The present invention can timely discover and correct deviations and omissions in the operation process, effectively avoid safety accidents caused by human negligence or improper operation, and significantly improve the safety and reliability of maintenance operations.

Description

Technical Field

[0001] The present invention relates to the technical field of power maintenance, and in particular to a maintenance method and system for power equipment. Background Art

[0002] With the continuous development of modern power maintenance technology, the maintenance and safety inspection of power equipment have received increasing attention and have become an indispensable part of ensuring the safe and stable operation of the power grid.

[0003] The existing power equipment maintenance process has exposed many shortcomings and limitations at the practical operation level. For example, operators often rely on work experience or paper documents to perform a series of complex power equipment maintenance operations. This maintenance method is relatively simple in presenting information, which can easily lead to omissions or misoperations of key steps. Once a necessary operation is skipped or performed improperly, it may not only affect the normal operation of the equipment, but may also cause serious safety accidents, resulting in immeasurable losses.

[0004] It can be seen that how to avoid optimizing the maintenance process of power equipment to improve the maintenance effect has become a technical problem that needs to be urgently solved by those skilled in the art. Summary of the Invention

[0005] The present invention provides a maintenance method and system for electric power equipment, so as to improve the maintenance process of electric power equipment.

[0006] In order to solve the above technical problems, an embodiment of the present invention provides a method for repairing electric power equipment, comprising:

[0007] A scene image during the maintenance process of the electric power equipment is acquired, wherein the scene image includes an image of a maintenance person and an image of the electric power equipment.

[0008] The electric power equipment image is identified to determine the type data of the current electric power equipment.

[0009] Based on each inspection behavior standard data determined by the type data, necessary preliminary operation data corresponding to each inspection behavior standard data is determined.

[0010] The maintenance personnel's maintenance actions are behaviorally identified based on the maintenance personnel image, and the maintenance actions are judged based on the identification results, the maintenance behavior standard data and the corresponding necessary pre-operation data. When it is judged that the maintenance action does not meet the preset safety standards, an abnormal warning signal is generated.

[0011] The maintenance personnel feedback information obtained during the current power equipment maintenance process is identified, and the abnormal warning signal is corrected according to the identification result.

[0012] A matching maintenance log result is generated using the corrected abnormal warning signal, wherein the maintenance log result at least includes a monitoring log result during the maintenance process of the current power equipment.

[0013] Furthermore, the identifying the electric power equipment image and determining the type data of the current electric power equipment includes:

[0014] A first electric power equipment image captured by an image capture device is acquired.

[0015] Denoising and image enhancement processing are performed on the first electric power equipment image to obtain a second electric power equipment image.

[0016] A histogram equalization algorithm is used to perform contrast equalization processing on the second electric power equipment image to obtain a third electric power equipment image.

[0017] Feature extraction is performed on the third power equipment image based on the YOLO algorithm to obtain equipment feature information.

[0018] The current power device in the third power device image is identified based on the device feature information to obtain a device type of the current power device.

[0019] Furthermore, the performing of behavior recognition on the maintenance action of the maintenance personnel based on the maintenance personnel image includes:

[0020] The pre-processed image of the maintenance personnel is input into the OpenPose model for human body recognition to obtain the key point positions of the maintenance personnel.

[0021] The posture information of the maintenance personnel is obtained according to the positions of the key points of the human body.

[0022] An operation action sequence is constructed according to the positions of the key points of the human body in the maintenance personnel images of adjacent frames.

[0023] Behavior recognition is performed on the maintenance actions of the maintenance personnel according to the operation action sequence.

[0024] Furthermore, the performing of behavior recognition on the maintenance actions of the maintenance personnel according to the operation action sequence includes:

[0025] Downsampling is performed on the first operation action sequence to obtain a second operation action sequence.

[0026] The second operation action sequence is input into a pre-built LSTM model to obtain probability distribution information of the behavior category output by the LSTM model.

[0027] The maintenance actions of the maintenance personnel are identified and judged based on the output results of the LSTM model.

[0028] Furthermore, the determining of necessary pre-operation data corresponding to each of the inspection behavior standard data determined by the type data includes:

[0029] According to the type data, corresponding inspection behavior standard data is retrieved from the database.

[0030] According to the directed dependency graph of power maintenance, necessary pre-operation data corresponding to each maintenance operation in the maintenance behavior standard data is obtained.

