Internet-based power transmission and transformation construction site intelligent monitoring and power distribution management system and method

Abstract

The application relates to an Internet-based power transmission and transformation construction site intelligent monitoring power distribution management method, and a system comprising: when an operation violation risk value corresponding to a target special operation site is greater than a preset risk threshold value, acquiring a power transmission and transformation construction site image containing a target operation personnel in the target special operation site; based on a construction area segmentation graph of the target operation personnel extracted from the power transmission and transformation construction site image, performing image recognition on the construction area segmentation graph based on an operation violation detection model to determine whether the target operation personnel has a special operation operation certificate and whether the target operation personnel wears protective equipment; when the target operation personnel does not have the special operation operation certificate and / or does not wear the protective equipment, an intelligent power distribution box executes a preset electrical control operation according to a violation type to perform power transmission and transformation construction site intelligent monitoring power distribution management. The application can avoid resource waste caused by invalid monitoring, realize accurate prevention and control in key areas, and simplify control in low-risk areas.

Description

Technical Field

[0001] This application relates to the field of construction site inspection, and in particular to an Internet-based intelligent monitoring and distribution management system and method for power transmission and transformation construction sites. Background Technology

[0002] Traditional intelligent monitoring and distribution management at power transmission and transformation construction sites largely rely on management personnel's experience or single electrical parameter thresholds (such as residual current exceeding the standard alarm) to determine the risk of electric shock. This approach fails to fully consider the spatial clustering characteristics and cumulative frequency effects of historical violations, leading to vague risk level classifications for different special operation locations. It makes it difficult to identify high-risk areas in advance, often resulting in a "one-size-fits-all" approach to control, which neither effectively prevents targeted prevention of key risks nor maintains normal construction efficiency. Furthermore, operational violations at power transmission and transformation construction sites (such as unlicensed operation of special tasks and failure to wear protective equipment correctly) exhibit clear spatial correlations and recurring patterns. Traditional management models do not structurally integrate and deeply analyze this historical data, failing to provide a scientific basis for risk prediction.

[0003] The response to and control of operational violations is slow and relies on manual intervention, resulting in low efficiency. Currently, the verification of whether workers have special operation certificates and whether they are wearing protective equipment such as insulating gloves, face shields, and insulating shoes properly is mainly completed by on-site management personnel during inspections. This not only consumes a lot of manpower but also suffers from problems such as missed inspections, misjudgments, and untimely responses.

[0004] In addition, the existing intelligent distribution boxes mostly adopt a crude "power outage upon abnormality" approach to control and manage the situation. They do not implement differentiated control measures based on the type of violation, which can easily cause interruptions to normal construction and make it difficult to balance the needs of safety control and construction continuity.

[0005] In summary, existing technologies often rely on the experience of management personnel or single electrical parameter thresholds to determine the risk of electric shock, failing to fully consider the spatial clustering characteristics and frequency accumulation effect of historical violations, as well as the verification of whether operators possess special operation certificates and protective equipment, which mainly relies on on-site management personnel for inspections. To address these issues, the applicant has made corresponding explorations. Summary of the Invention

[0006] The purpose of this application is to solve the above problems by providing an Internet-based intelligent monitoring and distribution management system and method for power transmission and transformation construction sites.

[0007] To achieve the above objectives, the present application adopts the following technical solution: A smart monitoring and power distribution management method for power transmission and transformation construction sites based on the Internet, proposed to meet one of the purposes of this application, includes: The violation dataset acquisition module is used to acquire historical operation violation datasets corresponding to power transmission and transformation construction sites. The power transmission and transformation construction sites include multiple special operation locations, and the historical operation violation datasets include multiple historical operation violation data points. Each historical operation violation data point contains the coordinates of the historical violation location where each type of operation violation event occurred and the operation violation frequency of each type of operation violation event at the coordinates of the historical violation location. The electric shock risk assessment module is used to construct an electric shock risk assessment model based on the coordinates of the historical violation locations and the frequency of the operation violation using Gaussian kernel density estimation. Based on the electric shock risk assessment model, the module determines the operation violation risk value corresponding to each special operation location in the power transmission and transformation construction site. The operation violation risk value is used to characterize the degree of risk of operation violation and electric shock accident at each special operation location. The on-site image acquisition module is used to acquire images of the power transmission and transformation construction site containing the target workers when the risk value of the violation of the target special operation site is greater than a preset risk threshold. The operation violation determination module is used to extract the construction area segmentation map corresponding to the target worker from the power transmission and transformation construction site image based on the training-to-converge construction area segmentation model, and to perform image recognition on the construction area segmentation map based on the training-to-converge operation violation detection model to determine whether the target worker has a special operation certificate and whether the target worker is wearing protective equipment. The power distribution electrical control module is used to perform preset electrical control operations based on the type of violation when the target operator does not have a special operation certificate and / or is not wearing protective equipment, so as to carry out intelligent monitoring and power distribution management at the power transmission and transformation construction site.

[0008] Optionally, extracting a segmentation map of the construction area corresponding to the target worker from the power transmission and transformation construction site image, including: The trained and converged construction area segmentation model is invoked to encode and decode the power transmission and transformation construction site images at multiple scales to obtain multiple image feature information that captures the contour features of the target workers. Based on the multiple image feature information, the construction area segmentation map corresponding to the target worker is extracted from the power transmission and transformation construction site image.

[0009] Optionally, extracting a segmentation map of the construction area corresponding to the target worker from the power transmission and transformation construction site image, including: The multi-level encoding of the construction area segmentation model, which has been trained to convergence, reduces the scale of the original image of the power transmission and transformation construction site image step by step, and generates intermediate feature information corresponding to each scale. The intermediate feature information is used to characterize the contour features of the construction area segmentation map corresponding to the target workers in the power transmission and transformation construction site image. The construction area segmentation model, which has been trained to convergence, is decoded in multiple levels. Based on the image feature information produced by the intermediate feature information at the smallest scale, the intermediate feature information produced by the same level encoding is used as reference information to decode higher scale image feature information. The image feature information is used to represent the contour features of the construction area segmentation map corresponding to the target workers in the power transmission and transformation construction site image in the form of a mask. The original image of the power transmission and transformation construction site is segmented using a masked image data fused from all image feature information by a construction area segmentation model, in order to extract the construction area segmentation map corresponding to the target workers from the power transmission and transformation construction site image.

