Digital intelligent supervision platform for electro-gas welding operation
Through the design of the digital intelligent supervision platform for electrical welding operations, using technologies such as face verification, environmental detection and positioning chips, the problems of safety hazards and low supervision efficiency in electrical welding operations are solved, and strict supervision and safety of electrical welding operations are achieved. Management significantly improves security and regulatory efficiency.
Patent Information
- Application Number
- CN202411087000.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-08
- Publication Date
- 2024-11-29
AI Technical Summary
There are safety hazards of improper operation or equipment failure in electrical welding operations, and the traditional supervision methods are inefficient, difficult to effectively prevent fire accidents, and there is a lack of strict supervision of the qualifications of operators and the environment.
A digital intelligent supervision platform for electrical welding operations was designed, including user management module, filing management module, approval management module, equipment management module and early warning management module. The technology of face verification, environmental detection, positioning chip and photosensitive chip is adopted. Strict qualification review, real-time monitoring and intelligent early warning to ensure safety of operations.
Through strict qualification review and real-time monitoring, the safety of electrical welding operations is significantly improved, the compliance of operation locations and equipment is ensured, and all-round supervision of electrical welding operations is achieved, reducing the risk of accidents and the occurrence of human errors.
Smart Images

Figure CN119048000A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of welding technology, and in particular to a digital intelligent monitoring platform for electric welding operations. Background Art
[0002] Electric welding plays an important role in industrial production. However, electric welding operations usually involve factors such as high temperature, electric arc and flammable materials. Improper operation or equipment failure may lead to fire and explosion, posing a serious threat to the safety of life and property of workers performing hot work. Traditional electric welding operation supervision methods are often limited to manual inspections and manual records. The supervision efficiency is low and it is prone to missed inspections and false inspections, making it difficult to effectively prevent fire accidents. In addition, the electric welding industry generally has a master-apprentice mechanism, and some practitioners have a weak awareness of holding a certificate before taking up their posts, making it difficult for relevant regulatory authorities to supervise uniformly. In view of the problems that electric welding operations are prone to fires and traditional management methods are inefficient, there is an urgent need for a more intelligent and efficient means of supervision.
[0003] With the rapid development of science and technology and industrial intelligence, the digital management of electric welding operations has become an important trend in the industrial field, and many companies and organizations are actively promoting the implementation of digital management. At present, some intelligent supervision platforms or applications for electric welding operations have emerged, which provide important support for the safety management and production efficiency of electric welding operations, and effectively improve the safety level of electric welding operations, but there are still shortcomings. Specifically, in terms of personnel management, although the existing platform provides supervision of electric welding operations, there is no strict qualification review of hot work personnel, which means that it is impossible to determine whether the hot work personnel have the necessary qualifications and skills, and there are safety hazards. Secondly, in terms of work site management, there is a lack of positioning supervision of the work environment, and it is impossible to ensure that the applied work site is consistent with the actual work site, and there is a possibility of false declaration of location information; at the same time, there is a lack of intelligent supervision of the construction environment, and it is difficult to eliminate the combustible and flammable hazards that may exist in the construction environment. In addition, in terms of supervision methods, the existing platform mainly focuses on supervision during operation, lacks an effective supervision mechanism in advance, and it is difficult to timely discover and prevent potential safety risks. The supervision effect needs to be further improved. Summary of the invention
[0004] In view of the deficiencies of the prior art, the object of the present invention is to provide a digital intelligent supervision platform for electric welding operations, comprising:
[0005] User management module, which has four system identities: hot work personnel, enterprise managers, fire emergency management personnel and fire law enforcement personnel. Users use it according to their own identities. Hot work personnel are responsible for the actual operation of electric welding operations. Enterprises are registered units of the platform and have management rights of the platform. Enterprise managers are responsible for managing users and related operations within the enterprise. Fire emergency management personnel are professionals responsible for supervising and managing the safety of electric welding operations. Fire law enforcement personnel are personnel from relevant departments responsible for law enforcement supervision and inspection of electric welding operations;