[0031] Furthermore, judging the maintenance action according to the recognition result, the maintenance behavior standard data and the corresponding necessary pre-operation data includes:

[0032] A maintenance safety standard diagram is constructed according to the type data, the maintenance behavior standard data and the necessary pre-operation data. The maintenance safety standard diagram includes standard behavior feature vectors and dependency relationships of each of the maintenance behavior standard data and each of the necessary pre-operation data.

[0033] Feature extraction is performed on the maintenance action to obtain a maintenance action feature vector of the maintenance personnel.

[0034] Based on the graph attention mechanism, the maintenance action feature vector is compared with the standard behavior feature vector to obtain the behavioral difference between the maintenance action and the safety standard.

[0035] Evaluate whether the maintenance action meets the preset safety standard based on the behavior difference.

[0036] Furthermore, the maintenance personnel feedback information includes voice feedback information.

[0037] Furthermore, the step of identifying the feedback information of the maintenance personnel obtained during the current maintenance process of the power equipment and correcting the abnormal warning signal according to the identification result includes:

[0038] The voice feedback information obtained by the voiceprint collection device is subjected to noise reduction and volume adjustment to obtain voice detection data.

[0039] Obtain the corresponding skip keyword preset for the maintenance action.

[0040] A keyword matching operation is performed on the voice detection data according to the skip keyword. If the match is successful, it is determined that the voice feedback information contains the skip keyword of the maintenance action, and the abnormal warning signal is corrected.

[0041] Furthermore, the maintenance log results also include abnormal warning details, maintenance task allocation and progress, maintenance personnel communication and collaboration records, and equipment historical maintenance records.

[0042] Another embodiment of the present invention provides a maintenance system for electric power equipment, comprising:

[0043] The image acquisition module is used to obtain scene images during the maintenance process of power equipment, and the scene images include maintenance personnel images and power equipment images.

[0044] The device identification module is used to identify the power device image and determine the type data of the current power device.

[0045] The standard data acquisition module is used to determine necessary pre-operation data corresponding to each of the maintenance behavior standard data according to each of the maintenance behavior standard data determined by the type data.

[0046] The signal warning module is used to identify the maintenance actions of the maintenance personnel based on the maintenance personnel image, judge the maintenance actions based on the recognition results, the maintenance behavior standard data and the corresponding necessary pre-operation data, and generate an abnormal warning signal when it is judged that the maintenance action does not meet the preset safety standards.

[0047] The signal correction module is used to identify the feedback information of the maintenance personnel obtained during the current power equipment maintenance process, and to correct the abnormal warning signal according to the identification result.

[0048] The log generation module is used to generate a matching maintenance log result based on the corrected abnormal warning signal, wherein the maintenance log result at least includes a monitoring log result during the maintenance process of the current power equipment.

[0049] Compared with the prior art, the embodiments of the present invention have the following advantages:

[0050] By real-time monitoring of the maintenance personnel's maintenance actions and voice information, and intelligently comparing them with preset safety standards and necessary operating procedures, it is possible to promptly discover and correct deviations and omissions in the operation process, effectively avoiding safety accidents caused by human negligence or improper operation, and significantly improving the safety and reliability of maintenance operations; by integrating intelligent means such as image recognition, voiceprint recognition, and image attention analysis, the maintenance process is standardized and automated, greatly reducing the burden on maintenance personnel and improving work efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 A flowchart of the steps of the method for repairing electric power equipment provided by an embodiment of the present invention;

[0052] Figure 2 This is a structural block diagram of the power equipment maintenance system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0053] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. The purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0054] In the description of this application, the terms "first," "second," "third," etc. are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of the technical features indicated. Therefore, a feature specified as "first," "second," "third," etc. may explicitly or implicitly include one or more of the features. In the description of this application, unless otherwise specified, "plurality" means two or more.

[0055] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installed", "connected" and "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or an indirect connection through an intermediate medium, or it can be a communication between the two components. The terms "vertical", "horizontal", "left", "right", "up", "down" and similar expressions used herein are for illustrative purposes only, and do not indicate or imply that the device or component referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0056] In the description of this application, it should be noted that, unless otherwise defined, all technical and scientific terms used in this application have the same meanings as those commonly understood by those skilled in the art. The terms used in this specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. Those skilled in the art will understand the specific meanings of the above terms in this application in specific circumstances.

