Intelligent power grid personnel management and control method, system, equipment and medium

Abstract

The invention discloses a smart power grid personnel management and control method, system and device and a medium, and the method comprises the steps: processing and recognizing a field operation image, and obtaining a face feature vector; performing identity verification on the face feature vector according to a preset database, determining a field operator according to an identity verification result, and performing qualification health verification on the field operator to determine an access state; performing operation process monitoring on the field operation personnel with the qualified admission state, identifying operation behaviors, comparing the operation behaviors with a preset safety rule base and operation task information, and determining an operation compliance state; and based on a preset multi-dimensional index scoring system, comprehensively evaluating the access state and the operation compliance state of the operation process to obtain a risk level. According to the invention, the problems of inaccurate field personnel identity qualification health verification, low illegal behavior identification efficiency, missing report and lack of full-process closed-loop management in the existing power grid safety supervision are solved.

Description

Technical Field

[0001] This invention relates to the field of power grid safety supervision technology, and in particular to a method, system, equipment and medium for personnel management in smart grids. Background Technology

[0002] With the deepening of smart grid construction, power operation sites are characterized by numerous locations, wide distribution of personnel, and cross-operation of multiple trades, which puts enormous pressure on traditional safety management models. To improve management efficiency, some power grid companies have introduced video monitoring systems and combined them with information management platforms to remotely supervise on-site operations.

[0003] However, existing video surveillance systems typically operate independently and fail to integrate with safety management data such as personnel qualifications and health records. This prevents the verification of personnel identities and work qualifications before operations begin, leading to numerous risks at work sites, including unlicensed work, impersonation, and discrepancies in health status. During operations, the system largely relies on manual review of video footage to identify violations such as helmet wearing and personnel crossing boundaries. This is not only inefficient but also prone to missed detection due to fatigue and negligence, resulting in delayed risk discovery. Furthermore, even when violations are occasionally detected, alarm information often remains at the record level, lacking a closed-loop management mechanism that integrates early warning triggering, response allocation, result feedback, and performance evaluation. This hinders disciplinary rectification at work sites, significantly diminishing the effectiveness of safety supervision. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a smart grid personnel management method, system, equipment, and medium to address the problems of inaccurate verification of on-site personnel's identity, qualifications, and health in existing power grid safety supervision, low efficiency in identifying violations and missed reports, and lack of closed-loop management throughout the entire process.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a method for personnel management in a smart grid, comprising: acquiring on-site operation images through video equipment, processing and recognizing the on-site operation images to obtain facial feature vectors; verifying the facial feature vectors against a preset database, identifying on-site personnel based on the verification results, and verifying the qualifications and health of the on-site personnel to determine their access status; monitoring the operation process of on-site personnel with qualified access status, identifying their operation behaviors and comparing them with a preset safety rule base and operation task information to determine their operation compliance status; comprehensively evaluating the access status and operation compliance status of the operation process based on a preset multi-dimensional indicator scoring system to obtain a risk level, and triggering an early warning and generating a disposal task based on the risk level.

[0007] As a preferred embodiment of the smart grid personnel management method of the present invention, the step of processing and recognizing the on-site operation image to obtain the facial feature vector includes: locating the facial region of the on-site operation image using anchor frame detection technology to obtain a facial region image, and performing frequency domain analysis and pose estimation on the facial region image to screen out qualified facial images; and extracting feature vectors from the qualified facial images to obtain the facial feature vector.

[0008] The beneficial effects of this preferred technical solution are as follows: By using anchor frame detection technology to locate the face region in the on-site operation image, the range of the face can be determined from the on-site operation image containing elements such as equipment and environment, and the region unrelated to the face can be separated. By performing frequency domain analysis and pose estimation on the face region image, it is possible to identify the blur, angle deviation and other factors that affect feature extraction in the image, thereby screening out qualified face images and preventing such unqualified on-site operation images from entering the feature extraction stage. Finally, feature vector extraction is performed on the qualified face images, and the obtained face feature vector can directly reflect the inherent attributes of the face and correspond to the actual face features.

[0009] As a preferred embodiment of the smart grid personnel management method of the present invention, the step of verifying the facial feature vector according to a preset database and determining the on-site workers based on the verification result includes: calculating the similarity between the facial feature vector and the facial feature vector in the preset database to obtain a similarity value; setting a similarity judgment threshold; when the similarity value reaches the similarity judgment threshold, marking the facial feature vector corresponding to the similarity value as an on-site worker and obtaining the corresponding personnel information from the preset database; when the similarity value does not reach the similarity judgment threshold, marking the facial feature vector corresponding to the similarity value as an unfamiliar person and generating an identity anomaly alarm information and pushing it to the on-site monitoring terminal.

[0010] The beneficial effects of this preferred technical solution are as follows: A similarity value is obtained by calculating the similarity between a facial feature vector and a facial feature vector in a preset database. This value represents the degree of matching between the two. A similarity judgment threshold is set as the basis for distinguishing matching states. When the similarity value reaches the similarity judgment threshold, the corresponding facial feature vector is marked as an on-site worker, and the corresponding personnel information is retrieved from the preset database, thus confirming the identity of on-site personnel and extracting relevant information. When the similarity value does not reach the similarity judgment threshold, the corresponding facial feature vector is marked as an unfamiliar person, and an identity anomaly alarm is generated and pushed to the on-site monitoring terminal, thus realizing the identification of non-workers and the immediate transmission of abnormal situations.

[0011] As a preferred embodiment of the smart grid personnel management method of the present invention, the step of verifying the qualifications and health of the on-site workers to determine their access status includes: determining the required job qualification type and occupational contraindications based on the type of the current job task, and determining whether the personnel information contains the required job qualification type and occupational contraindications, thereby obtaining a judgment result; marking the access status of the on-site workers corresponding to the personnel information whose judgment result is that it contains the required job qualification type and does not contain occupational contraindications as qualified, marking the access status of all other on-site workers corresponding to the judgment results as unqualified, and generating qualification and health abnormality alarm information and pushing it to the on-site monitoring terminal.

[0012] The beneficial effects of this preferred technical solution are as follows: Based on the type of the current work task, the required work qualification type and occupational contraindications are determined, clarifying the core criteria for work access. The judgment result is obtained by checking whether the personnel information contains the required work qualification type and occupational contraindications, providing a specific direction and clear standard for access status determination. On-site workers whose judgment result includes the required work qualification type but does not contain occupational contraindications are marked as qualified for access, while all other cases are marked as unqualified. This achieves a clear definition of whether workers can enter the site. Simultaneously, for unqualified cases, qualification and health abnormality alarm information is generated and pushed to the on-site monitoring terminal, allowing relevant on-site personnel to promptly know the specific circumstances of non-compliance with access requirements.