[0006] Filing management module, which is responsible for receiving qualification documents of hot work personnel for filing and providing approval process for filing applications;
[0007] Approval management module, which includes four submodules: face verification submodule, data receiving submodule, environment detection submodule and review authorization submodule. The face verification submodule is responsible for verifying whether the information of the personnel applying for hot work is consistent with the information filed in the system. The data receiving submodule is responsible for receiving and verifying the location, photos and video data of the construction environment submitted by the personnel doing hot work. The environment detection submodule is responsible for verifying whether the location submitted by the personnel doing hot work is in the specified construction area, and whether there are combustibles and flammable and explosive risks in the photos and videos. The review authorization submodule is responsible for reviewing the application submitted by the personnel doing hot work;
[0008] The equipment management module includes a positioning chip, a charging chip and a light sensing chip installed in the electric welding equipment. The module obtains the location-related information of the electric welding equipment in real time through the positioning chip and GNSS satellite technology, monitors in real time whether the electric welding equipment deviates from the predetermined area, monitors in real time the battery charging status and working time of the equipment through the charging chip, detects whether there is any illegal operation, and monitors the light perception of the equipment through the light sensing chip to detect whether the equipment is disassembled;
[0009] The early warning management module includes working environment early warning, equipment status early warning and intelligent early warning. It is used to monitor abnormal situations that occur during the application and process of electric welding operations, and issue early warnings in time to remind relevant personnel to pay attention and take necessary measures.
[0010] Furthermore, the environmental detection submodule in the approval management module adopts a target detection algorithm to detect combustibles. By extracting the characteristics of various objects in the construction environment, it identifies and marks potential combustible targets in the construction environment, evaluates the potential fire and explosion hazards, and generates a hazard assessment report.
[0011] Furthermore, the target detection algorithm adds a CBAM (Convolutional Block Attention Module) attention mechanism that combines channel attention and spatial attention. The specific steps for detecting combustibles are as follows:
[0012] S301, standardizing the images collected from the working environment, including scaling to a size required for model input, and normalizing brightness and contrast to ensure consistency of the model input;
[0013] S302, input the preprocessed image into the YOLOv8 model, extract features through the backbone network of the model, and then use the feature maps of three different scales, P3, P4, and P5, to perform preliminary target detection to obtain candidate combustible target areas;
[0014] S303, using the channel attention module in CBAM, through the two branches of global average pooling and global maximum pooling, capturing the statistical information between channels, inputting the captured statistical information into two MLP (multi-layer perceptron) networks, obtaining the importance weight of each channel, multiplying the channel weight by the original feature channel, strengthening the feature representation of important channels, and suppressing unimportant channels;
[0015] S304, using global average pooling and global maximum pooling in the spatial dimension to capture spatial position information, and converting the spatial information into a spatial attention map through 1x1 convolution and ReLU activation function, and finally applying the spatial attention map to each channel of the feature map to enhance the feature response of the key area;
[0016] S305, fusing the feature maps processed by channel attention and spatial attention to obtain optimized feature representation, and inputting the fused feature maps into the detection head of the YOLOv8 model to perform more refined object detection and bounding box regression;
[0017] S306, encapsulating the final detection result, including the bounding box coordinates, the predicted category and the confidence information;
[0018] S307: According to the feedback results, the model structure, attention mechanism parameters and training strategy are adjusted to further improve the detection performance.
[0019] Furthermore, in the user management module, enterprise managers are responsible for managing users within the enterprise and related operations including adding, deleting, and modifying user information, viewing the electric welding operations within the enterprise through the platform, and performing data analysis and report generation operations; fire emergency management personnel view the details of electric welding operations within their jurisdiction, regional data overviews, review of declaration-related information, and alarm records of illegal operations through the platform; fire law enforcement personnel view the filing status and approval records of electric welding operations through the platform, conduct on-site inspections and assessments, and handle and punish illegal behaviors.
[0020] Furthermore, the filing management module also provides real-time tracking and management functions for the filing application status, so that hot work personnel and management personnel can understand the processing progress and results of the filing application at any time.