[0010] Optionally, image recognition is performed on the construction area segmentation map based on a converged operational violation detection model to determine whether the target worker possesses a special operation certificate and whether the target worker is wearing protective equipment, including: The pre-trained operation violation detection model is invoked, and features are extracted from the construction area segmentation map through a convolutional backbone network to obtain multi-layer feature maps of different scales. A region generation network is used to generate multiple candidate regions of interest for the feature maps of different scales in the multi-layer structure. A region of interest alignment operation is performed on the candidate regions of interest to obtain the aligned feature map. The detection head network performs bounding box regression, recognition, and mask image prediction on the feature map aligned with the region of interest. The bounding box regression is used to accurately locate the target worker region in the construction area segmentation map. The mask image prediction is used to perform instance segmentation of the target worker region. The recognition is used to extract deep features from the segmented target worker region. Based on the results of the identification process, the document identification features, document wearing features, and document display features of the special operation certificate of the target operator are extracted, and the determination result of whether the target operator possesses a special operation certificate is output. At the same time, the wearing position and wearing standard features of the protective equipment of the target operator are extracted, and the determination result of whether the target operator is wearing protective equipment is output.

[0011] Optionally, an electric shock risk assessment model is constructed, including: Based on Gaussian kernel density estimation, the square of the Euclidean distance between the monitoring coordinates corresponding to the special operation location at the power transmission and transformation construction site and the coordinates of the historical violation location is calculated, and the first ratio between the square of the Euclidean distance and twice the square of the bandwidth parameter is calculated. The natural exponential function value of the first ratio is calculated and determined. The first product between the natural exponential function value and the frequency of work violations is calculated and determined. At the same time, the normalized values ​​of the electrical operating parameters, equipment operating parameters and environmental parameters of the power transmission and transformation construction site are integrated. The weighted value of each normalized value and the corresponding weight coefficient is calculated. The first product is multiplied by each weighted value and summed to determine the risk value of work violations, so as to construct an electric shock risk assessment model.

[0012] Optionally, the risk value of work violation corresponding to each special operation location in the power transmission and transformation construction site is determined according to the electric shock risk assessment model, including: The monitoring coordinates, real-time electrical operating parameters, real-time equipment operating parameters, and real-time environmental parameters of each special operation site in the power transmission and transformation construction site are collected. The real-time electrical operating parameters, real-time equipment operating parameters, and real-time environmental parameters are normalized to obtain the normalized values ​​corresponding to the real-time electrical operating parameters, the real-time equipment operating parameters, and the real-time environmental parameters. The monitoring coordinates and normalized values ​​of each special operation location are input into the electric shock risk assessment model to calculate the quantitative value corresponding to each special operation location. The quantified value is used as the corresponding risk value for operational violations at each special operation site in the power transmission and transformation construction site.

[0013] Optionally, when the target operator does not possess a special operation certificate and / or is not wearing protective equipment, the intelligent distribution box executes preset electrical control operations according to the type of violation, in order to conduct intelligent monitoring and power distribution management at the power transmission and transformation construction site, including: When it is determined that the target workers at the special operation site of the power transmission and transformation construction site do not have a special operation certificate and / or are not wearing protective equipment in accordance with regulations, a corresponding violation judgment signal will be output. The intelligent distribution box deployed at the power transmission and transformation construction site receives the violation judgment signal and identifies the violation type corresponding to the violation judgment signal. The violation type includes only not wearing protective equipment properly, only not having a special operation certificate, and neither having a special operation certificate nor wearing protective equipment properly. The protective equipment includes insulating gloves, protective face mask, and insulating shoes. The intelligent distribution box performs pre-set electrical control operations on the power supply circuit corresponding to the target special operation location based on the correspondence between the violation type and the electrical control operation, so as to carry out intelligent monitoring and power distribution management of the power transmission and transformation construction site. The electrical control operation includes one or more combinations of current limiting and capacity reduction of the power supply circuit, instantaneous power outage and linkage audible and visual alarm.

[0014] Optionally, the intelligent distribution box, based on the correspondence between violation types and electrical control operations, executes pre-set electrical control operations matching the power supply circuits corresponding to the target special operation location, in order to conduct intelligent monitoring and power distribution management at the power transmission and transformation construction site, including: When the violation judgment signal is only the failure to wear protective equipment properly, electrical control operations are performed to limit current and reduce capacity, and trigger audible and visual alarms. When the violation judgment signal is that the person only lacks a special operation certificate, the electrical control operation is to perform an instantaneous power outage of the power supply circuit and trigger an audible and visual alarm. When someone does not have a special operation certificate and does not wear protective equipment properly, they shall perform electrical control operations that involve momentarily cutting off the power supply circuit, triggering audible and visual alarms, and locking the closing authority.

[0015] Optionally, the basic network architecture of the construction area segmentation model is the U²-Net model; the basic network architecture of the operation violation detection model includes the Mask-RCNN model.

[0016] A digital distribution box, proposed for another purpose of this application, is applied to the Internet-based intelligent monitoring and distribution management method for power transmission and transformation construction sites as described in any of the above claims, comprising: The violation dataset acquisition module is used to acquire historical operation violation datasets corresponding to power transmission and transformation construction sites. The power transmission and transformation construction sites include multiple special operation locations, and the historical operation violation datasets include multiple historical operation violation data points. Each historical operation violation data point contains the coordinates of the historical violation location where each type of operation violation event occurred and the operation violation frequency of each type of operation violation event at the coordinates of the historical violation location. The electric shock risk assessment module is used to construct an electric shock risk assessment model based on the coordinates of the historical violation locations and the frequency of the operation violation using Gaussian kernel density estimation. Based on the electric shock risk assessment model, the module determines the operation violation risk value corresponding to each special operation location in the power transmission and transformation construction site. The operation violation risk value is used to characterize the degree of risk of operation violation and electric shock accident at each special operation location. The on-site image acquisition module is used to acquire images of the power transmission and transformation construction site containing the target workers when the risk value of the violation of the target special operation site is greater than a preset risk threshold. The operation violation determination module is used to extract the construction area segmentation map corresponding to the target worker from the power transmission and transformation construction site image based on the training-to-converge construction area segmentation model, and to perform image recognition on the construction area segmentation map based on the training-to-converge operation violation detection model to determine whether the target worker has a special operation certificate and whether the target worker is wearing protective equipment. The power distribution electrical control module is used to perform preset electrical control operations based on the type of violation when the target operator does not have a special operation certificate and / or is not wearing protective equipment, so as to carry out intelligent monitoring and power distribution management at the power transmission and transformation construction site.

[0017] Compared to existing technologies, this application addresses the problems of existing technologies that rely heavily on management personnel experience or single electrical parameter thresholds to determine electric shock risks, fail to fully consider the spatial clustering characteristics and frequency accumulation effects of historical violations, and lack sufficient verification of whether operators possess special operation certificates and protective equipment, relying mainly on on-site management personnel for inspections. This application offers the following benefits, including but not limited to: Firstly, this application integrates the coordinates and frequency of historical violations, and constructs an electric shock risk assessment model by combining Gaussian kernel density estimation. This model accurately quantifies the risk value of each special operation location, captures the spatial clustering characteristics and frequency accumulation effect of violation risks, and solves the drawbacks of "one-size-fits-all" control.