[0057] An embodiment of the present invention provides a method for repairing electric power equipment. For details, see Figure 1 , Figure 1 The flowchart of the method for repairing electric power equipment in one embodiment of the present invention includes steps S11 to S16:

[0058] Step S11 : acquiring a scene image during the maintenance process of the power equipment, wherein the scene image includes an image of a maintenance person and an image of the power equipment.

[0059] In order to monitor the maintenance process in real time and ensure the safety of maintenance personnel and the correct operation of equipment, it is necessary to obtain scene images during the maintenance process of power equipment.

[0060] Specifically, this embodiment uses a high-definition camera as an image acquisition device to capture scene images. During the maintenance process, it is necessary to identify the power equipment and the maintenance personnel, so the scene image includes the maintenance personnel image and the power equipment image.

[0061] The acquisition of scene images provides visual data for monitoring the maintenance process, which is used for subsequent image recognition and behavior analysis, improving the visualization and traceability of the maintenance process and reducing human errors.

[0062] Step S12: Identify the image of the electric power equipment and determine the type data of the current electric power equipment.

[0063] Since different types of power equipment have different maintenance standards and safety requirements, determining the equipment type is necessary to match the corresponding necessary maintenance operations. Therefore, it is necessary to identify the type of the current power equipment in the power equipment image.

[0064] The specific recognition process is: obtaining the first power equipment image captured by the image acquisition device. The first power equipment image is used as the original image. The image needs to be preprocessed and checked, specifically checking the image integrity, checking the image quality, etc. The process of checking the image integrity is to confirm that the image has no damage or missing parts, and checking the image quality is to evaluate the clarity, brightness, etc. of the image to ensure that it meets the subsequent processing requirements.

[0065] The image may contain noise, which affects the recognition accuracy. Therefore, the first power equipment image is denoised and image enhancement techniques such as sharpening and contrast adjustment are applied to improve the image quality to obtain a second power equipment image.

[0066] The second power equipment image is resized to a uniform size to meet the input requirements of the YOLO algorithm, ensuring that all input images have a consistent size.

[0067] A histogram equalization algorithm is used to perform contrast equalization on the second power equipment image to obtain a third power equipment image. Histogram equalization is a commonly used image enhancement technique used to improve image contrast, making details in the image more clearly visible. It redistributes the brightness values of pixels in the image to make the image's grayscale histogram more uniform. In some cases, images captured during power maintenance may appear gray and lack contrast due to poor lighting conditions or limitations of the shooting equipment. Histogram equalization can stretch the dynamic range of the image, making bright areas brighter and dark areas darker, thereby significantly improving the visual quality of the image.

[0068] Feature extraction is a key step in identifying the type of power equipment. The YOLO algorithm can simultaneously perform target detection and feature extraction in a single forward pass. When processing images of power equipment, the YOLO algorithm accurately detects the power equipment in the image and extracts key features. This feature information is used in subsequent identification steps. In this embodiment, the YOLO algorithm is used to extract features from the third power equipment image, improving recognition speed and accuracy and obtaining device feature information.

[0069] The current power device in the third power device image is identified based on the device feature information to obtain a device type of the current power device.

[0070] After extracting device feature information, the next step is to identify these features to determine the type of power equipment. This step is typically accomplished using classifiers such as support vector machines (SVMs) and neural networks. During the identification process, the extracted feature information must be matched against a known library of device features to find the most similar device type. Due to the wide variety of power equipment and the unique characteristics of each device, this step requires ensuring the accuracy and robustness of the classifier. Ultimately, the identification results will determine the type of the current power equipment, providing important information for subsequent equipment management and maintenance.

[0071] In step S13, necessary pre-operation data corresponding to each inspection behavior standard data is determined based on each inspection behavior standard data determined by the type data.

[0072] Based on the type data, the corresponding maintenance behavior standard data is retrieved from the database. This step is critical to ensuring the accuracy of the type data and the integrity of the database. In this embodiment, the database used is an internal database within the power system that records maintenance operation information for power equipment. The type data represents information such as the type of power equipment, possible fault types, or maintenance requirements. This information serves as the basis for retrieving the maintenance behavior standard data. The required maintenance behavior standard data for the current power equipment is obtained based on the type data.

