Electric power operator behavior monitoring method and system based on AI vision

Abstract

The invention discloses a power worker behavior monitoring method and system based on AI vision, and the method comprises the steps: collecting and preprocessing continuous real-shot images of a power worker in a working process, obtaining a preprocessed image, determining the limb features of a human body, and carrying out the prediction through a preset AI model and the limb features, thereby obtaining an action prediction image; determining motion characteristics of a human body in the motion prediction image to determine a motion type; based on the working scene and the action type in the preprocessed image, evaluating the behavior of the power operator to obtain a behavior safety evaluation value; and determining a state risk level based on the behavior safety evaluation value, and sending out a corresponding safety early warning prompt. According to the invention, the AI vision is applied to behavior monitoring of the electric power operation personnel, potential dangerous behaviors or unsafe states can be detected in time, corresponding safety early warning prompts can be given out, and the working personnel can be reminded to take measures in time to avoid accidents, so that the accident risk is effectively reduced, and the personnel safety is guaranteed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of personnel management, in particular to an AI vision-based power operation personnel behavior monitoring method and system. BACKGROUND

[0002] With the continuous development and application of artificial intelligence (AI) technology, its application in various fields is also increasingly widespread. In the power industry, the safety of power operation personnel is crucial, because power operation involves dangerous environments such as high-voltage power lines and high-altitude operations, and once an accident occurs, it can cause serious injury or even life-threatening.

[0003] However, traditional methods usually rely on manual inspection or monitoring, which has the risk of subjective judgment, and may result in inaccurate or missed monitoring results due to human factors. Moreover, traditional methods cannot achieve 24 / 7 all-weather monitoring, have slow monitoring response speed, cannot timely discover potential safety risks, and lack intelligent analysis and prediction capabilities, making it difficult to accurately predict potential risks. SUMMARY

[0004] To solve the above technical problems, the present application provides an AI vision-based power operation personnel behavior monitoring method and system, comprising: Collecting continuous real-time images of power operation personnel during work, and pre-processing the continuous real-time images to obtain pre-processed images; Determining the limb features of the human body in the pre-processed images, and predicting human body movements based on the limb features through a pre-set AI behavior prediction model to obtain movement prediction images; Determining the movement features of the human body in the movement prediction images, and analyzing the movement features to determine the movement type corresponding to the movement features; Determining the work scene in the pre-processed images, and performing safety assessment on the behavior of the power operation personnel based on the work scene and the movement type to obtain a behavior safety assessment value; Determining the state risk level of the power operation personnel based on the behavior safety assessment value, and issuing a corresponding safety warning prompt based on the state risk level.

[0005] Further, the collecting of continuous real-time images of power operation personnel during work, and the pre-processing of the continuous real-time images to obtain pre-processed images, comprises: Collecting continuous real-time images of power operation personnel during work, and pre-processing the continuous real-time images to obtain pre-processed images, the pre-processing including denoising, image enhancement, distortion correction, and resolution standardization.

[0006] Further, the limb features of the human body in the preprocessed image are determined, and human body action prediction is performed based on the limb features through a preset AI behavior prediction model to obtain an action prediction image, including: An edge detection algorithm is used to perform edge detection on the preprocessed image, and the preprocessed image is divided into multiple object contours according to the edge detection result; A human body recognition algorithm is used to recognize the human body from the multiple object contours, and a key point detection algorithm is used to locate the limb features of the human body; The limb features are input into the preset AI behavior prediction model, human body action prediction is performed by the preset AI behavior prediction model to output, and an action prediction image of the human body is obtained.

[0007] Further, the action features of the human body in the action prediction image are determined, and the action features are analyzed to determine the action type corresponding to the action features, including: A posture analysis algorithm is used to analyze the human body posture in each action prediction image to determine the human body posture, and a limb movement trajectory of the human body is constructed based on the human body postures in the continuous multiple action prediction images; The action features of the human body are extracted from the limb movement trajectory of the human body, and the action features of the human body are compared and analyzed with preset action types in an action feature library, and the preset action type consistent with the action features is determined as the action type corresponding to the action features.