[0013] As a preferred embodiment of the smart grid personnel management method of the present invention, the step of identifying work behavior and comparing it with a preset safety rule base and work task information to determine the work compliance status includes: performing real-time analysis of the work process using target detection and tracking technology to identify the work behavior, safety equipment wearing status, and location information of the on-site workers; comparing the work behavior and safety equipment wearing status with the preset safety rule base, and determining that the behavior is compliant when the comparison results are consistent, otherwise determining that the behavior is non-compliant; comparing the location information and work behavior with the work task information, and determining that the work is in accordance with the work plan when the comparison results are consistent, otherwise determining that the work is not in accordance with the work plan; marking the work compliance status of on-site workers whose comparison results simultaneously meet both the behavior compliance and work plan compliance as qualified, and marking the work compliance status of all other on-site workers as unqualified.

[0014] The beneficial effects of this preferred technical solution are as follows: By using target detection and tracking technology to analyze the work process in real time, it can comprehensively capture the work behavior, safety equipment wearing status, and location information of on-site workers, providing comprehensive and real-time basic information for determining the compliance status of the work. The work behavior and safety equipment wearing status are compared with the preset safety rule base, and the consistency judgment clarifies whether the work behavior meets the safety operation requirements. The location information and work behavior are compared with the work task information, and the consistency judgment clarifies whether the work is carried out in accordance with the established plan. Then, based on the standard of simultaneously meeting behavioral compliance and work plan compliance, the work compliance status of on-site workers is divided into qualified and unqualified categories, realizing a multi-dimensional and all-round judgment of the compliance of the work process, and providing a clear and unified judgment basis for the on-site work performance of workers.

[0015] As a preferred embodiment of the smart grid personnel management method described in this invention, the step of comprehensively evaluating the access status and compliance status of the work process based on a preset multi-dimensional indicator scoring system to obtain a risk level includes: extracting multiple scoring indicators from the multi-dimensional indicator scoring system, and calculating the score value of each scoring indicator according to the access status and compliance status; weighting and summing the score values ​​of each scoring indicator to obtain a comprehensive score, and mapping the comprehensive score to the corresponding risk level according to a preset risk level classification rule.

[0016] The beneficial effects of this preferred technical solution are as follows: multiple scoring indicators are extracted from the preset multi-dimensional indicator scoring system, and the score values ​​of each scoring indicator are calculated in combination with the access status and compliance status of the operation process. This allows the access and compliance-related status information to be transformed into quantifiable evaluation data. The score values ​​of each scoring indicator are weighted and summed to obtain a comprehensive score, which integrates multi-dimensional evaluation factors and avoids the one-sidedness of single-dimensional judgment. Then, according to the preset risk level classification rules, the comprehensive score is mapped to the corresponding risk level, so that the safety status of the operation process is presented in the form of a clear risk level. This allows the overall safety situation at the work site to form a unified and quantifiable evaluation result, and realizes an integrated evaluation of the safety status of the operation.

[0017] As a preferred embodiment of the smart grid personnel management method of the present invention, the step of triggering an early warning and generating a disposal task according to the risk level includes: triggering an early warning information of the corresponding level according to the risk level, generating a disposal task based on the early warning information, and pushing the disposal task to a designated responsible person; receiving disposal evidence uploaded by the designated responsible person, determining the closed-loop status of the disposal task based on the disposal evidence, generating a disposal record based on the closed-loop status, and using the disposal record as a scoring indicator for the next evaluation cycle.

[0018] The beneficial effects of this preferred technical solution are as follows: Triggering corresponding early warning information based on the risk level ensures that risk alerts match the actual degree of risk. Generating and pushing disposal tasks based on the early warning information to designated responsible personnel clarifies the specific targets and implementing entities for risk disposal, ensuring a clear direction for risk response. Subsequently, receiving disposal evidence uploaded by designated responsible personnel verifies the actual implementation of disposal actions. The closed-loop status of the disposal task is then determined based on the evidence, clearly defining whether the task has met the predetermined requirements. Finally, a disposal record is generated based on the closed-loop status, completely preserving information from the entire risk disposal process. This record is also used as a scoring indicator for the next evaluation cycle, integrating the disposal results into the subsequent evaluation system. This linkage between risk disposal and evaluation ensures that every risk response is accurately recorded and participates in the construction of subsequent evaluation dimensions.

[0019] Secondly, the present invention provides a smart grid personnel management system, comprising: Personnel access control module: used to capture on-site operation images through video equipment, process and recognize facial feature vectors, complete identity verification and qualification and health verification, and determine the personnel's access status; The work behavior monitoring module is used to monitor the work process of on-site workers whose entry status is qualified, identify work behaviors and compare them with the preset safety rule base and work task information to determine the work compliance status. Risk Situation Assessment Module: Based on a preset multi-dimensional indicator scoring system, this module comprehensively assesses the access status and compliance status of the work process, calculates the comprehensive score, and determines the risk level. Early warning and response closed-loop module: used to trigger early warnings based on the risk level, generate and push response tasks, and track response feedback to achieve closed-loop management.

[0020] Thirdly, the present invention provides an electronic device, comprising: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions, which, when executed by the processor, implement the steps of the smart grid personnel management method.

[0021] Fourthly, the present invention provides a computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the steps of the smart grid personnel management method.

[0022] Compared with existing technologies, the beneficial effects of this invention are as follows: By constructing a full-process smart grid personnel management system, it achieves a complete chain management upgrade from work access to process monitoring, risk assessment, and closed-loop handling, improving the accuracy and timeliness of personnel management; in the work access stage, through the linkage of facial feature recognition and a preset database, it achieves rapid verification of the worker's identity and matching of qualifications and health status, controlling the safety line from the source, avoiding potential risks such as identity theft, unlicensed work, and inconsistent health status, and ensuring the suitability of workers for work tasks; in the work process management, through target detection and tracking technology, it achieves real-time monitoring and judgment of work behavior, safety equipment wearing, and work location, eliminating the need for traditional manual checks. Overcoming the limitations of video surveillance improves the efficiency and accuracy of identifying violations, reduces omissions and delays caused by human negligence, and ensures the standardization and safety of the operation process. In the risk assessment stage, based on a pre-set multi-dimensional indicator scoring system, the safety status of the entire operation process is quantified, forming an accurate risk level classification. Combined with a graded early warning and task push mechanism, a closed-loop management system is built to ensure that risks and hidden dangers are responded to and properly handled in a timely manner, promoting the transformation of safety supervision from passive response to proactive prevention. In addition, this invention breaks down the information silos between traditional monitoring systems and management data, realizing the deep integration and collaborative linkage of personnel information, operation tasks, safety rules, and other data, and improving adaptability to smart grid operation sites. Attached Figure Description

[0023] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a schematic diagram of the overall process of a smart grid personnel management method according to an embodiment of the present invention. Detailed Implementation

[0025] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0026] Example 1, referring to Figure 1 As an embodiment of the present invention, a smart grid personnel management method is provided, including steps S100 to S400: S100: Acquire on-site operation images through video equipment, process and recognize the on-site operation images, and obtain facial feature vectors.