[0021] Furthermore, the face verification submodule in the approval management module adopts face recognition technology and liveness detection technology based on convolutional neural network (CNN) to analyze the feature points and facial contour information of the face, and compare and verify the information with the pre-established hot work personnel information database. If the verification is successful, the system will immediately feedback to the person applying for the hot work and allow the person applying for the hot work to further operate. If the verification fails, the person applying for the hot work will be prompted to re-upload the identity information or contact the relevant management personnel for processing.
[0022] Furthermore, the data receiving submodule in the approval management module is connected to the communication interface of the platform. Once the hot work personnel upload the data through the platform, the data receiving submodule will respond immediately and receive the data. The data receiving submodule is compatible with different image and video formats, and is equipped with an exception handling mechanism, which can detect the uploaded data abnormalities in time and notify the personnel applying for the hot work to handle it.
[0023] Furthermore, in the device management module, the acquired location-related information is calibrated and updated through convolutional neural networks and differential positioning technology.
[0024] Further, the calibrating the acquired location-related information specifically includes the following steps:
[0025] S401, using the Beidou positioning system, obtain the specific positioning of the electric welding construction area and multiple measurements of each electric welding equipment, and obtain the initial GNSS data set required by the network model. The data includes the equipment longitude, latitude, altitude, pseudorange, and deviation information. The data is processed by least squares and loaded into the input layer of the model in sequence;
[0026] S402, setting a convolution kernel of appropriate size and using a convolutional neural network to extract features from the GNSS data;
[0027] S403, through the loss layer and the fully connected layer, the model is helped to converge, the output result is obtained, the model training is completed, and the appropriate noise parameters are obtained;
[0028] S404, using Kalman filtering combined with noise parameters output by the network to correct the predicted value of the electric welding equipment position;
[0029] S405, calculate the error between the predicted value and the true value of the filter output, use the error evaluation score to back-propagate to the network, adjust the network weights, and optimize the network model. After repeated adjustments and updates, the optimal solution is obtained;
[0030] S406, obtaining appropriate noise parameters, and inferring the position information according to the noise parameters to obtain corrected position information.
[0031] Furthermore, in the working environment warning, when combustible materials are detected in the working environment, a warning is issued to the approval personnel to remind them to reject the application for hot work, and the information is fed back to the person applying for the hot work; in the equipment status warning, based on the monitoring of the equipment by the equipment management module, a warning is issued for not moving the electric welding equipment according to regulations, turning on the electric welding equipment, and disassembling the electric welding equipment; in the intelligent warning, big data and artificial intelligence technology are combined to perform intelligent approval and intelligent warning.
[0032] Compared with the prior art, the present invention has the following beneficial effects:
[0033] 1. Hot work personnel must pass strict qualification review and work application review before they can obtain work permits. Both qualification review and work application review require strict information screening, comparison, face verification, liveness verification and construction environment testing, and must be approved by auditors and supervisors at all levels to effectively ensure the professional quality and skill level of hot work personnel, reduce illegal operations such as impersonation and unlicensed operations, and significantly improve the safety of electric welding operations.
[0034] 2. By installing positioning chips, charging chips and light-sensing chips on electric welding equipment, the consistency between the applied work location and the actual work location can be ensured, the compliance of the electric welding operation can be ensured, the integrity of the electric welding equipment can be ensured, and reliable technical support can be provided for the safe management of electric welding operations.
[0035] 3. Establish a complete supervision mechanism throughout the entire process of electric welding operations, including pre-qualification review and filing, and in-process safety supervision and early warning, to ensure all-round control and real-time supervision of electric welding operations. This pre-emptive management strategy can timely discover and prevent potential risks before the operation, and continuously track and monitor the operation status during the operation, which maximizes the supervision efficiency and safety of electric welding operations, and lays a solid foundation for the safety management of electric welding operations.