[0008] Further, the working scene in the preprocessed image is determined, and the safety of the electric power worker behavior is evaluated based on the working scene and the action type to obtain a behavior safety evaluation value, including: The working environment in which the electric power worker is located is recognized from the multiple object contours of the preprocessed image, and the working scene of the electric power worker is determined according to the working environment; It is judged whether the electric power worker is performing work operation according to the action type, and after it is judged that the electric power worker is performing work operation, the scene type of the working scene is determined; The historical data corresponding to the scene type and the action type are respectively determined, and the safety of the scene type and the action type is respectively evaluated based on the historical data to obtain the weight of the scene type and the action type; The behavior safety evaluation value of the electric power worker is obtained based on the evaluation values and the weights of the scene type and the action type.

[0009] Further, the historical data corresponding to the scene type and the action type are respectively determined, and the safety of the scene type and the action type is respectively evaluated based on the historical data to obtain the weight of the scene type and the action type, including: Determine the historical data corresponding to the scene type and the action type respectively, and determine the accident severity and the accident frequency under the scene type and the action type from the corresponding historical data respectively; Perform safety evaluation on the accident severity under the scene type and the action type respectively to obtain the scene safety evaluation value of the scene type and the action safety evaluation value of the action type; Perform normalization processing on the accident frequency under the scene type and the action type respectively to obtain the weight of the scene type and the weight of the action type.

[0010] Further, the calculation formula of the behavior safety evaluation value is: F=a*X+b*Y, Wherein, F is the behavior safety evaluation value, a is the weight of the scene type, M is the scene safety evaluation value of the scene type, b is the weight of the action type, and N is the action safety evaluation value of the action type.

[0011] Further, the predicted behavior safety evaluation value determines the state risk level of the power operation personnel, comprising: Pre-set state risk level-behavior safety evaluation value interval corresponding relationship, and the state risk level-behavior safety evaluation value interval corresponding relationship is associated with the corresponding state risk level for each change interval; Obtain the behavior safety evaluation value of the power operation personnel, and based on the mapping relationship of the behavior safety evaluation value interval in the state risk level-behavior safety evaluation value interval corresponding relationship, select the state risk level corresponding to the behavior safety evaluation value interval as the state risk level of the power operation personnel.

[0012] Further, the corresponding safety warning prompt is issued based on the state risk level, comprising: Determine a plurality of preset level ranges, and determine the preset level range where the state risk level of the power operation personnel is located; According to the preset level range where the state risk level of the power operation personnel is located, set the corresponding safety warning prompt; If the state risk level of the power operation personnel is in the first preset level range, a low-risk safety warning prompt is issued; If the state risk level of the power operation personnel is in the second preset level range, a medium-risk safety warning prompt is issued; If the state risk level of the power operation personnel is in the third preset level range, a high-risk safety warning prompt is issued.

[0013] The application also provides an AI vision-based power operation personnel behavior monitoring system, comprising: The acquisition module is configured to collect continuous real-shot images of the power operation personnel in the working process, and to pre-process the continuous real-shot images to obtain pre-processed images. The prediction module is configured to determine limb features of the human body in the pre-processed images, and to perform human action prediction based on the limb features through a preset AI behavior prediction model to obtain an action prediction image. The analysis module is configured to determine action features of the human body in the action prediction image, and to analyze the action features to determine an action type corresponding to the action features. The evaluation module is configured to determine a working scene in the pre-processed image, and to perform safety evaluation on the behavior of the power operation personnel based on the working scene and the action type to obtain a behavior safety evaluation value. The early warning module is configured to determine a state risk level of the power operation personnel based on the behavior safety evaluation value, and to issue a corresponding safety early warning prompt based on the state risk level.

[0014] Compared with the prior art, the power operation personnel behavior monitoring method and system based on AI vision has the following advantages: The real-time image data acquisition and processing can realize real-time monitoring and recognition of the power operation personnel in the working process, and provide timely data support for safety management. The computer vision technology can accurately extract the limb features of the human body in the image, and provide basic data for subsequent action prediction and analysis. The AI behavior prediction model can effectively predict the actions of the power operation personnel, help monitor the behavior of the personnel, and identify potential safety risks. Through action feature analysis and prediction, the action type of the power operation personnel can be accurately identified, so that their behavior patterns in the working process can be better understood. The image recognition technology is used to automatically identify the working scene, which helps to better understand the working environment of the power operation personnel and provides more information for safety evaluation. The behavior safety evaluation is performed in combination with the working scene and the action type, which helps to quickly and accurately evaluate the safety behavior level of the power operation personnel and timely discover potential risks. The state risk level of the power operation personnel is determined based on the behavior safety evaluation value, which helps the management personnel better understand the safety risk degree faced by the staff. According to the state risk level, the corresponding safety early warning prompt is issued to remind the staff to pay attention to safety and take timely measures to reduce the risk, thereby reducing the possibility of accidents. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1is a flow structure schematic diagram of an AI vision-based power worker behavior monitoring method in an embodiment of the present application. Figure 2 is a composition schematic diagram of an AI vision-based power worker behavior monitoring system in an embodiment of the present application. DETAILED DESCRIPTION

[0016] The specific embodiments of the present application will be further described in conjunction with the accompanying drawings and embodiments. The following embodiments are used to illustrate the present application, but not to limit the scope of the present application.