[0018] Risk assessment results directly guide the allocation of monitoring resources, and subsequent image acquisition and verification are only initiated for target areas with risk values ​​exceeding the threshold, avoiding resource waste caused by ineffective monitoring, and achieving "precise prevention and control in key areas and simplified management and control in low-risk areas", which is suitable for the differentiated management needs of multiple special operation sites in power transmission and transformation construction sites.

[0019] Secondly, this application first uses the U²-Net model to remove the complex background of dense equipment and cables at the construction site, extracting a clean segmentation map of the worker areas. Then, it uses the Mask-RCNN model to accurately identify the special operation certificate qualifications and the wearing status of protective equipment, solving the pain points of missed detections and misjudgments in traditional manual inspections and the interference of background in single model recognition. It realizes automated, all-weather monitoring of qualification verification and protection inspection, eliminating the need for manual on-site verification, significantly reducing labor costs. At the same time, through fine-grained feature extraction, it effectively distinguishes ambiguous scenarios such as "certificate holder not displaying" and "improper wearing," improving the precision of violation judgment.

[0020] Thirdly, this application identifies various types of violations, such as simply not wearing protective equipment properly, simply lacking a special operation certificate, or lacking both a special operation certificate and not wearing protective equipment properly. It then links the intelligent distribution box to perform differentiated control actions, including current limiting and capacity reduction, instantaneous power outage, locking of closing permissions, and audible and visual alarms. This not only eliminates serious safety hazards but also avoids indiscriminate power outages that interfere with normal construction, achieving a balance between safety standards and construction continuity.

[0021] Fourth, the Internet-based intelligent monitoring and power distribution management method for power transmission and transformation construction sites proposed in this application can be directly implemented based on the digital distribution boxes, intelligent cameras, and cloud management systems already deployed at the power transmission and transformation construction sites without large-scale modifications. At the same time, it supports the dynamic adjustment of risk thresholds and control parameters and can adapt to the personalized needs of different special operation scenarios such as high-voltage wiring, tower erection, and welding operations. Attached Figure Description

[0022] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is a flowchart illustrating the Internet-based intelligent monitoring and power distribution management method for power transmission and transformation construction sites in this application embodiment; Figure 2 This is a schematic diagram illustrating the process of constructing an electric shock risk assessment model in an embodiment of this application; Figure 3 This is an exemplary structural diagram of the Internet-based intelligent monitoring and distribution management system for power transmission and transformation construction sites in this application embodiment. Detailed Implementation

[0023] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain this application, and should not be construed as limiting this application.

[0024] Unless otherwise expressly stated, the various embodiments disclosed in this application can be combined in a cross-cutting manner to flexibly construct new embodiments, as long as such combination does not depart from the inventive spirit of this application and can meet the needs of the prior art or solve a certain deficiency in the prior art. Those skilled in the art should be aware of such modifications.

[0025] Please see Figure 1 In one embodiment of the Internet-based intelligent monitoring and power distribution management method for power transmission and transformation construction sites, this application includes: Step S10: Obtain the historical operation violation dataset corresponding to the power transmission and transformation construction site. The power transmission and transformation construction site includes multiple special operation locations, and the historical operation violation dataset includes multiple historical operation violation data points. Each historical operation violation data point contains the coordinates of the historical violation location where each type of operation violation event occurred and the operation violation frequency of each type of operation violation event in the coordinates of the historical violation location. Specifically, the focus is on multiple special operation locations at power transmission and transformation construction sites, including but not limited to high-voltage wiring areas, tower erection areas, welding operation areas, and cable laying areas. The historical violation location coordinates record the specific spatial location of each type of operational violation event; the frequency of each operational violation event at the corresponding historical violation location coordinates is statistically analyzed, i.e., the violation frequency. For example, the historical number of occurrences of the violation "protective equipment not worn" in a certain high-voltage wiring area. These operational violations include unlicensed operation of special tasks; improper wearing of protective equipment; abnormal power distribution line parameters; and abnormal equipment operating conditions, such as equipment immersion in water, overheating, or short-circuit precursors.

[0026] Step S20: Using Gaussian kernel density estimation, a risk assessment model for electric shock is constructed based on the coordinates of the historical violation locations and the frequency of the operation violation. The risk value of the operation violation corresponding to each special operation location in the power transmission and transformation construction site is determined based on the risk assessment model. The risk value of the operation violation is used to characterize the degree of risk of operation violation and electric shock accident at each special operation location. After obtaining the historical operation violation dataset corresponding to the power transmission and transformation construction site, Gaussian kernel density estimation is used to construct an electric shock risk assessment model based on the coordinates of the historical violation locations and the frequency of the operation violations. The electric shock risk assessment model is used to determine the operation violation risk value corresponding to each special operation location in the power transmission and transformation construction site. The operation violation risk value is used to characterize the degree of risk of operation violations occurring at each special operation location and causing electric shock accidents. In some embodiments, please refer to Figure 2 Construct an electric shock risk assessment model, including: Step S201: Based on Gaussian kernel density estimation, calculate the square of the Euclidean distance between the monitoring coordinates corresponding to the special operation location at the power transmission and transformation construction site and the coordinates of the historical violation location, and calculate the first ratio between the square of the Euclidean distance and twice the square of the bandwidth parameter. Specifically, based on the spatial correlation principle of Gaussian kernel density estimation, this method quantifies the spatial proximity between current special operation locations and historical violation locations. Using the monitoring coordinates of a specific special operation location at a power transmission and transformation construction site as a benchmark, the Euclidean distance between this coordinate and the coordinates of each historical violation location is calculated, and then this distance is squared. The core objective is to highlight the impact of spatial distance on risk; the closer the current location is to a historical violation location, the smaller the squared distance value, and the stronger the risk correlation reflected in subsequent calculations. This aligns with the spatial clustering characteristic of violation risks at power transmission and transformation construction sites, such as the likelihood of similar violations occurring around a high-voltage work area.

[0027] By adjusting the spatial influence range using bandwidth parameters, a standardized spatial similarity coefficient is obtained. First, a bandwidth parameter suitable for the power transmission and transformation construction site scenario is determined. Those skilled in the art can flexibly set this parameter based on the site size and the distance between special operation locations. For example, the bandwidth can be set to 5 meters in densely populated work areas and 10 meters in open work areas. The square of this bandwidth parameter is calculated and multiplied by 2 to obtain a baseline value. Then, the square of the Euclidean distance calculated in the above steps is divided by this baseline value to obtain the first ratio. This first ratio is essentially a standardization process for spatial distance, enabling a unified comparison of distance data under different scenarios. The smaller the ratio, the higher the spatial similarity between the current location and historical violation locations.

[0028] Step S202: Calculate and determine the natural exponential function value of the first ratio, calculate and determine the first product between the natural exponential function value and the frequency of work violations, and simultaneously integrate the normalized values ​​of the electrical operating parameters, equipment operating parameters, and environmental parameters of the power transmission and transformation construction site, calculate the weighted value of each normalized value and the corresponding weight coefficient, multiply the first product by each weighted value and sum them to determine the work violation risk value, so as to construct an electric shock risk assessment model.