[0073] The directed dependency graph of power maintenance is a graphical tool that represents the dependency relationship between maintenance operations. It at least includes each maintenance operation and the necessary prerequisite operations corresponding to each maintenance operation, as well as the dependency, sequence and logical relationship between them.

[0074] A directed dependency graph is used to analyze the various maintenance operations in the maintenance behavior standard data, determine the dependencies between them, and extract the necessary prerequisite operation data corresponding to each maintenance operation based on this.

[0075] Step S14: Perform behavioral recognition on the maintenance personnel's maintenance actions based on the maintenance personnel image, and judge the maintenance actions based on the recognition results, maintenance behavior standard data and corresponding necessary pre-operation data. When it is judged that the maintenance actions do not meet the preset safety standards, an abnormal warning signal is generated.

[0076] As an advanced human pose estimation algorithm, the OpenPose model can accurately identify the positions of key points of the human body in an image, such as joints, head, hands, etc. These key point positions provide key data for subsequent acquisition of pose information and construction of operation action sequences.

[0077] The pre-processed maintenance personnel image is input into the OpenPose model for human body recognition to obtain the positions of the maintenance personnel's human body key points. Through accurate human body key point detection, a reliable data foundation is provided for subsequent steps, which helps to accurately analyze the posture and movements of the maintenance personnel.

[0078] After obtaining the positions of the key points on the human body, we need to use this information to extract the maintenance worker's posture information. This involves analyzing the relative positions, angles, and dynamic changes of the key points to reflect the maintenance worker's overall movement trends and local details. The maintenance worker's posture information is then obtained based on the positions of the key points on the human body.

[0079] To analyze the maintenance worker's continuous movements, it's necessary to connect the key points of the human body in adjacent frames to construct a sequence of actions. This step ensures the consistency and accuracy of the sequence so that the subsequent behavior recognition model can correctly identify the maintenance worker's movements. This sequence provides dynamic data support for subsequent behavior recognition, facilitating accurate analysis of the maintenance worker's continuous movements.

[0080] When processing continuous action sequences, due to the large amount of data and the possible presence of redundant information, the first operation action sequence needs to be downsampled. Downsampling can reduce the amount of data and improve processing efficiency, while retaining key action information to obtain the second operation action sequence. Through downsampling, a more concise and efficient action sequence can be obtained.

[0081] The downsampled second operation action sequence is input into the pre-built LSTM model, which outputs the probability distribution information of the behavior category. The LSTM model can capture the long-term dependencies in the sequence data and is suitable for processing the data of continuous actions of maintenance personnel. Through the application of the LSTM model, the action categories of maintenance personnel can be accurately identified.

[0082] The last step is to identify and judge the maintenance actions of the maintenance personnel based on the output results of the LSTM model. It is necessary to match the probability distribution information output by the model with the preset behavior category to determine the specific action category of the maintenance personnel.

[0083] Integrate type data, maintenance behavior standard data and necessary pre-operation data to construct a maintenance safety standard diagram. The maintenance safety standard diagram not only contains the standard behavior feature vectors of each of the aforementioned maintenance behavior standard data and each of the aforementioned necessary pre-operation data, but also includes the dependency relationships between these behaviors. Constructing the maintenance safety standard diagram provides a clear standard and framework for subsequent maintenance action comparisons, helps standardize the maintenance process, and ensures the objectivity and consistency of the evaluation process.

[0084] In addition, the visual nature of the maintenance safety standards diagram helps maintenance personnel better understand the maintenance safety standards, thereby improving overall safety awareness.

[0085] After obtaining the actual actions of the maintenance personnel, they need to be converted into feature vectors through feature extraction technology. The accuracy of feature extraction directly affects the results of subsequent comparison and evaluation, and provides the necessary data basis for subsequent comparison and evaluation. Through precise feature extraction, quantitative analysis of maintenance actions can be achieved, thereby more objectively evaluating whether they meet safety standards.

[0086] The graph attention mechanism is used to compare the maintenance action feature vector with the standard behavior feature vector to determine the degree of behavioral difference between the maintenance action and the safety standard. The graph attention mechanism is a deep learning technique that automatically learns the correlation weights between nodes and aggregates information based on these weights. During the comparison process, the mechanism dynamically adjusts its focus to more accurately capture the differences between the maintenance action and the standard behavior. This enables detailed analysis of the maintenance action, leading to a more accurate assessment of its differences from the safety standard. This not only improves the accuracy and efficiency of the assessment, but also helps identify potential safety hazards and improper operations.