[0017] In the description of the present application, it should be understood that the terms "center", "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the platform or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application.

[0018] The terms "first", "second" are only for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the technical features indicated. Therefore, the features defined with "first", "second" can explicitly or implicitly include one or more of the features. In the description of the present application, unless otherwise specified, the meaning of "a plurality of" is two or more.

[0019] In the description of the present application, it should be noted that unless otherwise specified and limited, the terms "mounting", "connecting", "connection" should be understood in a broad sense, for example, it can be fixed connection, or detachable connection, or integral connection; it can be mechanical connection, or electrical connection; it can be directly connected, or indirectly connected through an intermediate medium, or the communication inside two elements. For ordinary skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0020] As Figure 1As shown, in the embodiments of the present application, an AI vision-based power operation personnel behavior monitoring method is provided, which comprises: S100: collecting continuous real shot images of power operation personnel in the working process, and pre-processing the continuous real shot images to obtain pre-processed images; S200: determining the limb features of the human body in the pre-processed images, and predicting the human body action based on the limb features through a pre-set AI behavior prediction model to obtain an action prediction image; S300: determining the action features of the human body in the action prediction image, and analyzing the action features to determine the action type corresponding to the action features; S400: determining the working scene in the pre-processed image, and performing safety evaluation on the behavior of the power operation personnel based on the working scene and the action type to obtain a behavior safety evaluation value; S500: determining the state risk level of the power operation personnel based on the behavior safety evaluation value, and issuing a corresponding safety warning prompt based on the state risk level.

[0021] Further, the present application can realize real-time monitoring and identification of power operation personnel in the working process through real-time image data acquisition and processing, and provide timely data support for safety management; the present application can accurately extract the limb features of the human body in the image by using computer vision technology, and provide basic data for subsequent action prediction and analysis; the present application can effectively predict the action of the power operation personnel by establishing an AI behavior prediction model, help monitor the behavior of the personnel and identify potential safety risks; the present application can accurately identify the action type of the power operation personnel through action feature analysis and prediction, so as to better understand their behavior patterns in the working process; the present application can automatically identify the working scene by using image recognition technology, which helps to better understand the working environment of the power operation personnel and provide more information for safety evaluation; the present application combines the working scene and the action type to perform behavior safety evaluation, which helps to quickly and accurately evaluate the safety behavior level of the power operation personnel and timely discover potential risks; the present application determines the state risk level of the power operation personnel based on the behavior safety evaluation value, which helps the management personnel better understand the safety risk degree faced by the staff; the present application issues a corresponding safety warning prompt according to the state risk level, reminding the staff to pay attention to safety and timely take measures to reduce risks, thereby reducing the possibility of accidents.

[0022] In the embodiments of the present application, an AI vision-based power operation personnel behavior monitoring method is provided, which collects continuous real shot images of power operation personnel in the working process, and pre-processes the continuous real shot images to obtain pre-processed images, and the pre-processing includes denoising, image enhancement, distortion correction and resolution standardization.

[0023] Specifically, collecting continuous real-time images of power operation personnel during work at the power operation site and preprocessing these images are to improve the accuracy and efficiency of subsequent image analysis and processing. The preprocessing process includes steps such as denoising, image enhancement, distortion correction, and resolution standardization. Removing noise from the image is the first step in preprocessing. Noise may come from image acquisition equipment, interference during transmission, and other factors. Denoising helps improve image quality and reduce errors in subsequent processing. Image enhancement aims to improve the visual quality of the image, making it clearer and easier to analyze. During image acquisition, due to factors such as camera lens distortion, images may have distortion problems such as distortion and stretching. Distortion correction can repair these distortions, making the image more consistent with the real scene. Standardizing the resolution of the image ensures that all images have similar pixel density and dimension, which helps maintain consistency and simplifies subsequent image processing and analysis. This step improves image quality through denoising and image enhancement, making subsequent analysis more accurate and reliable. Preprocessing helps reduce noise and distortion in images, reducing errors in subsequent processing and improving algorithm accuracy. Image enhancement can improve the visual effect of the image, making it clearer and easier to observe and analyze. Resolution standardization ensures that all images have similar features, which is beneficial for subsequent algorithms to process images consistently. In summary, preprocessing continuous real-time images of power operation personnel during work can improve image quality, reduce errors, enhance visualization, and unify image features, providing a better foundation for subsequent image analysis and processing, thereby improving the accuracy and efficiency of power operation personnel behavior monitoring and safety warning systems.