[0027] S200. Verify the identity of the facial feature vector based on the preset database, determine the on-site workers through the identity verification results, and verify the qualifications and health of the on-site workers to determine their access status.

[0028] S300 monitors the work process of qualified on-site personnel, identifies their work behavior, and compares it with the preset safety rule base and work task information to determine the compliance status of the work.

[0029] S400, based on a preset multi-dimensional indicator scoring system, comprehensively assesses the access status and compliance status of the operation process to obtain the risk level, and triggers early warning and generates disposal tasks according to the risk level.

[0030] It should be noted that smart grid field operations involve high voltage and live equipment, posing a high safety risk. If personnel with incorrect identities, expired qualifications, or substandard health conditions enter the work area, it can easily lead to safety accidents. Traditional personnel management relies heavily on manual registration and verification, which is not only inefficient but also prone to identity theft and untimely qualification verification. Furthermore, the compliance of the work process lacks real-time monitoring and comparison, and the risk assessment is based on a single dimension, resulting in delayed early warnings and untimely responses, making it difficult to ensure the safety and compliance of on-site operations. Therefore, precise and comprehensive management of smart grid personnel is crucial.

[0031] Therefore, in response to the aforementioned issues such as loopholes in identity verification, lagging operation monitoring, and insufficient risk assessment, steps S100 to S400 are used to collect on-site operation images, identify personnel identities, complete qualification and health verification to determine access status, monitor the operation process of accessed personnel in real time, compare safety rules and task information to determine compliance, and then generate risk levels and handling tasks through multi-dimensional comprehensive evaluation. This achieves precise control over the entire process of smart grid personnel from access to operation, fills the loopholes of traditional manual control, improves the timeliness of operation compliance monitoring and risk handling, and ensures the safe and orderly conduct of smart grid on-site operations.

[0032] Example 2, refer to Figure 1 As an embodiment of the present invention, a smart grid personnel management method is provided based on the above embodiment.

[0033] In this embodiment of the application, the step S100, which involves processing and recognizing the on-site operation image to obtain the facial feature vector, includes steps A1 to A2: A1. The face region is located in the on-site operation image by anchor frame detection technology to obtain the face region image. The face region image is then subjected to frequency domain analysis and pose estimation to screen out qualified face images. In this embodiment, an anchor frame detection algorithm is used to process the acquired on-site operation images. The algorithm locates face regions within the on-site operation images using preset anchor frame parameters, obtaining rectangular region images containing faces. Specifically, the anchor frame detection algorithm sets multiple candidate frames at different positions and scales in the on-site operation images. Face localization is performed by calculating the matching degree between the candidate frames and facial features. When the matching degree exceeds a preset threshold, the face region is identified, and a face region image is obtained. Next, frequency domain analysis is performed on the face region image. A Fast Fourier Transform (FFT) is used to transform the face region image from the spatial domain to the frequency domain, and the energy proportion of high-frequency components is calculated to evaluate image sharpness. Simultaneously, a pose estimation model is used to calculate the head pose angles of the face region image, obtaining three pose parameters: pitch angle, yaw angle, and roll angle. When the sharpness evaluation value of the face region image is higher than a preset sharpness threshold, and the absolute values ​​of all head pose angles are less than preset pose thresholds, the face region image is determined to be a qualified face image; otherwise, it is determined to be an unqualified image and is discarded.

[0034] For example, in a certain work site image acquisition, the anchor frame detection algorithm detected three face regions in a frame of the work site image, corresponding to three workers. After frequency domain analysis, the sharpness evaluation values ​​of two of the face region images were 0.82 and 0.75, respectively, both higher than the preset sharpness threshold of 0.7. The sharpness evaluation value of the third face region image was 0.55, which was lower than the preset sharpness threshold, and it was rejected. Next, the pose estimation of the first two face region images was performed. The pitch angle of the first face was 5 degrees, the yaw angle was 8 degrees, and the roll angle was 3 degrees, all within the preset pose threshold of 15 degrees, and it was judged as a qualified face image. The yaw angle of the second face was 22 degrees, which exceeded the threshold range and was judged as an unqualified image. Finally, a qualified face image was selected from the frame of the work site image.

[0035] A2. Extract feature vectors from qualified face images to obtain face feature vectors; First, the qualified face image obtained in step A1 is input into a pre-trained convolutional neural network model. Through forward propagation calculations using multiple layers of convolution, pooling, and fully connected layers, a fixed-dimensional face feature vector is output from the feature extraction layer of the convolutional neural network model. The face feature vector is a numerical representation of the high-dimensional semantic features of the face image, effectively capturing key facial recognition features such as facial contours, facial proportions, and texture details. In this embodiment, the dimension of the face feature vector is set to 512 or 1024 dimensions to ensure a balance between feature representation capability and computational efficiency. Simultaneously, the convolutional neural network model trained through contrastive learning optimizes the feature space distribution, ensuring that face feature vectors extracted from the same person under different lighting, angles, and expressions are closer together in the feature space, while face feature vectors from different people are farther apart.

[0036] For example, the qualified face image selected in step A1 is input into a pre-trained convolutional neural network model for face feature vector extraction. The resolution of the qualified face image is 112×112 pixels. After processing by the convolutional layer, batch normalization layer, and activation layer of the convolutional neural network model, a 512-dimensional face feature vector is output at the feature extraction layer. The numerical value of this face feature vector is represented as a set of floating-point numbers, for example... A total of 512 values ​​were collected, which comprehensively reflect the recognition features of the qualified face image.

[0037] In an optional implementation, the processing and recognition of the on-site operation images in step S100 to obtain facial feature vectors can also be achieved through multi-frame image quality comprehensive evaluation and optimal frame selection. Specifically, multiple frames of on-site operation images of the same person are continuously acquired within a certain time window, for example, 10 to 15 frames are acquired within a 3-second time window to form a multi-frame image sequence. Then, facial region localization is performed on each frame in the multi-frame image sequence to obtain multiple facial region images. Subsequently, an image quality evaluation score is calculated for each facial region image. This image quality evaluation score needs to comprehensively consider multiple quality dimensions such as the clarity of the on-site operation image, the uniformity of illumination, the integrity of the face, and the frontal angle. Each quality dimension is assigned a corresponding weight and then weighted and summed to obtain a comprehensive quality evaluation score. Then, the facial region image with the highest comprehensive quality evaluation score is selected as the qualified facial image, and feature vector extraction is performed on the qualified facial image to obtain the facial feature vector.