[0036] 4. The platform integrates advanced positioning technology, Internet of Things technology, and artificial intelligence algorithms. It can conduct intelligent review of the data in the environmental hazard assessment report and realize automatic assessment of the safety of the electric welding operation environment. At the same time, it combines historical warning data with artificial intelligence to achieve timely warning, greatly improving approval efficiency and reducing the risk of accidents and human errors. BRIEF DESCRIPTION OF THE DRAWINGS
[0037] In order to more clearly illustrate the specific implementation of the present invention or the technical solution in the prior art, the drawings required for describing the specific implementation or the prior art will be briefly introduced below. Obviously, the drawings described below are only one implementation of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0038] Figure 1 It is an architecture diagram of a digital intelligent supervision platform for electric welding operations;
[0039] Figure 2 It is a schematic diagram of a user management module of a digital intelligent supervision platform for electric welding operations;
[0040] Figure 3 It is a schematic diagram of the filing management module of the digital intelligent supervision platform for electric welding operations;
[0041] Figure 4 It is a schematic diagram of the approval management module of the digital intelligent supervision platform for electric welding operations;
[0042] Figure 5 It is a working diagram of the face verification submodule in the approval management module of the digital intelligent supervision platform for electric welding operations;
[0043] Figure 6 It is a working diagram of a data receiving submodule in the approval management module of a digital intelligent supervision platform for electric welding operations;
[0044] Figure 7 It is a working diagram of the environmental detection submodule in the approval management module of the digital intelligent supervision platform for electric welding operations;
[0045] Figure 8 It is a working diagram of the review and authorization submodule in the approval management module of the digital intelligent supervision platform for electric welding operations;
[0046] Fig. 9 It is a schematic diagram of the equipment management module of a digital intelligent supervision platform for electric welding operations;
[0047] Fig.10It is a schematic diagram of positioning work in the equipment management module of the digital intelligent supervision platform for electric welding operations;
[0048] Fig.11 It is a schematic diagram of the early warning management module of the digital intelligent supervision platform for electric welding operations. DETAILED DESCRIPTION
[0049] In order to make the technical means, creative features, objectives and effects achieved by the present invention easy to understand, the technical scheme in the specific implementation mode of the present invention is clearly and completely described below to further illustrate the present invention. Obviously, the specific implementation mode described is only a part of the implementation mode of the present invention, rather than all styles.
[0050] like Figure 1 As shown, a digital intelligent supervision platform for electric welding operations includes a user management module, a filing management module, an approval management module, an equipment management module and an early warning management module.
[0051] like Figure 2 As shown in the figure, the user management module of a digital intelligent supervision platform for electric welding operations has four identities: hot work personnel, enterprise managers, fire emergency management personnel and fire law enforcement personnel. Users use or manage according to their own identities. Among them, hot work personnel are operators responsible for actually performing electric welding operations. They submit the filing application for electric welding operations through the platform and upload necessary materials such as relevant qualification certificates and license certificates. After the filing application is approved, the hot work personnel obtain the corresponding operation authority and can perform electric welding operations at the designated time and place. Enterprise managers are responsible for managing users and related operations within the enterprise, including adding, deleting, and modifying user information, and view the electric welding operations within the enterprise through the platform, and perform data analysis and report generation. Fire emergency management personnel are professionals responsible for supervising and managing the safety of electric welding operations. Through the platform, they can view the details of electric welding operations within the jurisdiction, regional data overviews, review of relevant information reported, and alarm records of illegal operations. Fire law enforcement personnel are department personnel responsible for law enforcement supervision and inspection of electric welding operations. Through the platform, they can view the filing status and approval records of electric welding operations, conduct on-site inspections and assessments, and handle and punish violations.
[0052] like Figure 3As shown, a filing management module of a digital intelligent supervision platform for electric welding operations is used to manage and supervise the filing application and qualification approval process for electric welding operations. Personnel performing hot work operations need to fill in relevant information according to the platform requirements, such as qualification certificates, special operation certificates, etc. The system uses OCR recognition to extract data such as name, photo, certificate number, etc. from the certificate, compares it with the applicant's information, and submits the qualification application information to the approval personnel. The approval personnel will review and approve it step by step, and the approval results will be fed back to the applicant. In this module, all passed qualification filing information will be stored in the database for subsequent query and use.