[0024] In an embodiment of the present application, an AI vision-based power operation personnel behavior monitoring method is provided, which determines the limb features of the human body in the preprocessed image and predicts the human body action based on the limb features through a preset AI behavior prediction model to obtain an action prediction image, including: performing edge detection on the preprocessed image through an edge detection algorithm, and dividing the preprocessed image into multiple object contours according to the edge detection result; identifying the human body from the multiple object contours through a human body recognition algorithm, and positioning the limb features of the human body through a key point detection algorithm; inputting the limb features into the preset AI behavior prediction model, and outputting the human body action prediction through the preset AI behavior prediction model to obtain the action prediction image of the human body.

[0025] Specifically, the pre-processed image is subjected to edge detection by an edge detection algorithm, and the pre-processed image is divided into multiple object contours according to the result of edge detection. The edge detection algorithm is used to detect edge information in the image, helps to determine the contour of the object, can effectively extract the edge information of the object, and is helpful for subsequent object recognition and positioning process; the human body is recognized from the multiple object contours by a human body recognition algorithm, and the limb features of the human body are positioned by a key point detection algorithm. The human body recognition algorithm can accurately recognize the human body contour in the image, and provide accurate positioning for subsequent limb feature detection. The key point detection algorithm is used to position the key parts of the human body, such as the head, arms, legs, etc., to extract the limb features. Through key point detection, the key parts of the human body can be accurately positioned, and rich limb feature information can be obtained, which provides important data support for subsequent action prediction. The AI behavior prediction model is used to predict the human action according to the input limb feature data. The AI behavior prediction model can accurately predict the human action by learning and analyzing the limb feature data, and provide important information for the monitoring and early warning system. In summary, this step combines edge detection, human body recognition, key point detection and AI behavior prediction technologies, and realizes the conversion from the original image to the human action prediction image through a series of processing and analysis of the pre-processed image. The object contour can be extracted, the human body can be accurately positioned, the limb features can be extracted, the human action can be accurately predicted, and strong technical support can be provided for the behavior monitoring and safety warning system of electric power operating personnel, which helps to realize accurate monitoring and early warning of the behavior of electric power operating personnel.

[0026] In the embodiment of the present application, an AI vision-based electric power operating personnel behavior monitoring method is provided, which determines the action features of the human body in the action prediction image, analyzes the action features, determines the action type corresponding to the action features, comprising: analyzing the human body posture in each action prediction image by a posture analysis algorithm, determining the human body posture, and constructing the limb movement trajectory of the human body based on the human body posture in the continuous multiple action prediction images; extracting the action features of the human body from the limb movement trajectory of the human body, and comparing and analyzing the action features of the human body with the preset action types in the action feature library, and determining the preset action type consistent with the action features as the action type corresponding to the action features.

[0027] Specifically, the posture analysis algorithm is used to analyze the posture of the human body in the action prediction image, determine the posture information of the human body, and through posture analysis, the posture of the human body can be accurately identified, providing basic data for subsequent limb movement trajectory construction; based on the posture information in the continuous multiple action prediction images, the limb movement trajectory of the human body is constructed, the movement path of the human body in space is described, and through the construction of the limb movement trajectory, the movement process of the human body can be recorded comprehensively, providing detailed data support for subsequent action feature extraction and analysis; the action features of the human body are extracted from the limb movement trajectory, the characteristics of the human body action are described, and the action feature extraction helps to describe the movement characteristics of the human body, providing specific data for subsequent action type comparison and analysis; the data extracted from the action features are compared and analyzed with the preset action types in the action feature library, and the actual action type of the human body is determined, and through comparison and analysis, the actual action type of the human body can be matched with the preset action type, realizing accurate classification and recognition of the human body action. In summary, this step combines posture analysis, limb movement trajectory construction, action feature extraction and comparison analysis, etc. Through the analysis of the posture and limb movement trajectory in multiple action prediction images, the feature extraction and type recognition of the human body action are realized, the human body posture can be accurately identified, the limb movement trajectory can be constructed, the action features can be extracted, and the action types can be compared and analyzed, etc. It provides a deeper action analysis and recognition ability for the behavior monitoring and safety warning system of electric power workers, which helps to realize accurate monitoring and warning of actions.