[0038] In this embodiment of the application, step S200, which involves verifying the facial feature vector based on a preset database and determining the on-site personnel based on the verification result, includes steps B1 to B3: B1. Calculate the similarity between the facial feature vector and the facial feature vector in the preset database to obtain a similarity value; In this embodiment, the preset database stores the facial feature vectors and corresponding personnel information of all registered workers. First, the cosine similarity algorithm is used to calculate the similarity between the facial feature vector extracted in step A2 and each facial feature vector in the preset database. The calculation method is to divide the dot product of the two facial feature vectors by the product of their respective moduli. The resulting value ranges from -1 to 1. The closer the value is to 1, the more similar the two facial feature vectors are. Then, the highest similarity value is used as the matching result of the facial feature vector, and the corresponding facial feature vector number in the preset database is recorded.

[0039] For example, the similarity calculation is performed between a 512-dimensional face feature vector extracted in step A2 and 2.8 million face feature vectors stored in the preset database. The highest similarity value is 0.92, which corresponds to the face feature vector numbered TZ-20251119 in the preset database.

[0040] B2. Set a similarity judgment threshold. When the similarity value reaches the similarity judgment threshold, mark the facial feature vector corresponding to the similarity value as the on-site worker and obtain the corresponding personnel information from the preset database. First, a similarity judgment threshold is set according to the accuracy requirements of the actual application scenario. In this embodiment, the similarity judgment threshold is set to 0.85. When the similarity value calculated in step B1 is greater than or equal to the similarity judgment threshold, it is determined that the facial feature information is successfully matched with a certain facial feature information in the preset database, and the facial feature information is marked as an on-site worker. Then, according to the matching facial feature vector number, the corresponding personnel information is queried and obtained from the preset database. The personnel information includes, but is not limited to, basic information and business information such as employee number, name, affiliated unit, contact information, job qualification type, and health record.

[0041] B3. When the similarity value does not reach the similarity judgment threshold, the facial feature vector corresponding to the similarity value is marked as an unfamiliar person, and an identity abnormality alarm message is generated and pushed to the on-site monitoring terminal. When the similarity value calculated in step B1 is less than the similarity judgment threshold, it is determined that the facial feature information has no matching record in the preset database or the matching is too low. The facial feature information is marked as an unfamiliar person, and an identity anomaly alarm is immediately generated. In this embodiment, the identity anomaly alarm information includes detailed information such as alarm time, alarm location, facial capture image, and similarity value. The identity anomaly alarm information is pushed to the on-site monitoring terminal for display through the network communication interface. After receiving the identity anomaly alarm information, the on-site monitoring terminal displays the alarm content in red on the screen and triggers the sound and light alarm device to remind the on-site safety management personnel to deal with it in time. The on-site safety management personnel can view the real-time video and facial capture images of the unfamiliar person through the monitoring terminal, verify the identity of the unfamiliar person, and take corresponding control measures, such as persuading them to leave, registering them, or reporting them.

[0042] For example, at a 500 kV substation work site, a video image of a worker is captured and facial feature information is extracted. After step B1, the similarity score is calculated to be 0.62, which is lower than the similarity judgment threshold of 0.85. At this time, the facial feature information is marked as an unknown person, and an identity anomaly alarm is generated. The alarm message reads "At 08:15 on January 21, 2024, an unknown person was found at the No. 1 access control point of Substation No. 5, with a similarity score of 0.62", and a facial capture image is attached. The identity anomaly alarm message is pushed to the on-site monitoring terminal. The monitoring terminal screen displays a red alarm pop-up and emits a buzzer. The on-site safety management personnel immediately check the monitoring screen and find that the person is an outsider who has mistakenly entered the work area. They promptly persuade him to leave and guide him to the visitor registration area.

[0043] In one optional implementation, step S200 involves verifying the facial feature vector against a preset database. Determining on-site personnel based on the verification results can also be achieved using a multi-level verification mechanism. First, the facial feature vector is coarsely screened against the facial feature vectors in the preset database. Vector dimensionality reduction and fast indexing techniques are used to select the top N candidate matching results based on similarity, for example, the top 10 candidate matching results. The value of N can be flexibly set according to the database size. Then, a refined similarity calculation is performed on the selected N candidate matching results. The Euclidean distance method is used to compare the facial feature vector with each candidate feature vector one by one, obtaining N precise similarity values. The highest value among the N precise similarity values ​​is selected as the final similarity value, and this final similarity value is compared with a similarity judgment threshold to complete the identity determination. If the final similarity value reaches the similarity judgment threshold, the person is marked as an on-site worker and the corresponding personnel information is obtained. If the similarity judgment threshold is not reached, the person is marked as an unknown person and an identity anomaly alarm is generated.

[0044] In this embodiment of the application, the step S200, which involves verifying the qualifications and health of on-site workers to determine their access status, includes C1~C2: C1. Based on the type of the current job task, determine the required job qualification type and occupational contraindications, and determine whether the personnel information contains the required job qualification type and occupational contraindications, and obtain the judgment result; In this embodiment, detailed information about the current work task is obtained by connecting to the power grid operation management platform, including the type of work task, work content, work location and work time, etc. Then, the required type of work qualification and occupational contraindications for the current work task are determined according to the type of the current work task. Specifically, the current job task types include, but are not limited to, special job types such as working at height, lifting operations, electrical testing operations, and live-line work. Different job types correspond to different required job qualification types. For example, working at height requires a working at height certificate, lifting operations require a crane operator's certificate, and electrical testing operations require an electrical testing qualification certificate. At the same time, occupational contraindications are determined based on the working environment and job content. For example, personnel with epilepsy, heart disease, high blood pressure, etc., are prohibited from participating in working at height. Then, the job qualification information and health record information obtained in step B2 are retrieved to determine whether the personnel information contains the required job qualification type and whether it contains any of the occupational contraindications. The determination result is divided into four situations: containing the required job qualification type and not containing occupational contraindications, containing the required job qualification type and containing occupational contraindications, not containing the required job qualification type and not containing occupational contraindications, and not containing the required job qualification type and containing occupational contraindications.

[0045] For example, in a high-altitude operation task at a 500 kV substation, the current operation task type is identified as high-altitude operation through the power grid operation management platform, and the required operation qualification type is determined to be a high-altitude operation certificate. Occupational contraindications include epilepsy, heart disease, and hypertension. For a certain on-site worker marked as an on-site worker in step B2, the unique identity information of the on-site worker is obtained through a preset database. The corresponding operation qualification information and health record information are retrieved from the intrinsic safety big data platform using the unique identity information. It is found that the on-site worker holds a crane operation certificate but does not hold a high-altitude operation certificate. The health record shows no occupational contraindications. Therefore, the judgment result is that the required operation qualification type is not included, but occupational contraindications are not included.