[0087] After the comparison is completed, it is necessary to evaluate whether the maintenance actions meet the preset safety standards based on the behavioral differences. Behavioral difference assessment and safety standard judgment are the core links of the plan and are directly related to the safety and efficiency of the maintenance work. Through accurate assessment, improper operations can be discovered and corrected in a timely manner, thereby avoiding potential safety risks.

[0088] Step S15: Identify the feedback information from the maintenance personnel obtained during the current maintenance process of the power equipment, and modify the abnormal warning signal according to the identification result.

[0089] Specifically, the maintenance personnel feedback information used in this embodiment is voice feedback information.

[0090] During the maintenance of power equipment, maintenance personnel may provide key information through voice feedback. This information is crucial for identifying maintenance progress, abnormal conditions, and operational recommendations. However, on-site environments often contain various noise interferences, such as equipment operation and personnel conversations, which can affect the clarity of voice feedback information. Therefore, the voice feedback information obtained by the voiceprint collection device needs to be noise-reduced to reduce background noise interference.

[0091] Furthermore, appropriate volume adjustment is required based on the original voice volume to ensure the accuracy and reliability of subsequent processing. Speech detection data obtained through noise reduction and volume adjustment can significantly improve the clarity of voice feedback information, providing an accurate data foundation for subsequent keyword matching and abnormal warning signal correction, helping to reduce false positives and missed negatives.

[0092] In the maintenance process of power equipment, sometimes the monitoring results will be inaccurate. For example, a necessary maintenance action and its necessary pre-operation have been completed but not detected, which may cause a false alarm and an erroneous warning. At this time, the operator can communicate with the voiceprint collection device in the monitoring equipment, such as "I have completed the electrical inspection work", to trigger keyword detection and skip this process.

[0093] Furthermore, during power maintenance, there are routine or known safe operations that don't require real-time system monitoring or early warning. To improve system flexibility and user experience, a series of pre-set skip keywords can be associated with specific maintenance actions. When maintenance personnel complete these operations, if the image recognition system fails to accurately capture or identify them, they can convey to the system that the operation is complete and safe by speaking the corresponding skip keyword. By pre-setting skip keywords, the system can intelligently identify and skip known safe operations, reducing unnecessary warnings and interruptions, and improving the continuity and efficiency of maintenance work.

[0094] After obtaining the pre-processed voice detection data and the preset skip keywords, the system needs to perform a keyword matching operation, and perform a keyword matching operation on the voice detection data according to the skip keywords. If the match is successful, it is determined that the voice feedback information contains the skip keywords of the maintenance action, and the abnormal warning signal is corrected, that is, the abnormal warning signal is canceled.

[0095] Through keyword matching operations, the system can respond to the maintenance personnel's voice commands in real time and accurately, and make intelligent corrections to abnormal warning signals. This not only reduces false alarms and missed alarms caused by misjudgment, improves the accuracy and reliability of maintenance, but also gives maintenance personnel greater initiative and control, enabling them to respond to various situations more flexibly and efficiently during the maintenance process, thereby improving the quality and efficiency of the overall maintenance work.

[0096] Step S16: Generate a matching maintenance log result using the corrected abnormal warning signal, wherein the maintenance log result at least includes a monitoring log result during the maintenance process of the current power equipment.

[0097] The maintenance log results also include abnormal warning details, maintenance task allocation and progress, maintenance personnel communication and collaboration records, and equipment historical maintenance records.

[0098] During the maintenance of power equipment, the accuracy and timeliness of abnormal warning signals are crucial to ensuring equipment safety and reducing downtime.

[0099] After identifying maintenance personnel feedback and correcting abnormality warning signals through the aforementioned steps, the maintenance system obtains more accurate abnormality warning information. Next, these corrected abnormality warning signals can be used as a starting point to generate maintenance log entries. Detailed information about the warning signals, including warning time, warning type, warning level, and associated devices, is accurately recorded in the maintenance log, providing data support for subsequent analysis and decision-making.