[0028] In the embodiment of the present application, an AI vision-based electric power worker behavior monitoring method is provided, which determines the working scene in the preprocessed image, performs safety evaluation on the behavior of the electric power worker based on the working scene and the action type, and obtains a behavior safety evaluation value, including: identifying the working environment in which the electric power worker is located from the multiple object contours of the preprocessed image, and determining the working scene of the electric power worker according to the working environment; determining whether the electric power worker is performing work operation according to the action type, and determining the scene type of the working scene after determining that the electric power worker is performing work operation; respectively determining the historical data corresponding to the scene type and the action type, and respectively performing safety evaluation on the scene type and the action type based on the historical data and determining the weights of the scene type and the action type; calculating based on the evaluation values and weights of the scene type and the action type to obtain the behavior safety evaluation value of the electric power worker.

[0029] Specifically, the working environment in which the power worker is located is identified from the plurality of object contours of the preprocessed image, i.e., the work site and the surrounding environment are determined, and by identifying the working environment, necessary background information and context can be provided for subsequent work scene determination and safety evaluation; the working scene of the power worker is determined according to the working environment, i.e., the current working state and environment of the power worker are described, and determining the working scene helps to more accurately analyze and evaluate the behavior of the power worker, and provides basic data for subsequent safety evaluation; whether the power worker is performing work operation is determined according to the action type, i.e., the specific behavior of the power worker is identified, and determining the work operation can help to determine the behavior state of the power worker, and provide important clues for subsequent safety evaluation; based on the historical data, the scene type and the action type are evaluated for safety, the weights of the scene type and the action type are determined, and the behavior safety evaluation value of the power worker is calculated, and through safety evaluation, the behavior safety of the power worker can be quantitatively evaluated, potential safety risks can be found and prevented in time, and the safety of the power worker is ensured. In summary, this step combines working environment recognition, scene determination, work operation judgment and safety evaluation technologies, and through comprehensive analysis of the behavior and environment of the power worker, the behavior safety of the power worker is evaluated, the working environment, the working scene, the work operation and the safety evaluation can be accurately identified, which provides a scientific basis and decision support for the safety management and monitoring of the power worker, and helps to improve the safety production consciousness and working environment safety level of the power worker.

[0030] In the embodiment of the present application, an AI vision-based power worker behavior monitoring method is provided, which respectively determines the historical data corresponding to the scene type and the action type, and respectively evaluates the scene type and the action type for safety based on the historical data and determines the weights of the scene type and the action type, comprising: respectively determining the historical data corresponding to the scene type and the action type, and respectively determining the accident severity and the accident occurrence frequency under the scene type and the action type from the corresponding historical data; respectively evaluating the accident severity under the scene type and the action type for safety to obtain the scene safety evaluation value of the scene type and the action safety evaluation value of the action type; respectively normalizing the accident occurrence frequency under the scene type and the action type to obtain the weight of the scene type and the weight of the action type.

[0031] Specifically, the historical data corresponding to the scene type and the action type are determined respectively, the accident conditions related to different scenes and actions, including the accident severity and the accident frequency, are recorded, through the analysis of the historical data, the specific conditions of the accidents under different scenes and action types can be understood, which provides data support for subsequent safety evaluation; according to the accident conditions recorded in the historical data, the accident severity under the scene type and the action type is evaluated, the important parameters of safety evaluation are determined, the accident severity evaluation is helpful to understand the severity of the accidents under different scenes and action types, which provides an important reference basis for safety evaluation; the accident frequency recorded in the historical data is analyzed, the accident frequency conditions under different scenes and action types are understood, the accident frequency analysis can help to determine the probability of the accidents under different scenes and action types, which provides data support for safety evaluation; the scene type and the action type are evaluated according to the accident severity and the accident frequency, the safety evaluation values of the scene type and the action type are obtained, the accident frequency is normalized to obtain the weight of the scene type and the action type, through the safety evaluation value and the weight calculation, the factors such as the accident severity and the accident frequency can be considered comprehensively, which provides comprehensive evaluation for the safety evaluation of the scene type and the action type, and is helpful to determine the behavior safety evaluation value and the behavior weight. In summary, this step combines the historical data analysis, the accident severity evaluation, the accident frequency analysis, the safety evaluation value and the weight calculation, through the in-depth analysis and evaluation of the historical data, the safety evaluation of the scene type and the action type is realized, the accident severity can be evaluated, the accident frequency can be analyzed, the safety evaluation value and the weight calculation can be determined, which provides a scientific basis and decision support for the evaluation of the behavior safety of the electric power operating personnel, and is helpful to improve the safety awareness and behavior standard of the electric power operating personnel.