[0053] like Figure 4 As shown in the figure, an approval management module of a digital intelligent supervision platform for electric welding operations is used to approve and process electric welding operation applications submitted by personnel engaged in hot work. Personnel engaged in hot work conduct facial verification through the platform, and determine the identity of the applicant through means such as face detection, alignment, and feature extraction. After the facial verification is passed, fill in the electric welding operation application, submit construction environment photos, construction environment videos, construction environment positioning, and the equipment required for application, etc. If the construction environment positioning is not in the specified area, an early warning is directly generated, requiring the personnel engaged in hot work to reposition in the designated operation area. The platform uses target detection technology to detect construction environment photos and videos, determine whether there are combustibles in the environment, generate a hazard assessment report based on the detection results, mark the possible fire or explosion risks in the construction environment, and provide the hazard assessment report and application form to the approval personnel. The approval personnel comprehensively consider the hazard assessment report and application information, approve or reject the application, and assign different color codes according to different results. Among them, if the operation application is approved, a green code is assigned, indicating that hot work can be started; if the application is rejected, a gray code is assigned, indicating that hot work cannot be started. Finally, the approval history of the registration application is recorded, including approval opinions, approval time, approver and other information, and the approval record query and statistical analysis functions are provided to provide decision-making reference for managers.
[0054] The approval management module includes a face verification submodule, a data receiving submodule, an environment detection submodule, and an audit authorization submodule, as follows:
[0055] The working diagram of the face verification submodule is as follows: Figure 5As shown in the figure, after the hot work personnel use the equipment carrying the system to enter the system, they face the camera to take a face photo, and then perform the face verification operation according to the face verification instructions. If the face passes the verification, the system prompts that the verification is successful and allows the hot work personnel to enter the next step; if the face fails the verification, the system prompts the operator to re-verify. In order to realize the function of the face verification module, the module adopts the face recognition technology based on the convolutional neural network (CNN), which involves processes such as face detection, face alignment, and feature extraction, and compares and matches the extracted face features with the pre-entered operator face database to ensure the accuracy and reliability of the recognition results. At the same time, in order to enhance the security and credibility of the system, the module also applies liveness detection algorithms, including blink detection and mouth movement detection. During the face verification process, the system prompts the operator to perform operations such as blinking and opening the mouth to ensure that the operator is a real live person. The comprehensive face detection and liveness detection algorithms help to improve the accuracy and security of the face verification module and provide a reliable basic guarantee for subsequent operations.
[0056] The working diagram of the data receiving submodule is as follows: Figure 6 As shown in the figure, the specific workflow of this submodule is as follows:
[0057] S101, receiving data: receiving the positioning, photo and video data of the construction environment submitted by the hot work personnel through the system interface or network connection.
[0058] S102, data verification: After receiving the data, the data receiving module performs data verification. First, the positioning data of the construction environment is verified to ensure the accuracy and legitimacy of the location information. Next, the photo and video data are checked to verify their file format, size, and clarity, etc., to ensure the integrity and availability of the data. In addition, the photo and video data need to be parsed and their metadata information, such as shooting time and location, needs to be checked to ensure the authenticity and timeliness of the data.
[0059] S103, exception handling: During the data verification process, if the positioning, photo or video data are found to be erroneous or incomplete, the data receiving submodule will process these abnormal data and issue an error prompt to the operator, informing him to re-upload the data so that the data meets the system requirements.
[0060] The working diagram of the environment detection submodule is as follows: Figure 7 As shown, its functions mainly include detecting and analyzing whether there are combustibles, flammable and explosive hazards, etc. in the pictures and videos uploaded by hot work personnel. Specifically, the workflow of the environmental detection submodule is as follows:
[0061] S201, Positioning detection: According to the positioning information received by the data receiving submodule, by analyzing the location information such as longitude and latitude in the positioning data, determine whether it meets the preset construction area range. If the positioning is within the preset construction area, further environmental detection can be carried out; if it exceeds the prescribed range, it is necessary to prompt the hot work personnel to reposition.