[0029] Specifically, the electrical operating parameters include three-phase voltage (phase A, phase B, and phase C), neutral voltage, three-phase current (phase A, phase B, and phase C), total circuit current, and residual current (leakage current); the equipment operating parameters include the operating current, working time, and start / stop status of power tools (such as wiring pliers and welding machines); the oil temperature, winding temperature, and insulation status of transformers and switchgear; the temperature of cable joints (to prevent overheating and fire); and the environmental parameters include ambient temperature, relative humidity, rainfall, water depth, and equipment immersion status.

[0030] The first ratio calculated above is then subjected to a natural exponential function value. This operation maps the first ratio to a value between 0 and 1, forming a spatial similarity coefficient. The closer the spatial distance, the closer the coefficient is to 1; the farther the spatial distance, the closer the coefficient is to 0. This spatial similarity coefficient is then multiplied by the frequency of operational violations at the corresponding historical violation location to obtain the first product. This step integrates "spatial location" and "historical violation frequency." For example, at the same spatial distance, the higher the frequency of historical violations, the larger the first product, initially reflecting the degree of risk accumulation at that location.

[0031] The calculated first product is multiplied by the weighted values ​​of all parameters, and then all the multiplication results are summed to obtain a quantitative and comprehensive operational violation risk value, which is used to construct an electric shock risk assessment model. Ultimately, this electric shock risk assessment model can directly output the operational violation risk value for each special operation location at the power transmission and transformation construction site. The higher the risk value, the greater the probability of an operational violation occurring at that location and causing an electric shock accident, providing a quantitative basis for subsequent graded management.

[0032] In a further embodiment, the risk value of work violation corresponding to each special operation location in the power transmission and transformation construction site is determined according to the electric shock risk assessment model, including: Step S2001: Collect the monitoring coordinates, real-time electrical operation parameters, real-time equipment operation parameters, and real-time environmental parameters of each special operation location in the power transmission and transformation construction site; normalize the real-time electrical operation parameters, real-time equipment operation parameters, and real-time environmental parameters to obtain the normalized values ​​corresponding to the real-time electrical operation parameters, the real-time equipment operation parameters, and the real-time environmental parameters. Step S2002: Input the monitoring coordinates and normalized values ​​of each special operation location into the electric shock risk assessment model to calculate the quantitative value corresponding to each special operation location; use the quantitative value as the operation violation risk value corresponding to each special operation location in the power transmission and transformation construction site.

[0033] Step S30: When the risk value of the violation of the target special operation location is greater than the preset risk threshold, obtain the power transmission and transformation construction site image containing the target workers at the target special operation location; A Gaussian kernel density estimation model is constructed based on the coordinates of the historical violation locations and the frequency of the operational violations. The operational violation risk value corresponding to each special operation location in the power transmission and transformation construction site is determined according to the operational violation risk assessment model. The operational violation risk value is used to characterize the degree of risk of operational violations and electric shock accidents occurring at each special operation location. When the operational violation risk value corresponding to the target special operation location is greater than a preset risk threshold, an image of the power transmission and transformation construction site containing the target personnel is acquired at the target special operation location.

[0034] Specifically, the images of the power transmission and transformation construction site can be obtained through intelligent cameras and mobile monitoring terminals deployed at the construction site. The equipment supports all-weather shooting (adapting to complex environments such as rain and night), and the data transmission stability can be ensured through the 4G / 5G communication module of the intelligent distribution box. It is linked with the "equipment positioning" function of the equivalent diversion cloud management system to ensure that the acquisition equipment is accurately aimed at the target special operation location. After the acquired images of the power transmission and transformation construction site are preprocessed (such as deblurring and noise reduction), they are transmitted to the downstream operation violation judgment module in real time. At the same time, the image acquisition time, location coordinates, corresponding risk values ​​and other metadata are synchronized and stored in association with the "electricity consumption record" and "alarm record" modules of the intelligent distribution box to provide a basis for tracing the source of subsequent violation judgment results. The acquisition action is only initiated for high-risk locations to avoid invalid acquisition of low-risk areas, reduce the data processing pressure of the system, and solve the problem of "multiple work points and large personnel flow" at the power transmission and transformation construction site, so as to ensure both accurate management and construction efficiency.

[0035] Step S40: Extract the construction area segmentation map corresponding to the target worker from the power transmission and transformation construction site image based on the training-to-converge construction area segmentation model, and perform image recognition on the construction area segmentation map based on the training-to-converge operation violation detection model to determine whether the target worker has a special operation certificate and whether the target worker is wearing protective equipment. When the risk value of a violation corresponding to a target special operation location exceeds a preset risk threshold, after acquiring an image of a power transmission and transformation construction site containing the target worker at the target special operation location, a construction area segmentation map corresponding to the target worker is extracted from the power transmission and transformation construction site image based on a pre-trained converged construction area segmentation model. Then, an image recognition is performed on the construction area segmentation map based on a pre-trained converged operation violation detection model to determine whether the target worker possesses a special operation certificate and whether the target worker is wearing protective equipment. The basic network architecture of the construction area segmentation model is a U²-Net model; the basic network architecture of the operation violation detection model includes a Mask-RCNN model.

[0036] In some embodiments, at power transmission and transformation construction sites, workers may be located under high-voltage towers (far-field small targets), next to equipment (mid-field targets), or performing wiring operations (near-field large targets), resulting in significant scale fluctuations. U²-Net, through its 7-level encoding / decoding structure and RFB (receptive field enhancement block) design, can simultaneously capture details of small targets and outlines of large targets, avoiding the problems of incomplete segmentation of distant personnel and blurred edges of near personnel found in traditional segmentation models (such as FCN and ordinary UNet), ensuring complete extraction of worker areas at different scales. Image acquisition and preliminary processing at power transmission and transformation construction sites rely on intelligent cameras or local computing modules in intelligent distribution boxes (such as 32-bit ARM microprocessors), placing extremely high demands on model inference speed and resource consumption. U²-Net has only 4.7M parameters, far lower than models such as DeepLabv3+ and PSPNet, with an inference time of less than 0.5 seconds for a single 1024×1024 image, achieving instantaneous "acquisition-segmentation" response, avoiding delays caused by cloud transmission, and meeting the functional requirements of intelligent distribution boxes for "real-time monitoring and rapid response."

[0037] Workers' protective equipment (such as insulated gloves and face shields), hand tools (such as crimping pliers), and surrounding cables are all thin structures, making them prone to segmentation breaks or background adhesion in traditional models. U²-Net enhances feature fusion through a dual U-shaped structure, achieving higher segmentation accuracy for thin edges than traditional models. It can accurately isolate background interference from workers, dense equipment, and cables, outputting a clean construction area segmentation map, thus eliminating background noise for subsequent violation detection.