[0046] C2. Mark the access status of on-site workers whose judgment results include the required job qualification type and do not contain occupational contraindications as qualified, and mark the access status of all other on-site workers whose judgment results include unqualified, and generate qualification and health abnormality alarm information and push it to the on-site monitoring terminal. Based on the judgment result obtained in step C1, the on-site worker's access status is determined. If the judgment result indicates that the worker possesses the required job qualification type and does not have any occupational contraindications, it means the worker meets the qualification requirements for the current job and their health condition is satisfactory; the worker's access status is marked as qualified. If the judgment result falls into one of the other three categories—containing the required job qualification type but also containing occupational contraindications, not containing the required job qualification type but also not containing occupational contraindications, or not containing both the required job qualification type and occupational contraindications—it means the worker does not meet the job requirements or there is a health risk; the worker's access status is marked as unqualified. Unqualified; For on-site workers whose access status is marked as unqualified, an abnormal qualification and health alarm message is immediately generated. The abnormal qualification and health alarm message includes details such as alarm time, alarm location, personnel information, missing work qualification type or existing occupational contraindications, and is pushed to the on-site monitoring terminal and mobile work terminal for display via network communication interface. After receiving the abnormal qualification and health alarm message, the on-site monitoring terminal displays the alarm content in red and triggers the audible and visual alarm device. At the same time, the work permission management interface of the mobile work terminal grays out and disables the work permission of the on-site worker, preventing him / her from participating in the current work task.

[0047] For example, in the example of step C1, the judgment result of a certain on-site worker is that the required work qualification type is not included but the occupational contraindications are not included. The access status of the on-site worker is marked as unqualified, and a qualification health abnormality alarm message is generated. The alarm message shows "At 09:30 on January 21, 2024, in the high-altitude work area of ​​Substation No. 5, personnel Zhang (employee number GZ-19821122) lacks a high-altitude work permit and is prohibited from participating in the work", and is accompanied by a facial capture image of Zhang and the work qualification information he currently holds. The qualification health abnormality alarm message is pushed to the on-site monitoring terminal and the mobile work terminal. The on-site monitoring terminal screen displays a red alarm pop-up window, and the high-altitude work permission button for Zhang on the mobile work terminal is grayed out and disabled. After receiving the alarm, the on-site safety management personnel immediately require Zhang to leave the work area and arrange for a person with a high-altitude work permit to replace him.

[0048] In this embodiment of the application, step S300, which involves identifying the work behavior and comparing it with the preset safety rule base and work task information to determine the work compliance status, includes steps D1 to D4: D1. Real-time analysis of the operation process is conducted through target detection and tracking technology to identify the work behavior, safety equipment wearing status, and location information of on-site workers; In this embodiment, the on-site workers whose access status is marked as qualified in step C2 are monitored during the work process. First, the work video stream is acquired in real time through high-definition video monitoring equipment at a frame rate of 25 frames per second. The work video stream is input into a pre-trained target detection model for frame-by-frame analysis to detect targets such as personnel, safety helmets, reflective vests, and safety belts. When the detection confidence is greater than 0.7, it is determined to be a valid target. Then, the DeepSORT algorithm is used to track the valid targets. The position of the valid targets is predicted by Kalman filtering, and the matching cost between the detection box and the predicted box is calculated. Then, the Hungarian algorithm is used to complete the association and assign a unique tracking number to each valid target to achieve continuous identification of the same on-site worker in consecutive video frames. Next, the type of work behavior is identified. When a person is detected bending over for more than 2 seconds, it is determined to be low-altitude work. When both hands are gripping an object above the head for more than 3 seconds, it is determined to be high-altitude work. Then, the wearing status of safety equipment is identified. If a certain piece of equipment is not detected for 5 consecutive frames, it is determined to be not worn. Finally, the image coordinates are converted into actual plane coordinates through camera calibration parameters, and the personnel position information is recorded.

[0049] For example, at a 220 kV substation work site, multiple on-site workers were detected. Among them, the on-site worker with tracking number T-001 was found to be wearing a safety helmet, reflective vest, and insulating gloves, and was determined to be working at height, at a location of (15.2 meters, 8.6 meters). The on-site worker with tracking number T-002 was not found to be wearing a safety helmet for 5 consecutive frames, and was determined to be working at low height, at a location of (12.8 meters, 10.3 meters).

[0050] D2. Compare the work behavior and safety equipment wearing status with the preset safety rule base. If the comparison results are consistent, the behavior is judged to be compliant; otherwise, the behavior is judged to be non-compliant. In this embodiment, a preset safety rule base stores safety regulations corresponding to different types of work behaviors. First, the corresponding regulations are queried based on the type of work behavior identified in step D1. For example, working at height requires wearing a safety helmet, reflective vest, and safety belt; lifting operations require wearing a safety helmet and reflective vest, and standing under the lifted object is prohibited; working with electricity requires wearing a safety helmet, insulating gloves, and insulating shoes. The wearing status of safety equipment identified in step D1 is compared with the regulations item by item. If the actual equipment fully covers the requirements, it is considered compliant; otherwise, missing equipment is recorded. Next, it is analyzed whether the work behavior exhibits a prohibited mode. For example, if the distance between the personnel and the underside of the lifted object is less than 1 meter for more than 10 seconds, it is considered a violation. When the equipment is worn correctly and there is no prohibited behavior, the behavior is considered compliant; otherwise, the behavior is considered a violation and a violation record is generated, including the violation time, personnel information, violation type, and video screenshot.

[0051] For example, the work behavior of the person with tracking number T-001 is working at height. The regulations require a safety helmet, reflective vest and safety belt. The person actually wore a safety helmet, reflective vest and insulating gloves, but lacked a safety belt. The behavior was judged to be in violation of the regulations, and the record "At 10:15 on January 21, 2024, the person Li (employee number GZ-19810121) was working at height without wearing a safety belt" was generated.

[0052] D3. Compare the location information and work behavior with the work task information. If the comparison results are consistent, it is determined that the work plan is met; otherwise, it is determined that the work plan is not met. In this embodiment, the daily work task information is obtained by connecting to the power grid operation management platform, including task number, work type, work area range, and planned personnel list. The work area range is defined by the vertex coordinates of a polygon. The personnel location identified in step D1 is compared with the work area range, and the ray method is used to determine whether the point is within the polygon. When the personnel location is within the work area range, the location information is determined to be consistent with the work plan; otherwise, it is marked as out-of-bounds construction. Next, the type of work behavior is compared with the type of task work. If the comparison results are consistent, it is determined to be consistent; otherwise, it is marked as inconsistent with the work content. Finally, the employee number obtained in step B2 is compared with the planned personnel list. If it exists in the list, it is determined to be consistent; otherwise, it is marked as unplanned work. When the personnel location, work behavior, and personnel identity are all consistent, it is determined to be consistent with the work plan; otherwise, it is determined to be inconsistent with the work plan, and an alarm record is generated.