[0062] S202, image detection: After the construction area is confirmed by positioning detection, the image uploaded by the hot work personnel is detected, and the meta-information in the image is extracted to assist positioning detection, such as shooting time, location information, etc., to ensure that the location of the image is within the specified construction area. At the same time, the target detection algorithm is used to detect combustibles, identify and locate combustible targets that may exist in the image, determine the location and category of the combustible targets, and generate image detection results.
[0063] S203, video detection: Similar to image detection, video detection also uses target detection algorithms to process the video frame by frame. The difference is that the detection of the construction environment through video is more comprehensive, with a clearer perspective and richer information. After the detection is completed, the combustible targets and locations detected in the video are summarized to generate video detection results.
[0064] S204, Hazard Assessment: Comprehensively analyze the detection results from image detection and video detection, conduct a comprehensive assessment of the detected combustible targets and possible fire and explosion risks, and generate an assessment report. The assessment report lists the information: combustibles detected by images and videos, and flammable and explosive risk levels.
[0065] In order to accurately identify combustibles in the construction environment, the CBAM (Convolutional Block Attention Module) attention mechanism combining channel attention and spatial attention is added to the target detection method. The specific steps for detecting combustibles are as follows:
[0066] S301, standardize the images collected from the working environment, including scaling to the size required for model input, normalizing brightness and contrast, etc., to ensure the consistency of the model input.
[0067] S302: Input the preprocessed image into the YOLOv8 model, extract features through the backbone network of the model, and then use the feature maps of three different scales, P3, P4, and P5, to perform preliminary target detection to obtain candidate combustible target areas.
[0068] S303, using the channel attention module in CBAM, through the two branches of global average pooling and global maximum pooling, to capture the statistical information between channels. The captured statistical information is input into two MLP (multi-layer perceptron) networks to obtain the importance weight of each channel. The channel weight is multiplied by the original feature channel to strengthen the feature representation of important channels and suppress unimportant channels.
[0069] S304 uses global average pooling and global maximum pooling in the spatial dimension to capture spatial position information, and converts the spatial information into a spatial attention map through 1x1 convolution and ReLU activation function. Finally, the spatial attention map is applied to each channel of the feature map to enhance the feature response of the key area.
[0070] S305, the feature maps processed by channel attention and spatial attention are fused to obtain optimized feature representation, and the fused feature maps are input into the detection head of the YOLOv8 model for more refined object detection and bounding box regression.
[0071] S306, encapsulating the final detection result, including information such as bounding box coordinates, predicted category and confidence.
[0072] S307: According to the feedback results, the model structure, attention mechanism parameters or training strategy are adjusted to further improve the detection performance.
[0073] The working diagram of the audit authorization submodule is as follows Figure 8 As shown in the figure, its main function is for the management personnel to review the applications submitted by the hot work personnel and decide whether to authorize the hot work personnel to use the electric welding equipment according to the actual situation of the construction environment. The review factors include the identity information of the hot work personnel, the equipment required, the safety of the construction environment, the risk assessment level, etc. If the application meets the requirements, the management personnel will apply and authorize the hot work personnel to use the electric welding equipment. Finally, the module will feedback the review results to the hot work personnel. If the application is approved, the hot work personnel will receive the corresponding authorization notice and can start the electric welding operation; if the application is not approved, the evaluation report will be sent to the hot work personnel to facilitate the hot work personnel to make rectifications. At the same time, the evaluation report and review results will be recorded in the system database for subsequent review and tracing.
[0074] In order to help the approval personnel make approval decisions more quickly and accurately and improve the approval efficiency and quality, the approval data of electric welding operations will be stored in the database as basic data to train the intelligent approval model. Based on the data accumulated from a large number of manually approved data tables, the data and approval results in the environmental hazard assessment report after combustible target detection are digitized and standardized, and trained through neural networks to obtain an intelligent approval model to achieve intelligent approval of job applications. At the same time, these data are used as a corpus to train a large assessment model with auxiliary approval functions, analyze and understand the text information in the environmental hazard assessment report, and extract key information and features.