[0038] The work scenarios (high-voltage wiring, welding) and equipment types (transformers, switchgear) at power transmission and transformation construction sites are unique, making general pre-trained models insufficiently adaptable. U²-Net, without relying on backbone networks like ResNet for pre-training, can be directly fine-tuned using a small number of labeled samples (around 2000 images) from the construction site, quickly adapting to the target features of special work scenarios and reducing the sample collection cost for model deployment. Therefore, the U²-Net model is used as the basic network architecture for the construction area segmentation model in this application. Once the construction area segmentation model has converged, it can be put into production use to extract the construction area segmentation map corresponding to the target workers from images of power transmission and transformation construction sites.

[0039] In some embodiments, Mask-RCNN employs a three-branch structure—bounding box regression, classification, and mask prediction—to achieve precise localization of worker areas (solving the target differentiation problem in multi-person operations) in a single inference step; instance segmentation (further refining worker body parts, such as hands and head, for easier identification of protective equipment); and deep feature extraction (for identifying identification documents and wearing status of special operation permits). Compared to the traditional two-step approach of YOLO detection combined with a separate classification model, this reduces model deployment complexity and inference latency.

[0040] Furthermore, special operation permits may be small targets such as those worn in pockets or held in hand. Traditional detection models (such as Faster R-CNN) using ROI pooling can lead to feature alignment errors, affecting the accuracy of small target recognition. Mask-RCNN employs ROI Align technology, using bilinear interpolation to maintain spatial alignment between the feature map and the original image, greatly improving the accuracy of small target recognition (such as ID cards and protective face shields). It can accurately capture key features such as document identification and wearing position, avoiding misjudgments such as "not displaying the permit" or "not wearing it correctly".

[0041] Furthermore, the model can simultaneously define multiple detection categories such as "special operation certificate," "insulating gloves," "protective face mask," and "insulating shoes." By outputting the confidence level of each category through classification branches, it directly determines "whether a certificate is held" and "whether protective equipment is worn," eliminating the need to build multiple separate single-classification models and simplifying model deployment and maintenance costs. Simultaneously, the mask prediction branch can further verify the completeness of protective equipment wearing (e.g., whether gloves are fully worn and whether the face mask fits snugly), improving the precision of violation detection.

[0042] Furthermore, power transmission and transformation construction sites present complex environments such as strong light, rain, and insufficient nighttime lighting, which can easily lead to a decline in the recognition performance of traditional models. Mask-RCNN, by combining a candidate region generation network (RPN) with multi-scale feature maps, exhibits better robustness to changes in lighting and target occlusion (such as equipment obscuring the upper body of workers) than single-scale detection models. It can stably output judgment results in complex environments, avoiding missed or false judgments caused by environmental interference.

[0043] Therefore, the Mask-RCNN model is used as the basic network architecture of the operation violation detection model in this application. After the operation violation detection model is trained to convergence, it can be put into production use to determine whether the target operator has a special operation certificate and whether the target operator is wearing protective equipment.

[0044] In some embodiments, extracting a segmentation map of the construction area corresponding to the target worker from images of a power transmission and transformation construction site includes: Step S401: Call the construction area segmentation model that has been trained to convergence to encode and decode the power transmission and transformation construction site image at multiple scales to obtain multiple image feature information that captures the contour features of the target workers in the image. Step S402: Extract the construction area segmentation map corresponding to the target worker from the power transmission and transformation construction site image based on the multiple image feature information.

[0045] In a specific embodiment, extracting the construction area segmentation map corresponding to the target worker from the power transmission and transformation construction site image includes: Step S4001: Through multi-level encoding of the construction area segmentation model that has been trained to convergence, the original image of the power transmission and transformation construction site image is scaled down step by step, and intermediate feature information corresponding to each scale is generated accordingly. The intermediate feature information is used to characterize the contour features of the construction area segmentation map corresponding to the target workers in the power transmission and transformation construction site image. Step S4002: The construction area segmentation model, which has been trained to convergence, is decoded in multiple levels. Based on the image feature information produced by the intermediate feature information at the smallest scale, the intermediate feature information produced by the same level of encoding is used as reference information to decode higher scale image feature information. The image feature information is used to represent the contour features of the construction area segmentation map corresponding to the target workers in the power transmission and transformation construction site image in the form of a mask. Step S4003: The construction area segmentation model performs image segmentation on the original image of the power transmission and transformation construction site image based on the mask image data fused from all image feature information, so as to extract the construction area segmentation map corresponding to the target workers from the power transmission and transformation construction site image.

[0046] As can be seen from steps S4001 to S4003 above, the multi-level encoding process reduces the image scale step by step, deeply mining the multi-scale contour features of the target workers, and avoiding interference from background elements such as dense equipment, cables, and complex terrain at the power transmission and transformation construction site; the decoding stage combines the intermediate feature information of the same level encoding to restore the target details, ensuring that the segmentation map completely covers key areas such as the workers' limbs and hand tools, thus solving the problems of blurred target edges and incomplete segmentation in traditional segmentation.

[0047] Furthermore, the encoding and decoding process covers the full scale range from distant small targets (such as workers under high-voltage towers) to close-up large targets (such as operators next to equipment). Through multi-scale feature fusion, it ensures that accurate segmentation results can be stably output under different work scenarios and shooting distances, adapting to the scene characteristics of power transmission and transformation construction sites with "multiple work points and strong personnel mobility".

[0048] Furthermore, the final generated construction area segmentation map retains only the core area of ​​the target workers, eliminating the interference of background noise on subsequent violation detection models (such as Mask-RCNN). This allows the model to focus on the identification of key features such as identification badges and the wearing position of protective equipment, significantly improving the accuracy of qualification verification and protective equipment inspection, and providing a reliable basis for hierarchical electrical control.

[0049] In some embodiments, image recognition is performed on the construction area segmentation map based on a converged work violation detection model to determine whether the target worker possesses a special operation certificate and whether the target worker is wearing protective equipment, including: Step S40001: Call the operation violation detection model that has been trained to convergence, and extract features from the construction area segmentation map through a convolutional backbone network to obtain feature maps of different scales in multiple layers; Step S40002: Generate multiple candidate regions of interest (ROIs) for the feature maps of different scales in the multi-layered structure using a region generation network, and perform an ROI alignment operation on the candidate ROIs to obtain an aligned feature map; Step S40003: The feature map aligned with the region of interest is processed by the detection head network through bounding box regression, recognition processing and mask map prediction. The bounding box regression is used to accurately locate the target worker area in the construction area segmentation map. The mask map prediction is used to perform instance segmentation of the target worker area. The recognition processing is used to extract deep features from the instance segmented target worker area. Step S40004: Based on the results of the identification process, extract the document identification features, document wearing features, and document display features of the special operation certificate of the target operator, and output the determination result of whether the target operator has a special operation certificate; at the same time, extract the wearing position and wearing standard features of the protective equipment of the target operator, and output the determination result of whether the target operator is wearing protective equipment.