[0100] The maintenance method for power equipment of the present invention monitors the maintenance actions and voice information of maintenance personnel in real time, and intelligently compares them with preset safety standards and necessary operating steps. It can timely discover and correct deviations and omissions in the operation process, effectively avoid safety accidents caused by human negligence or improper operation, and significantly improve the safety and reliability of maintenance operations; by integrating intelligent means such as image recognition, voiceprint recognition, and image attention analysis, the maintenance process is standardized and automated, which greatly reduces the burden on maintenance personnel and improves work efficiency.

[0101] The embodiment of the present invention further provides a maintenance device for electric power equipment, which is used to perform the maintenance method for electric power equipment as described above. Figure 2 This is a structural block diagram of a maintenance device for electric power equipment according to an embodiment of the present invention, the device comprising:

[0102] The image acquisition module 21 is used to obtain scene images during the maintenance process of the power equipment, and the scene images include images of maintenance personnel and images of the power equipment.

[0103] The device identification module 22 is used to identify the power device image and determine the type data of the current power device.

[0104] The standard data acquisition module 23 is used to determine necessary pre-operation data corresponding to each of the maintenance behavior standard data according to each of the maintenance behavior standard data determined by the type data.

[0105] The signal warning module 24 is used to identify the maintenance personnel's maintenance actions based on the maintenance personnel image, judge the maintenance actions based on the identification results, the maintenance behavior standard data and the corresponding necessary pre-operation data, and generate an abnormal warning signal when it is judged that the maintenance action does not meet the preset safety standards.

[0106] The signal correction module 25 is used to identify the feedback information of the maintenance personnel obtained during the current power equipment maintenance process, and correct the abnormal warning signal according to the identification result.

[0107] The log generation module 26 is configured to generate a matching maintenance log result based on the corrected abnormal warning signal, wherein the maintenance log result at least includes a monitoring log result during the maintenance process of the current power equipment.

[0108] The technical features and technical effects of the system proposed in the embodiment of the present invention are the same as those of the method proposed in the embodiment of the present invention and are not described in detail here. Each module in the above system can be implemented in whole or in part by software, hardware, or a combination thereof. Each of the above modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software so that the processor can call and execute the operations corresponding to the above modules.

[0109] The above-described embodiments merely illustrate several implementations of the present invention, and while their descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims.

Claims

1. A method for repairing electric power equipment, characterized in that: include: Acquire a scene image during the maintenance process of the power equipment, wherein the scene image includes an image of a maintenance person and an image of the power equipment; Identify the electric power equipment image and determine the type data of the current electric power equipment; determining necessary pre-operation data corresponding to each of the inspection behavior standard data according to each of the inspection behavior standard data determined by the type data; Performing behavioral recognition of the maintenance personnel's maintenance actions based on the maintenance personnel image, judging the maintenance actions based on the recognition results, the maintenance behavior standard data, and the corresponding necessary pre-operation data, and generating an abnormality warning signal when it is judged that the maintenance actions do not meet the preset safety standards; The judgment of the maintenance action based on the recognition result, the maintenance behavior standard data and the corresponding necessary pre-operation data includes: constructing a maintenance safety standard diagram based on the type data, the maintenance behavior standard data and the necessary pre-operation data, the maintenance safety standard diagram including standard behavior feature vectors and dependency relationships of each maintenance behavior standard data and each necessary pre-operation data; extracting features of the maintenance action to obtain a maintenance action feature vector of the maintenance personnel; comparing the maintenance action feature vector with the standard behavior feature vector based on a graph attention mechanism to obtain a behavioral difference between the maintenance action and the safety standard; and evaluating whether the maintenance action meets the preset safety standard based on the behavioral difference. Identifying feedback information from maintenance personnel obtained during the current power equipment maintenance process, and correcting the abnormal warning signal based on the identification result; A matching maintenance log result is generated using the corrected abnormal warning signal, wherein the maintenance log result at least includes a monitoring log result during the maintenance process of the current power equipment.

2. The method for repairing electric power equipment according to claim 1, wherein: The identifying the electric power equipment image and determining the type data of the current electric power equipment includes: Acquire a first electric power equipment image acquired by an image acquisition device; performing denoising and image enhancement processing on the first electric power equipment image to obtain a second electric power equipment image; performing contrast equalization processing on the second power equipment image using a histogram equalization algorithm to obtain a third power equipment image; Performing feature extraction on the third power equipment image based on the YOLO algorithm to obtain equipment feature information; The current power device in the third power device image is identified based on the device feature information to obtain a device type of the current power device.