[0032] In the embodiment of the application, an AI vision-based electric power operating personnel behavior monitoring method is provided, and the calculation formula of the behavior safety evaluation value is: F=a*X+b*Y, Wherein, F is the behavior safety evaluation value, a is the weight of the scene type, M is the scene safety evaluation value of the scene type, b is the weight of the action type, and N is the action safety evaluation value of the action type.

[0033] In the embodiments of the present application, an AI vision-based power operation personnel behavior monitoring method is provided, and the predicted behavior safety evaluation value determines the state risk level of the power operation personnel, including: pre-setting a state risk level-behavior safety evaluation value interval correspondence relationship, and each change interval is associated with a corresponding state risk level; obtaining the behavior safety evaluation value of the power operation personnel, and based on the mapping relationship of the behavior safety evaluation value interval of the behavior safety evaluation value in the state risk level-behavior safety evaluation value interval correspondence relationship, selecting the state risk level corresponding to the behavior safety evaluation value interval as the state risk level of the power operation personnel.

[0034] Specifically, the correspondence relationship between different state risk levels and behavior safety evaluation value intervals is pre-set, so as to determine the state risk level of the power operation personnel according to the behavior safety evaluation value of the power operation personnel. By setting the correspondence relationship, the behavior safety evaluation value can be converted into the state risk level, so that the safety condition of the power operation personnel can be more intuitively understood. According to the behavior safety evaluation value of the power operation personnel and the pre-set state risk level-behavior safety evaluation value interval correspondence relationship, the behavior safety evaluation value is mapped to the corresponding state risk level. Through the mapping relationship, the abstract behavior safety evaluation value can be converted into the specific state risk level, which provides a practical reference basis for safety management and decision-making. In summary, the behavior safety evaluation value of the power operation personnel is mapped to the state risk level, which realizes the intuitive description and evaluation of the safety condition. The setting of the correspondence relationship and the mapping relationship between the state risk level and the behavior safety evaluation value provides an intuitive judgment standard for the safety state evaluation of the power operation personnel, which helps to discover potential safety risks in time and take corresponding safety measures, can effectively convert the abstract evaluation value into the actual risk level, and improves the operability and effectiveness of safety management.

[0035] In the embodiments of the present application, an AI vision-based power operation personnel behavior monitoring method is provided, and the corresponding safety warning prompt is issued based on the state risk level, including: determining a plurality of preset level ranges, and determining the preset level range in which the state risk level of the power operation personnel is located; setting the corresponding safety warning prompt according to the preset level range in which the state risk level of the power operation personnel is located; if the state risk level of the power operation personnel is in the first preset level range, a low-risk safety warning prompt is issued; if the state risk level of the power operation personnel is in the second preset level range, a medium-risk safety warning prompt is issued; and if the state risk level of the power operation personnel is in the third preset level range, a high-risk safety warning prompt is issued.

[0036] Specifically, a plurality of preset level ranges are determined to divide the state risk level of the power operation personnel into different levels so as to classify and give early warning prompts according to the state risk level of the power operation personnel. By setting different level ranges, the safety status of the power operation personnel can be more clearly understood, and a basis is provided for subsequent safety early warning prompts. According to the preset level range in which the state risk level of the power operation personnel is located, corresponding safety early warning prompts are set, including safety prompts for low risk, medium risk and high risk. By setting safety early warning prompts, corresponding safety prompts can be issued in a timely manner according to the state risk level of the power operation personnel, reminding relevant personnel to take necessary safety measures. According to the preset level range in which the state risk level of the power operation personnel is located, the safety early warning level is divided into low risk, medium risk and high risk, so as to issue corresponding safety prompts according to the risk degree. By dividing different safety early warning levels, safety risks can be classified and handled according to actual conditions, which helps to improve the efficiency and pertinence of safety management. In summary, by setting a plurality of preset level ranges and corresponding safety early warning prompts, different levels of safety prompts are issued in a timely manner according to the state risk level of the power operation personnel, which helps to improve the timeliness and effectiveness of safety management. The setting of the preset level range, the setting of the safety early warning prompt and the division of the safety early warning level provide more detailed and comprehensive control for the safety management of the power operation personnel, helping to reduce potential safety risks and protect the safety of the workers.