[0050] As can be seen from steps S40001 to S40004 above, by extracting multi-layer feature maps of different scales through a convolutional backbone network, it is possible to cover the differences between the distant and near views of workers in the power transmission and transformation construction site, as well as the differences in the size of their certificates and protective equipment. This avoids the problem of missing small targets and incomplete features of large targets caused by single-scale feature extraction, and ensures the comprehensive capture of target features in complex scenarios.

[0051] Furthermore, by using a region generation network to filter candidate regions of interest and perform alignment operations, the system can accurately locate the areas where workers, their documents, and protective equipment are located, eliminate minor background noise (such as equipment edges and cable fragments) remaining in the construction area segmentation map, reduce interference from irrelevant areas to subsequent identification, and improve identification efficiency and accuracy.

[0052] By working collaboratively through three branches—boundary box regression, recognition processing, and mask image prediction—this system can accurately locate the overall area of ​​the worker using bounding boxes, refine body parts (head, hands, torso) through instance segmentation, and extract key identification information through deep feature extraction. This avoids the accuracy loss of the traditional "first locate, then recognize" process, enabling multiple tasks to be completed in one step, and adapting to the response requirements of real-time management of intelligent power distribution boxes.

[0053] Furthermore, fine-grained features are extracted for qualification verification and protection inspection respectively. Qualification judgment focuses on certificate identification, wearing status, and display status, while protection judgment focuses on wearing location and standardization. This can effectively distinguish between ambiguous scenarios such as "certificate holder not displaying" and "improper wearing" to avoid misjudgment. At the same time, it forms data linkage with the "certificate query" and "face management" functions of the equivalent diversion cloud management system to further enhance the reliability of the judgment results and provide accurate basis for subsequent graded electrical control.

[0054] Step S50: When the target operator does not have a special operation certificate and / or is not wearing protective equipment, the intelligent distribution box performs preset electrical control operations according to the type of violation, so as to carry out intelligent monitoring and power distribution management at the power transmission and transformation construction site.

[0055] Based on a converged construction area segmentation model, the construction area segmentation map corresponding to the target worker is extracted from the power transmission and transformation construction site image. Based on a converged operation violation detection model, image recognition is performed on the construction area segmentation map to determine whether the target worker has a special operation certificate and whether the target worker is wearing protective equipment. If the target worker does not have a special operation certificate and / or is not wearing protective equipment, the intelligent distribution box executes preset electrical control operations according to the violation type to carry out intelligent monitoring and power distribution management of the power transmission and transformation construction site.

[0056] In some embodiments, when the target worker does not possess a special operation certificate and / or is not wearing protective equipment, the intelligent distribution box executes preset electrical control operations based on the type of violation to perform intelligent monitoring and power distribution management at the power transmission and transformation construction site, including: Step S501: When it is determined that the target worker at the target special operation location of the power transmission and transformation construction site does not have a special operation certificate and / or is not wearing protective equipment in accordance with regulations, a corresponding violation judgment signal is output; the intelligent distribution box deployed at the power transmission and transformation construction site receives the violation judgment signal and identifies the violation type corresponding to the violation judgment signal, wherein the violation type includes only not wearing protective equipment in accordance with regulations, only not having a special operation certificate, and neither having a special operation certificate nor wearing protective equipment in accordance with regulations, and the protective equipment includes insulating gloves, protective face mask, and insulating shoes; Step S502: The intelligent distribution box performs a matching preset electrical control operation on the power supply circuit corresponding to the target special operation location according to the correspondence between the violation type and the electrical control operation, so as to carry out intelligent monitoring and power distribution management of the power transmission and transformation construction site. The electrical control operation includes one or more combinations of current limiting and capacity reduction of the power supply circuit, instantaneous power outage and linkage audible and visual alarm.

[0057] In a specific embodiment, the intelligent distribution box performs pre-set electrical control operations matching the type of violation and the corresponding electrical control operations on the power supply circuit corresponding to the target special operation location, in order to carry out intelligent monitoring and power distribution management of the power transmission and transformation construction site, including: Step S5001: When the violation judgment signal is only that the protective equipment is not worn properly, perform electrical control operations such as current limiting and capacity reduction and trigger audible and visual alarms. Step S5002: When the violation judgment signal is that the operator only lacks a special operation certificate, perform an electrical control operation to momentarily cut off the power supply circuit and trigger an audible and visual alarm. Step S5003: When a person does not have a special operation certificate and does not wear protective equipment in accordance with regulations, perform electrical control operations such as instantaneous power outage of the power supply circuit, linkage of audible and visual alarms, and locking of closing authority.

[0058] As can be seen from the above embodiments, the classification of violations includes only not wearing protective equipment properly, only not possessing a special operation certificate, and neither possessing a special operation certificate nor wearing protective equipment properly, and the limitation of specific types of protective equipment, avoiding the drawbacks of the traditional "one-size-fits-all" approach to control. After the intelligent distribution box accurately identifies the type of violation, it ensures that subsequent control operations are accurately matched with the risk level, solving the problem of the crude control of the same type of violation risk.

[0059] Based on the correspondence between violation types and control operations, a combination of operations such as current limiting and capacity reduction, instantaneous power outage, and linkage of audible and visual alarms are implemented. For low-risk violations (only failure to wear protective equipment), electrical control operations of current limiting and capacity reduction with linkage of audible and visual alarms are implemented, which both warns of violations and does not interrupt construction. For high-risk violations (lack of special operation certificate, lack of special operation certificate and failure to wear protective equipment properly), electrical control operations of instantaneous power outage of the power supply circuit, linkage of audible and visual alarms, and locking of closing authority are implemented, eliminating serious safety hazards and achieving a balance between "not breaching the safety bottom line and not being redundant in construction efficiency".

[0060] As can be seen from the above embodiments, compared with the prior art, this application addresses the problems of the prior art, which relies heavily on the experience of management personnel or a single electrical parameter threshold to determine the risk of electric shock, does not fully consider the spatial clustering characteristics and frequency accumulation effect of historical violations, and does not fully consider whether the operators have special operation certificates and protective equipment verification, mainly relying on on-site management personnel to complete the inspection. This application has the following beneficial effects, including but not limited to: Firstly, this application integrates the coordinates and frequency of historical violations, and constructs an electric shock risk assessment model by combining Gaussian kernel density estimation. This model accurately quantifies the risk value of each special operation location, captures the spatial clustering characteristics and frequency accumulation effect of violation risks, and solves the drawbacks of "one-size-fits-all" control.

[0061] Risk assessment results directly guide the allocation of monitoring resources, and subsequent image acquisition and verification are only initiated for target areas with risk values ​​exceeding the threshold, avoiding resource waste caused by ineffective monitoring, and achieving "precise prevention and control in key areas and simplified management and control in low-risk areas", which is suitable for the differentiated management needs of multiple special operation sites in power transmission and transformation construction sites.