3. The method for repairing electric power equipment according to claim 1, wherein: The performing behavior recognition on the maintenance action of the maintenance personnel according to the maintenance personnel image includes: Input the pre-processed image of the maintenance personnel into the OpenPose model for human body recognition to obtain the key point positions of the maintenance personnel; Acquiring posture information of the maintenance personnel according to the positions of the key points of the human body; Constructing an operation action sequence according to the positions of the key points of the human body in the maintenance personnel images of adjacent frames; Behavior recognition is performed on the maintenance actions of the maintenance personnel according to the operation action sequence.

4. The method for repairing electric power equipment according to claim 3, wherein: The performing of behavior recognition on the maintenance action of the maintenance personnel according to the operation action sequence includes: Downsampling the first operation action sequence to obtain a second operation action sequence; Inputting the second operation action sequence into a pre-built LSTM model to obtain probability distribution information of the behavior category output by the LSTM model; The maintenance actions of the maintenance personnel are identified and judged based on the output results of the LSTM model.

5. The method for repairing electric power equipment according to claim 1, wherein: The determining of necessary pre-operation data corresponding to each of the inspection behavior standard data determined by the type data includes: Retrieving corresponding inspection behavior standard data from a database according to the type data; According to the directed dependency graph of power maintenance, necessary pre-operation data corresponding to each maintenance operation in the maintenance behavior standard data is obtained.

6. The method for repairing electric power equipment according to claim 1, wherein: The maintenance personnel feedback information includes voice feedback information.

7. The method for repairing electric power equipment according to claim 6, wherein: The identifying of the maintenance personnel feedback information obtained during the current power equipment maintenance process and correcting the abnormal warning signal according to the identification result includes: Performing noise reduction and volume adjustment on the voice feedback information obtained by the voiceprint collection device to obtain voice detection data; Obtaining a corresponding skip keyword preset for the maintenance action; A keyword matching operation is performed on the voice detection data according to the skip keyword. If the match is successful, it is determined that the voice feedback information contains the skip keyword of the maintenance action, and the abnormal warning signal is corrected.

8. The method for repairing electric power equipment according to claim 1, wherein: The maintenance log results also include abnormal warning details, maintenance task allocation and progress, maintenance personnel communication and collaboration records, and equipment historical maintenance records.

9. A maintenance system for electric power equipment, characterized in that: include: An image acquisition module is used to obtain scene images during the maintenance process of power equipment, wherein the scene images include images of maintenance personnel and images of power equipment; An equipment identification module, configured to identify the electric equipment image and determine the type data of the current electric equipment; a standard data acquisition module, configured to determine necessary pre-operation data corresponding to each of the maintenance behavior standard data according to each of the maintenance behavior standard data determined by the type data; a signal warning module, configured to identify the maintenance personnel's maintenance actions based on the maintenance personnel image, judge the maintenance actions based on the identification result, the maintenance behavior standard data, and the corresponding necessary pre-operation data, and generate an abnormality warning signal when it is determined that the maintenance actions do not meet the preset safety standards; The judgment of the maintenance action based on the recognition result, the maintenance behavior standard data and the corresponding necessary pre-operation data includes: constructing a maintenance safety standard diagram based on the type data, the maintenance behavior standard data and the necessary pre-operation data, the maintenance safety standard diagram including standard behavior feature vectors and dependency relationships of each maintenance behavior standard data and each necessary pre-operation data; extracting features of the maintenance action to obtain a maintenance action feature vector of the maintenance personnel; comparing the maintenance action feature vector with the standard behavior feature vector based on a graph attention mechanism to obtain a behavioral difference between the maintenance action and the safety standard; and evaluating whether the maintenance action meets the preset safety standard based on the behavioral difference. A signal correction module is used to identify the feedback information of the maintenance personnel obtained during the current power equipment maintenance process, and to correct the abnormal warning signal according to the identification result; The log generation module is used to generate a matching maintenance log result based on the corrected abnormal warning signal, wherein the maintenance log result at least includes a monitoring log result during the maintenance process of the current power equipment.

Images