[0053] For example, in a 500 kV substation lifting operation task, the vertices of the operation area are (10 m, 5 m), (20 m, 5 m), (20 m, 12 m), and (10 m, 12 m). The planned personnel employee numbers are QZ-19851208, QZ-19920315, and QZ-20001122. The personnel position (18.5 m, 6.9 m) obtained in step D1 with tracking number T-003 is within the area. The operation behavior of lifting is consistent with the task, and the employee number QZ-19851208 is in the list, which is determined to be in accordance with the operation plan.

[0054] D4. Mark the work compliance status of on-site workers who simultaneously meet the requirements of behavioral compliance and work plan compliance as qualified, and mark the work compliance status of all other on-site workers as unqualified. In this embodiment, the final judgment is made by combining the behavioral compliance judgment in step D2 and the work plan compliance judgment in step D3. When the on-site operator meets both behavioral compliance and work plan compliance, the work compliance status is marked as qualified; if neither condition is met, it is marked as unqualified. Then, the behavioral violation records generated in step D2 and the plan inconsistency alarm records generated in step D3 are summarized to form comprehensive violation alarm information. The comprehensive violation alarm information includes alarm time, alarm personnel information, alarm type, specific description of violation or inconsistency, and alarm video screenshot. The comprehensive violation alarm information is pushed to the on-site monitoring terminal, mobile work terminal, and the safety supervision module of the power grid operation management platform through the network communication interface. After receiving the comprehensive violation alarm information, the on-site monitoring terminal highlights the alarm content on the screen and triggers the audible and visual alarm device to remind the on-site safety management personnel to deal with it in time. The mobile work terminal pushes an alarm notification to the work supervisor, requiring immediate rectification measures. The safety supervision module of the power grid operation management platform records the comprehensive violation alarm information to the safety event database to form a safety event ledger.

[0055] For example, the on-site worker with tracking number T-003 is judged to be compliant in step D2 and to be in compliance with the work plan in step D3. Since both conditions are met, the work compliance status is marked as qualified, and the work process is recorded normally. The on-site worker with tracking number T-001 is judged to be in violation of the rules in step D2 because he / she is not wearing a safety belt. The work compliance status is marked as unqualified. The violation records are summarized to generate a comprehensive violation alarm message. The alarm type is "working at height without wearing a safety belt". The alarm is pushed to each terminal and displays "At 10:15 on January 21, 2024, worker Li (employee number GZ-19810121) was working at height without wearing a safety belt. It is recommended to stop the work immediately and wear a safety belt before continuing".

[0056] In this embodiment of the application, step S400, which involves comprehensively assessing the access status and compliance status of the work process based on a preset multi-dimensional index scoring system to obtain the risk level, includes steps E1 to E2: E1. Extract multiple scoring indicators from the multi-dimensional indicator scoring system, and calculate the score value of each scoring indicator based on the access status and operation compliance status. In this embodiment, the multi-dimensional indicator scoring system pre-constructs multiple scoring indicators, covering aspects such as work plan management, alarm handling, and behavioral compliance. First, scoring indicators are extracted from the multi-dimensional indicator scoring system, including work plan indicators (e.g., the proportion of temporary work plans, plan execution rate, etc.), work preparation indicators (e.g., qualification and health anomaly rate), behavioral compliance indicators (e.g., safety equipment wearing rate, boundary crossing incident rate, work violation rate, etc.), and alarm handling indicators (e.g., alarm handling efficiency, closed-loop compliance rate, etc.). Then, based on the access status obtained in step C2 and the work compliance status obtained in step D4, corresponding data indicator values ​​are calculated within the monthly statistical period. For example, the qualification and health anomaly rate is calculated by calculating the proportion of personnel marked as unqualified in the access status; the work violation rate is calculated by calculating the proportion of personnel marked as unqualified in the work compliance status; and the alarm handling efficiency is calculated by calculating the proportion of comprehensive violation alarm information generated in step D4 that has been handled. For each scoring indicator, a piecewise linear scoring rule is used to calculate the score value. The scoring rule sets multiple scoring intervals and corresponding scores. When the data indicator value is within the scoring interval, the score value is calculated by linear interpolation. When it exceeds the interval, the boundary score is taken.

[0057] For example, in a certain monthly evaluation period, the proportion of temporary work plans is 5%. The scoring rules for this indicator are set as follows: 120 points for a proportion ≤3%, linear interpolation for a proportion between 3% and 25%, and 70 points for a proportion ≥25%. The current proportion of 5% falls within the 3% to 25% range, and the score calculated by linear interpolation is approximately 116 points. The safety equipment wearing rate is 92%. The scoring rules for this indicator are set as follows: no points are awarded for a wearing rate ≤90%, 20 points are added for every 1% increase in the wearing rate between 90% and 95%, and 4 points are added for every 1% increase in the wearing rate >95%. The current wearing rate of 92% results in a score of 40 points.

[0058] E2. The scores of each scoring indicator are weighted and summed to obtain a comprehensive score, and the comprehensive score is mapped to the corresponding risk level according to the preset risk level classification rules. The scores of each scoring indicator calculated in step E1 are weighted and summed. First, the weight coefficients corresponding to each scoring indicator are obtained from the multi-dimensional index scoring system. The weight coefficients are set according to the degree of influence of the scoring indicator on operational safety. For example, the weight of temporary operation plan proportion is 20%, the weight of plan execution rate is 15%, the weight of disposal efficiency is 25%, and the weight of safety equipment wearing rate is 10%. The sum of all weight coefficients is equal to 100%. Then, the score of each scoring indicator is multiplied by the corresponding weight coefficient and summed to obtain the comprehensive score. In this embodiment, the comprehensive score is mapped to the corresponding risk level according to the preset risk level classification rules. The comprehensive score between 800 and 1000 points is judged as low risk level, the comprehensive score between 600 and 800 points is judged as medium risk level, and the comprehensive score below 600 points is judged as high risk level.

[0059] For example, in a monthly assessment, the weighted sum of all scoring indicators results in a comprehensive score of 580 points, which is lower than 600 points. According to the risk level classification rules, this score is mapped to a high-risk level.