[0037] As shown in Figure 2 In the embodiment of the present application, an AI vision-based power operation personnel behavior monitoring system is provided, which comprises: an acquisition module for collecting continuous real image of power operation personnel in the working process, and pre-processing the continuous real image to obtain a pre-processed image; a prediction module for determining the limb features of the human body in the pre-processed image, and performing human action prediction based on the limb features through a preset AI behavior prediction model to obtain an action prediction image; an analysis module for determining the action features of the human body in the action prediction image, and analyzing the action features to determine the action type corresponding to the action features; an evaluation module for determining the working scene in the pre-processed image, and performing safety evaluation on the power operation personnel behavior based on the working scene and the action type to obtain a behavior safety evaluation value; and an early warning module for determining the state risk level of the power operation personnel based on the behavior safety evaluation value, and issuing corresponding safety early warning prompts based on the state risk level.

[0038] In summary, the embodiment of the present application provides an AI vision-based power worker behavior monitoring method and system, which comprises: collecting and preprocessing continuous real image of power workers in the working process to obtain a preprocessed image, determining the limb features of the human body, and predicting the action prediction image through the preset AI model and the limb features; determining the action features of the human body in the action prediction image to determine the action type; evaluating the behavior of the power worker based on the working scene in the preprocessed image and the action type to obtain a behavior safety evaluation value; determining the state risk level based on the behavior safety evaluation value, and issuing a corresponding safety warning prompt. The present application applies AI vision to power worker behavior monitoring, can timely detect potential dangerous behavior or unsafe state, and issue corresponding safety warning prompts to remind workers to take measures in time to avoid accidents, thereby effectively reducing the risk of accidents and protecting personnel safety.

[0039] Finally, it should be noted that: obviously, those skilled in the art can make various modifications and variations to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and its equivalent technology, the present application also intends to include these modifications and variations.

[0040] The above only describes one embodiment of the present application, but cannot limit the scope of the present application, and any structural changes made according to the present application should be considered to fall within the scope of the present application and be restricted. Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the platform and the related description described above can refer to the corresponding process in the foregoing platform embodiment, which will not be described here.

[0041] The term "comprising" or any other similar word is intended to cover non-exclusive inclusion, so that the process, platform, article or device / platform including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or includes the elements inherent to these processes, platforms, articles or devices / platforms.

[0042] So far, the technical solutions of the present application have been described in combination with the further embodiments shown in the drawings, but those skilled in the art can easily understand that the protection scope of the present application is obviously not limited to these specific embodiments. Without deviating from the principles of the present application, those skilled in the art can make equivalent changes or replacements to closely related technical features, and the technical solutions after these changes or replacements will fall within the protection scope of the present application.

[0043] The above only describes the preferred embodiments of the present application, and is not intended to limit the protection scope of the present application.

Claims

1. An AI vision-based power operator behavior monitoring method, characterized in that, The method comprises the following steps: Collecting continuous real-time images of power operation personnel during work, and pre-processing the continuous real-time images to obtain pre-processed images; Determine the limb features of the human body in the pre-processed image, and predict the human body action based on the limb features through the preset AI behavior prediction model to obtain the action prediction image; Determine the action features of the human body in the action prediction image, and analyze the action features to determine the action type corresponding to the action features; Determine the working scene in the pre-processed image, and evaluate the safety of the power operation personnel behavior based on the working scene and the action type to obtain the behavior safety evaluation value; Determine the state risk level of the power operation personnel based on the behavior safety evaluation value, and issue corresponding safety warning prompts based on the state risk level.

2. The AI vision-based power worker behavior monitoring method according to claim 1, wherein, The method comprises the following steps: Collecting continuous real-time images of power operation personnel during work, and pre-processing the continuous real-time images to obtain pre-processed images, the pre-processing including denoising, image enhancement, distortion correction and resolution standardization.