[0062] Secondly, this application first uses the U²-Net model to remove the complex background of dense equipment and cables at the construction site, extracting a clean segmentation map of the worker areas. Then, it uses the Mask-RCNN model to accurately identify the special operation certificate qualifications and the wearing status of protective equipment, solving the pain points of missed detections and misjudgments in traditional manual inspections and the interference of background in single model recognition. It realizes automated, all-weather monitoring of qualification verification and protection inspection, eliminating the need for manual on-site verification, significantly reducing labor costs. At the same time, through fine-grained feature extraction, it effectively distinguishes ambiguous scenarios such as "certificate holder not displaying" and "improper wearing," improving the precision of violation judgment.

[0063] Thirdly, this application identifies various types of violations, such as simply not wearing protective equipment properly, simply lacking a special operation certificate, or lacking both a special operation certificate and not wearing protective equipment properly. It then links the intelligent distribution box to perform differentiated control actions, including current limiting and capacity reduction, instantaneous power outage, locking of closing permissions, and audible and visual alarms. This not only eliminates serious safety hazards but also avoids indiscriminate power outages that interfere with normal construction, achieving a balance between safety standards and construction continuity.

[0064] Fourth, the Internet-based intelligent monitoring and power distribution management method for power transmission and transformation construction sites proposed in this application can be directly implemented based on the digital distribution boxes, intelligent cameras, and cloud management systems already deployed at the power transmission and transformation construction sites without large-scale modifications. At the same time, it supports the dynamic adjustment of risk thresholds and control parameters and can adapt to the personalized needs of different special operation scenarios such as high-voltage wiring, tower erection, and welding operations.

[0065] A digital distribution box, proposed for another purpose of this application, is applied to the Internet-based intelligent monitoring and distribution management method for power transmission and transformation construction sites as described in any of the above claims, comprising: The violation dataset acquisition module 1100 is used to acquire the historical operation violation dataset corresponding to the power transmission and transformation construction site. The power transmission and transformation construction site includes multiple special operation locations, and the historical operation violation dataset includes multiple historical operation violation data points. Each historical operation violation data point contains the coordinates of the historical violation location where each type of operation violation event occurred and the operation violation frequency of each type of operation violation event in the coordinates of the historical violation location. The electric shock risk assessment module 1200 is used to construct an electric shock risk assessment model based on the coordinates of the historical violation locations and the frequency of the operation violation using Gaussian kernel density estimation. Based on the electric shock risk assessment model, the module determines the operation violation risk value corresponding to each special operation location in the power transmission and transformation construction site. The operation violation risk value is used to characterize the degree of risk of operation violation and electric shock accident at each special operation location. The on-site image acquisition module 1300 is used to acquire images of the power transmission and transformation construction site containing the target workers when the risk value of the violation of the operation corresponding to the target special operation site is greater than a preset risk threshold. The operation violation judgment module 1400 is used to extract the construction area segmentation map corresponding to the target operator from the power transmission and transformation construction site image based on the training-to-converge construction area segmentation model, and to perform image recognition on the construction area segmentation map based on the training-to-converge operation violation detection model to determine whether the target operator has a special operation certificate and whether the target operator is wearing protective equipment. The power distribution electrical control module 1500 is used to perform preset electrical control operations based on the type of violation when the target operator does not have a special operation certificate and / or is not wearing protective equipment, so as to carry out intelligent monitoring and power distribution management at the power transmission and transformation construction site.

[0066] The above description is only a partial embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

[0067] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for intelligent monitoring and power distribution management at power transmission and transformation construction sites based on the Internet, characterized in that, include: Obtain a historical operation violation dataset corresponding to the power transmission and transformation construction site. The power transmission and transformation construction site includes multiple special operation locations. The historical operation violation dataset includes multiple historical operation violation data points. Each historical operation violation data point contains the coordinates of the historical violation location where each type of operation violation event occurred and the operation violation frequency of each type of operation violation event at the coordinates of the historical violation location. Gaussian kernel density estimation is used to construct an electric shock risk assessment model based on the coordinates of the historical violation locations and the frequency of the operation violation. The operation violation risk value corresponding to each special operation location in the power transmission and transformation construction site is determined according to the electric shock risk assessment model. The operation violation risk value is used to characterize the degree of risk of operation violation and electric shock accident at each special operation location. When the risk value of violation of the target special operation location is greater than the preset risk threshold, the power transmission and transformation construction site image containing the target workers is obtained at the target special operation location; Based on the converged construction area segmentation model, the construction area segmentation map corresponding to the target worker is extracted from the power transmission and transformation construction site image. Based on the converged operation violation detection model, the construction area segmentation map is used for image recognition to determine whether the target worker has a special operation certificate and whether the target worker is wearing protective equipment. When the target operator does not have a special operation certificate and / or is not wearing protective equipment, the intelligent distribution box will perform preset electrical control operations according to the type of violation, so as to carry out intelligent monitoring and power distribution management at the power transmission and transformation construction site.

2. The method for intelligent monitoring and power distribution management at power transmission and transformation construction sites based on the Internet, as described in claim 1, is characterized in that... Extracting the construction area segmentation map corresponding to the target workers from the power transmission and transformation construction site images, including: The trained and converged construction area segmentation model is invoked to encode and decode the power transmission and transformation construction site images at multiple scales to obtain multiple image feature information that captures the contour features of the target workers. Based on the multiple image feature information, the construction area segmentation map corresponding to the target worker is extracted from the power transmission and transformation construction site image.

3. The method for intelligent monitoring and power distribution management at power transmission and transformation construction sites based on the Internet, as described in claim 2, is characterized in that... Extracting the construction area segmentation map corresponding to the target workers from the power transmission and transformation construction site images, including: The multi-level encoding of the construction area segmentation model, which has been trained to convergence, reduces the scale of the original image of the power transmission and transformation construction site image step by step, and generates intermediate feature information corresponding to each scale. The intermediate feature information is used to characterize the contour features of the construction area segmentation map corresponding to the target workers in the power transmission and transformation construction site image. The construction area segmentation model, which has been trained to convergence, is decoded in multiple levels. Based on the image feature information produced by the intermediate feature information at the smallest scale, the intermediate feature information produced by the same level encoding is used as reference information to decode higher scale image feature information. The image feature information is used to represent the contour features of the construction area segmentation map corresponding to the target workers in the power transmission and transformation construction site image in the form of a mask. The original image of the power transmission and transformation construction site is segmented using a masked image data fused from all image feature information by a construction area segmentation model, in order to extract the construction area segmentation map corresponding to the target workers from the power transmission and transformation construction site image.