[0060] In an optional implementation, the risk level obtained in step S400 by comprehensively evaluating the access status and compliance status of the work process based on a preset multi-dimensional indicator scoring system can also be achieved using a dynamic benchmark comparison method. First, scoring indicators are extracted from the multi-dimensional indicator scoring system. Based on the access status obtained in step C2 and the compliance status obtained in step D4, the data indicator values ​​for the current period are statistically analyzed. Simultaneously, historical data indicator values ​​from the same period are retrieved as benchmark reference values. The deviation rate between the current data indicator value and the benchmark reference value is calculated. The deviation rate is equal to the current data indicator value minus the benchmark reference value, divided by the benchmark reference value. For positive indicators such as safety equipment wearing rate, a positive deviation rate indicates improvement. When the deviation rate is negative, it indicates deterioration. For inverse indicators such as the occurrence rate of out-of-bounds events, a positive deviation rate indicates deterioration, and a negative deviation rate indicates improvement. The risk level is then determined based on the absolute value of the deviation rate: a deviation rate less than 10% is considered low risk, between 10% and 30% is considered medium risk, and greater than 30% is considered high risk. Finally, the risk levels of each scoring indicator are comprehensively assessed: a low risk level is achieved when all scoring indicators are low risk, a high risk level is achieved when any scoring indicator is high risk, and a medium risk level is achieved in all other cases.

[0061] In this embodiment of the application, step S400, which involves triggering an early warning and generating a response task based on the risk level, includes steps F1 to F2: F1. Trigger the corresponding level of early warning information based on the risk level, generate a response task based on the early warning information, and push the response task to the designated responsible personnel. Based on the risk level obtained in step E2, corresponding warning information is triggered. In this embodiment, a low-risk level generates a green alert warning information which is only logged; a medium-risk level generates a yellow reminder warning information which includes scoring anomaly indicators and improvement suggestions; and a high-risk level generates a red warning warning information which includes scoring anomaly indicators, related comprehensive violation alarm information records, and mandatory rectification requirements. Next, a handling task is generated based on the warning information. The handling task includes a task number, task type, handling requirements, handling deadline, and related warning information. Then, the designated responsible personnel are queried based on the work tasks involved in the warning information. For low-risk levels, the task is pushed to the on-site safety management personnel; for medium-risk levels, it is pushed to the work supervisor and on-site safety management personnel; and for high-risk levels, it is pushed to the work supervisor, on-site safety management personnel, and the safety supervisor of the affiliated unit. Finally, the handling task is pushed to the designated responsible personnel's mobile work terminal and PC management interface via a network communication interface. The mobile work terminal displays a task card containing basic task information and a deadline countdown.

[0062] For example, in a monthly assessment, a high-risk level comprehensive score of 580 is obtained, and a red warning message is generated showing "the scores for safety equipment wearing rate and boundary crossing incident rate are abnormal, and 12 related comprehensive violation alarm messages are not closed loop". The task number CZ-20240121-01 is generated, with a deadline of 48 hours, and is pushed to the mobile operation terminals of the operation leader Zhang, the on-site safety management personnel Li, and the safety supervisor of the unit Wang.

[0063] F2. Receive the disposal evidence uploaded by the designated responsible personnel, determine the closed-loop status of the disposal task based on the disposal evidence, generate disposal records based on the closed-loop status, and use the disposal records as the scoring indicators for the next evaluation cycle. In this embodiment, designated personnel view the handling tasks pushed in step F1 through a mobile work terminal or PC management interface. After carrying out rectification according to the handling requirements, they upload handling evidence, which includes photos of the rectification site, rectification instructions, and documents signed by relevant personnel. After receiving the handling evidence, verification and review are performed. Verification determines the completeness of the handling evidence and whether the upload time is before the deadline. Review is conducted by superior management to confirm the rectification effect and determine the closed-loop status of the handling task based on the handling evidence: if the verification and review are passed, the closed-loop status is determined to be closed; otherwise, it is determined to be not closed and the task is returned for supplementary materials. Then, a handling record is generated based on the closed-loop status, including task number, handling time, handling personnel, closed-loop status, review comments, etc. The handling record is synchronized to the security event database and associated with the comprehensive violation alarm information generated in step D4. When step E1 is executed in the next evaluation cycle, the handling record of this cycle is retrieved to calculate the percentage of closed-loop tasks and the closed-loop compliance rate. The handling time is calculated to determine the handling efficiency. These scoring indicators are included in the comprehensive scoring calculation of the next cycle.

[0064] For example, Zhang, the person in charge of the operation, uploaded the disposal evidence, including photos of the rectification site and rectification instructions, 36 hours before the disposal deadline. After the verification was passed, it was submitted for review. The review confirmed that the rectification effect was good, and the closed loop status was determined to be closed. A disposal record was generated showing that the disposal took 12 hours. This disposal record will be included in the calculation of the closed loop standardization rate and disposal efficiency scoring indicators in the next month's evaluation.

[0065] In an optional implementation, step S400, which triggers an early warning and generates a response task based on the risk level, can also be implemented using a dynamic risk trend early warning escalation mechanism. First, historical risk level data from multiple recent assessment periods is retrieved, and the trend of risk level changes is calculated. A stable risk state is defined as the risk level remaining unchanged for three consecutive assessment periods; an upward risk trend is defined as the risk level rising for two consecutive assessment periods; and a downward risk trend is defined as the risk level falling for two consecutive assessment periods. The early warning level is determined based on a comprehensive assessment of the current risk level and the risk trend. When the risk level is low and the risk is stable or decreasing, only a log is recorded without triggering an early warning. When the risk level is medium and the risk is rising, a level two early warning is triggered, and a general response task is generated. When the risk level is high or medium and the risk level has risen continuously for more than two periods, a level one early warning is triggered, and an emergency response task is generated. The deadline for emergency response tasks is shortened to 24 hours, and the approval authority for related tasks is suspended until the response task is completed and approved. For cases of a downward risk trend, a risk improvement report is generated and pushed to management personnel, and improvement measures are recorded as an experience case library.

[0066] In summary, by constructing a comprehensive smart grid personnel management system, a full-chain management upgrade has been achieved, encompassing everything from work access to process monitoring, risk assessment, and closed-loop handling, thus improving the accuracy and timeliness of personnel management. During the work access phase, the linkage between facial recognition and a pre-set database enables rapid verification of personnel identities and matching of qualifications and health status, controlling safety from the source and mitigating potential risks such as identity theft, unlicensed work, and discrepancies in health status, ensuring the suitability of personnel for their assigned tasks. In the work process management, target detection and tracking technologies enable real-time monitoring and assessment of work behavior, safety equipment wearing, and work location, overcoming the limitations of traditional manual video review. This invention improves the efficiency and accuracy of identifying violations, reduces omissions and delays caused by human negligence, and ensures the standardization and safety of the operation process. In the risk assessment stage, based on a preset multi-dimensional indicator scoring system, the safety status of the entire operation process is quantified, forming an accurate risk level classification. Combined with a graded early warning and task push mechanism, a closed-loop management system is built to ensure that risks and hidden dangers are responded to and properly handled in a timely manner, promoting the transformation of safety supervision from passive response to proactive prevention. In addition, this invention breaks down the information silos between traditional monitoring systems and management data, realizing deep integration and collaborative linkage of personnel information, operation tasks, safety rules and other data, and improving adaptability to smart grid operation sites.