3. The AI vision-based power worker behavior monitoring method according to claim 2, wherein, The method comprises the following steps: Edge detection algorithm is used to perform edge detection on the pre-processed image, and the pre-processed image is divided into multiple object contours according to the edge detection result; The human body is identified from the multiple object contours through a human body recognition algorithm, and the limb features of the human body are located through a key point detection algorithm; The limb features are input into the preset AI behavior prediction model, and the human body action prediction output is obtained by the preset AI behavior prediction model to obtain the action prediction image of the human body.

4. The AI vision-based power worker behavior monitoring method according to claim 3, wherein, The method comprises the following steps: The human body posture in each action prediction image is analyzed through a posture analysis algorithm to determine the human body posture, and the limb movement trajectory of the human body is constructed based on the human body postures in the continuous multiple action prediction images; The action features of the human body are extracted from the limb movement trajectory of the human body, and the action features of the human body are compared and analyzed with the preset action types in the action feature library, and the preset action type consistent with the action features is determined as the action type corresponding to the action features.

5. The AI vision-based power worker behavior monitoring method according to claim 4, wherein, The method comprises the following steps: The working environment of the power operation personnel is identified from the multiple object contours of the pre-processed image, and the working scene of the power operation personnel is determined according to the working environment; Determine whether the power operation personnel is performing work operation according to the action type, and determine the scene type of the working scene after determining that the power operation personnel is performing work operation; Determine the historical data corresponding to the scene type and the action type respectively, and evaluate the safety of the scene type and the action type based on the historical data respectively, and determine the weight of the scene type and the action type. The behavior safety assessment value of the power operation personnel is calculated based on the evaluation values and the weights of the scene type and the action type.

6. The AI vision-based power worker behavior monitoring method according to claim 5, wherein, The historical data corresponding to the scene type and the action type are determined respectively, and the safety assessment values of the scene type and the action type are determined based on the historical data. The historical data corresponding to the scene type and the action type are determined respectively, and the accident severity and the accident frequency under the scene type and the action type are determined from the corresponding historical data. The accident severity under the scene type and the action type is respectively evaluated to obtain the scene safety assessment value of the scene type and the action safety assessment value of the action type. The accident frequency under the scene type and the action type is respectively normalized to obtain the weight of the scene type and the weight of the action type.

7. The AI vision-based power worker behavior monitoring method according to claim 6, wherein, The calculation formula of the behavior safety assessment value is: F=a*X+b*Y, wherein, F is the behavior safety assessment value, a is the weight of the scene type, M is the scene safety assessment value of the scene type, b is the weight of the action type, and N is the action safety assessment value of the action type.

8. The AI vision-based power worker behavior monitoring method according to claim 5, wherein, The state risk level of the power operation personnel is determined based on the predicted behavior safety assessment value. A state risk level-behavior safety assessment value correspondence relationship is preset, and each change interval is associated with a corresponding state risk level. The behavior safety assessment value of the power operation personnel is obtained, and the state risk level corresponding to the behavior safety assessment value interval is selected as the state risk level of the power operation personnel based on the mapping relationship of the behavior safety assessment value interval in the state risk level-behavior safety assessment value correspondence relationship.

9. The AI vision-based power worker behavior monitoring method according to claim 8, wherein, The corresponding safety warning prompt is issued based on the state risk level. A plurality of preset level ranges are determined, and the preset level range in which the state risk level of the power operation personnel is located is determined. The corresponding safety warning prompt is set according to the preset level range in which the state risk level of the power operation personnel is located. If the state risk level of the power operation personnel is in the first preset level range, a low-risk safety warning prompt is issued. If the state risk level of the power operation personnel is in the second preset level range, a medium-risk safety warning prompt is issued. If the state risk level of the power operation personnel is in the third preset level range, a high-risk safety warning prompt is issued.

10. An AI vision-based power operator behavior monitoring system, characterized in that, The method comprises the following steps: An acquisition module is configured to collect continuous real-time images of the power operation personnel during work and pre-process the continuous real-time images to obtain pre-processed images. A prediction module is configured to determine the limb features of the human body in the pre-processed images and perform human action prediction based on the limb features through a pre-set AI behavior prediction model to obtain an action prediction image. An analysis module is configured to determine the action features of the human body in the action prediction image and analyze the action features to determine the action type corresponding to the action features. An evaluation module is configured to determine a work scene in the preprocessed image, perform a safety evaluation on the behavior of the power worker based on the work scene and the action type, and obtain a behavior safety evaluation value; An early warning module is configured to determine a state risk level of the power worker based on the behavior safety evaluation value, and issue a corresponding safety early warning prompt based on the state risk level.

Images