4. The method for intelligent monitoring and power distribution management at power transmission and transformation construction sites based on the Internet, as described in claim 1, is characterized in that... Image recognition is performed on the construction area segmentation map based on a converged operational violation detection model to determine whether the target workers possess special operation certificates and whether they are wearing protective equipment, including: The pre-trained operation violation detection model is invoked, and features are extracted from the construction area segmentation map through a convolutional backbone network to obtain multi-layer feature maps of different scales. A region generation network is used to generate multiple candidate regions of interest for the feature maps of different scales in the multi-layer structure. A region of interest alignment operation is performed on the candidate regions of interest to obtain the aligned feature map. The detection head network performs bounding box regression, recognition, and mask image prediction on the feature map aligned with the region of interest. The bounding box regression is used to accurately locate the target worker region in the construction area segmentation map. The mask image prediction is used to perform instance segmentation of the target worker region. The recognition is used to extract deep features from the segmented target worker region. Based on the results of the identification process, the document identification features, document wearing features, and document display features of the special operation certificate of the target operator are extracted, and the determination result of whether the target operator possesses a special operation certificate is output. At the same time, the wearing position and wearing standard features of the protective equipment of the target operator are extracted, and the determination result of whether the target operator is wearing protective equipment is output.

5. The method for intelligent monitoring and power distribution management at power transmission and transformation construction sites based on the Internet, as described in claim 1, is characterized in that... Constructing an electric shock risk assessment model includes: Based on Gaussian kernel density estimation, the square of the Euclidean distance between the monitoring coordinates corresponding to the special operation location at the power transmission and transformation construction site and the coordinates of the historical violation location is calculated, and the first ratio between the square of the Euclidean distance and twice the square of the bandwidth parameter is calculated. The natural exponential function value of the first ratio is calculated and determined. The first product between the natural exponential function value and the frequency of work violations is calculated and determined. At the same time, the normalized values ​​of the electrical operating parameters, equipment operating parameters and environmental parameters of the power transmission and transformation construction site are integrated. The weighted value of each normalized value and the corresponding weight coefficient is calculated. The first product is multiplied by each weighted value and summed to determine the risk value of work violations, so as to construct an electric shock risk assessment model.

6. The method for intelligent monitoring and power distribution management at power transmission and transformation construction sites based on the Internet, as described in claim 5, is characterized in that... Based on the electric shock risk assessment model, the corresponding work violation risk value for each special operation location in the power transmission and transformation construction site is determined, including: The monitoring coordinates, real-time electrical operating parameters, real-time equipment operating parameters, and real-time environmental parameters of each special operation site in the power transmission and transformation construction site are collected. The real-time electrical operating parameters, real-time equipment operating parameters, and real-time environmental parameters are normalized to obtain the normalized values ​​corresponding to the real-time electrical operating parameters, the real-time equipment operating parameters, and the real-time environmental parameters. The monitoring coordinates and normalized values ​​of each special operation location are input into the electric shock risk assessment model to calculate the quantitative value corresponding to each special operation location. The quantified value is used as the corresponding risk value for operational violations at each special operation site in the power transmission and transformation construction site.

7. The method for intelligent monitoring and power distribution management at power transmission and transformation construction sites based on the Internet, as described in claim 1, is characterized in that... When the target operator does not possess a special operation certificate and / or is not wearing protective equipment, the intelligent distribution box executes preset electrical control operations according to the type of violation, in order to conduct intelligent monitoring and power distribution management at the power transmission and transformation construction site, including: When it is determined that the target workers at the special operation site of the power transmission and transformation construction site do not have a special operation certificate and / or are not wearing protective equipment in accordance with regulations, a corresponding violation judgment signal will be output. The intelligent distribution box deployed at the power transmission and transformation construction site receives the violation judgment signal and identifies the violation type corresponding to the violation judgment signal. The violation type includes only not wearing protective equipment properly, only not having a special operation certificate, and neither having a special operation certificate nor wearing protective equipment properly. The protective equipment includes insulating gloves, protective face mask, and insulating shoes. The intelligent distribution box performs pre-set electrical control operations on the power supply circuit corresponding to the target special operation location based on the correspondence between the violation type and the electrical control operation, so as to carry out intelligent monitoring and power distribution management of the power transmission and transformation construction site. The electrical control operation includes one or more combinations of current limiting and capacity reduction of the power supply circuit, instantaneous power outage and linkage audible and visual alarm.

8. The method for intelligent monitoring and power distribution management at power transmission and transformation construction sites based on the Internet, as described in claim 1, is characterized in that... The intelligent distribution box, based on the correspondence between violation types and electrical control operations, executes pre-set electrical control operations matching the power supply circuits corresponding to the target special operation location, thereby enabling intelligent monitoring and power distribution management at the power transmission and transformation construction site, including: When the violation judgment signal is only the failure to wear protective equipment properly, electrical control operations are performed to limit current and reduce capacity, and trigger audible and visual alarms. When the violation judgment signal is that the person only lacks a special operation certificate, the electrical control operation is to perform an instantaneous power outage of the power supply circuit and trigger an audible and visual alarm. When someone does not have a special operation certificate and does not wear protective equipment properly, they shall perform electrical control operations that involve momentarily cutting off the power supply circuit, triggering audible and visual alarms, and locking the closing authority.

9. The method for intelligent monitoring and power distribution management at power transmission and transformation construction sites based on the Internet, as described in any one of claims 1, is characterized in that... The basic network architecture of the construction area segmentation model is the U²-Net model; the basic network architecture of the operation violation detection model includes the Mask-RCNN model.

10. An Internet-based intelligent monitoring and distribution management system for power transmission and transformation construction sites, employing the Internet-based intelligent monitoring and distribution management method for power transmission and transformation construction sites as described in any one of claims 1 to 9, comprising: The violation dataset acquisition module is used to acquire historical operation violation datasets corresponding to power transmission and transformation construction sites. The power transmission and transformation construction sites include multiple special operation locations, and the historical operation violation datasets include multiple historical operation violation data points. Each historical operation violation data point contains the coordinates of the historical violation location where each type of operation violation event occurred and the operation violation frequency of each type of operation violation event at the coordinates of the historical violation location. The electric shock risk assessment module is used to construct an electric shock risk assessment model based on the coordinates of the historical violation locations and the frequency of the operation violation using Gaussian kernel density estimation. Based on the electric shock risk assessment model, the module determines the operation violation risk value corresponding to each special operation location in the power transmission and transformation construction site. The operation violation risk value is used to characterize the degree of risk of operation violation and electric shock accident at each special operation location. The on-site image acquisition module is used to acquire images of the power transmission and transformation construction site containing the target workers when the risk value of the violation of the target special operation site is greater than a preset risk threshold. The operation violation determination module is used to extract the construction area segmentation map corresponding to the target worker from the power transmission and transformation construction site image based on the training-to-converge construction area segmentation model, and to perform image recognition on the construction area segmentation map based on the training-to-converge operation violation detection model to determine whether the target worker has a special operation certificate and whether the target worker is wearing protective equipment. The power distribution electrical control module is used to perform preset electrical control operations based on the type of violation when the target operator does not have a special operation certificate and / or is not wearing protective equipment, so as to carry out intelligent monitoring and power distribution management at the power transmission and transformation construction site.

Images