[0067] Example 3 illustrates a schematic scheme for a smart grid personnel management method. It should be noted that the technical solution of this smart grid personnel management system belongs to the same concept as the technical solution of the aforementioned smart grid personnel management method. Details not described in detail in this embodiment can be found in the description of the technical solution of the aforementioned smart grid personnel management method.

[0068] This embodiment also provides a smart grid personnel management system, including: Personnel access control module: used to capture on-site operation images through video equipment, process and recognize facial feature vectors, complete identity verification and qualification and health verification, and determine the personnel's access status; The work behavior monitoring module is used to monitor the work process of on-site workers whose entry status is qualified, identify work behaviors and compare them with the preset safety rule base and work task information to determine the work compliance status. Risk Situation Assessment Module: Based on a preset multi-dimensional indicator scoring system, this module comprehensively assesses the access status and compliance status of the work process, calculates the comprehensive score, and determines the risk level. Early warning and response closed-loop module: used to trigger early warnings based on risk levels, generate and push response tasks, and track response feedback to achieve closed-loop management.

[0069] This embodiment also provides an electronic device suitable for personnel management in smart grids, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the personnel management method for smart grids as proposed in the above embodiments.

[0070] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, it implements the smart grid personnel management method proposed in the above embodiments.

[0071] The storage medium proposed in this embodiment belongs to the same inventive concept as the method for implementing personnel management in smart grids proposed in the above embodiments. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0072] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0073] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for managing personnel of a smart grid, characterized in that, The method comprises the following steps: acquiring a face feature vector by collecting a live operation image through a video device, processing and recognizing the live operation image; verifying the face feature vector according to a preset database, determining a live operation personnel through a verification result, and determining an access state by performing a qualification and health check on the live operation personnel; monitoring the live operation personnel in the access state, recognizing an operation behavior, and comparing the operation behavior with a preset safety rule library and operation task information respectively to determine an operation compliance state; based on a preset multi-dimensional index scoring system, comprehensively evaluating the access state and the operation compliance state of the operation process to obtain a risk level, triggering a warning according to the risk level, and generating a disposal task.

2. The smart grid personnel management method of claim 1, wherein, The step of acquiring the face feature vector by processing and recognizing the live operation image comprises the following steps: positioning a face region of the live operation image through an anchor frame detection technology to obtain a face region image, and performing frequency domain analysis and pose estimation on the face region image to screen out a qualified face image; extracting a feature vector from the qualified face image to obtain the face feature vector.

3. The smart grid personnel management method of claim 2, wherein, The step of verifying the face feature vector according to the preset database and determining the live operation personnel through the verification result comprises the following steps: calculating a similarity between the face feature vector and a face feature vector in the preset database to obtain a similarity value; setting a similarity judgment threshold, when the similarity value reaches the similarity judgment threshold, marking the face feature vector corresponding to the similarity value as a live operation personnel, and obtaining corresponding personnel information from the preset database; when the similarity value does not reach the similarity judgment threshold, marking the face feature vector corresponding to the similarity value as a stranger, and generating an identity abnormality warning information and pushing it to a live monitoring terminal.

4. The smart grid personnel management method of claim 3, wherein, The step of determining the access state of the live operation personnel by performing the qualification and health check comprises the following steps: determining a required operation qualification type and occupational contraindication according to a type of a current operation task, and judging whether the required operation qualification type and the occupational contraindication are included in the personnel information to obtain a judgment result; marking the access state of the live operation personnel corresponding to the personnel information including the required operation qualification type and not including the occupational contraindication as qualified, marking the access state of the live operation personnel corresponding to all the remaining judgment results as unqualified, and generating a qualification and health abnormality warning information and pushing it to the live monitoring terminal.

5. The smart grid personnel management method of claim 4, wherein, The step of recognizing the operation behavior, comparing the operation behavior with the preset safety rule library and the operation task information respectively, and determining the operation compliance state comprises the following steps: real-time analyzing the operation process through target detection and tracking technology to recognize an operation behavior, a safety equipment wearing condition, and position information of the live operation personnel; comparing the operation behavior and the safety equipment wearing condition with the preset safety rule library, when the comparison results are consistent, determining that the behavior is compliant, otherwise, determining that the behavior is non-compliant; comparing the position information and the operation behavior with the operation task information, when the comparison results are consistent, determining that the operation behavior conforms to the operation plan, otherwise, determining that the operation behavior does not conform to the operation plan; Mark the operation compliance state of the on-site operation personnel who simultaneously meet the behavior compliance and the operation plan compliance as qualified, and mark the operation compliance state of all the remaining on-site operation personnel as unqualified.

6. The smart grid personnel management method of claim 5, wherein, The step of obtaining the risk level based on the preset multi-dimensional index scoring system and the comprehensive evaluation of the access state and the operation compliance state includes: Extracting a plurality of scoring indexes from the multi-dimensional index scoring system, and calculating the scoring values of the scoring indexes according to the access state and the operation compliance state; Summing up the scoring values of the scoring indexes to obtain a comprehensive score, and mapping the comprehensive score to a corresponding risk level according to a preset risk level division rule.

7. The smart grid personnel management method of claim 6, wherein, The step of triggering an early warning and generating a disposal task according to the risk level includes: Triggering the early warning information of the corresponding level according to the risk level, generating the disposal task based on the early warning information, and pushing the disposal task to the designated responsible person; Receiving the disposal evidence uploaded by the designated responsible person, judging the closed-loop state of the disposal task according to the disposal evidence, and generating a disposal record based on the closed-loop state, which is used as a scoring index for the next evaluation period.

8. A smart grid personnel management system, applying the method of any one of claims 1-7, characterized in that, It includes: A personnel access control module for capturing on-site operation images through a video device, processing and identifying face feature vectors, completing identity verification and qualification health verification, and determining the access state of personnel; An operation behavior monitoring module for monitoring the operation process of on-site operation personnel with a qualified access state, identifying operation behaviors, and comparing them with a preset safety rule library and operation task information to determine the operation compliance state; A risk situation assessment module for comprehensive evaluation of the access state and the operation compliance state of the operation process based on a preset multi-dimensional index scoring system, calculation of a comprehensive score, and determination of a risk level; An early warning and disposal closed-loop module for triggering an early warning according to the risk level, generating and pushing a disposal task, and tracking disposal feedback to achieve closed-loop management. 9.An electronic device, comprising: a memory and a processor; The memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions, which realize the steps of the smart grid personnel management method in any one of claims 1 to 7. 10.A computer readable storage medium storing computer executable instructions, which realize the steps of the smart grid personnel management method in any one of claims 1 to 7 when executed by a processor.

Images