[0075] like Fig. 9 As shown in the figure, an equipment management module of a digital intelligent supervision platform for electric welding operations is used to manage and monitor electric welding equipment. Positioning chips, charging chips and light-sensing chips are implanted in the electric welding equipment. The location-related information of the electric welding equipment is obtained in real time through the positioning chip and Beidou GNSS satellite technology. The positioning information is calibrated and updated through neural network, differential positioning and other technologies to timely detect whether the electric welding equipment deviates from the predetermined area; the battery charging status and working time of the equipment are understood in real time through the charging chip to facilitate the detection of illegal operations; the light perception of the equipment is sensed through the light-sensing chip to verify whether the equipment is disassembled. Combining the three chips, all-round management of the location tracking, power status monitoring and equipment integrity verification of the electric welding equipment is realized, abnormal conditions of the equipment are detected in time, and the safety and quality of the electric welding operation are guaranteed.
[0076] like Fig.10 As shown in the figure, in the equipment management module, the positioning chip mainly cooperates with Beidou GNSS satellite technology to realize the positioning of electric welding equipment. The working location of electric welding is complex and changeable, and the positioning of equipment is easily affected by factors such as magnetic field and terrain. The positioning data can be corrected by convolutional neural network and Kalman filter to reduce noise interference. The structure of convolutional neural network is 1 input layer, 5 convolutional layers, 1 loss layer, 2 fully connected layers and 1 output layer. The specific steps are:
[0077] S401, through the Beidou positioning system, obtain the specific positioning of the electric welding construction area and multiple measurements of each electric welding equipment, and obtain the initial GNSS data set required by the network model. The data includes equipment longitude, latitude, altitude, pseudorange, deviation and other information. The data is least squares processed and loaded into the input layer of the model in sequence.
[0078] S402, setting a convolution kernel of appropriate size and using a convolutional neural network to extract features from the GNSS data.
[0079] S403, through the loss layer and the fully connected layer, helps the model converge as quickly as possible, obtains the output result, completes the model training, and obtains the appropriate noise parameters.
[0080] S404, using Kalman filtering combined with the noise parameters output by the network to correct the predicted value of the electric welding equipment position.
[0081] S405, calculate the error between the predicted value and the true value of the filter output, use the error evaluation score to back-propagate to the network, adjust the network weights, and optimize the network model. After repeated adjustments and updates, the optimal solution is obtained.
[0082] S406, obtaining appropriate noise parameters, and inferring the position information according to the noise parameters to obtain corrected position information.
[0083] like Fig.11 As shown, an early warning management module of a digital intelligent supervision platform for electric welding operations is used to monitor abnormal situations that may occur during or during the application of electric welding operations, and issue early warnings in a timely manner to remind relevant personnel to pay attention and take necessary measures. According to the location and working status of the equipment, illegal work that does not comply with the application process is detected. When the operator does not submit an application for the operation or the application is rejected, according to the positioning chip information, if the position of the electric welding equipment changes, a position offset alarm is issued; according to the charging chip information, if the electric welding is started without submitting the operation application or without passing the application, a red code is assigned, and an alarm is issued to the supervisor; if the operation application is submitted but has not been approved and the electric welding is not started, a yellow code is assigned. According to the light sensing chip information, if the light sensing state of the equipment is abnormal and has not been approved or has not passed the approval, a red code is assigned and an alarm is issued.
[0084] The early warning method is pushed through a variety of methods, including mobile phone text messages, emails, App push, AI intelligent agents, etc.
[0085] To ensure that relevant personnel can receive and respond in a timely manner, it is preferred to use the AI agent push method, the main steps of which are:
[0086] S501, select the relevant platform and create an early warning AI agent.
[0087] S502, upload the platform warning information and the relevant personnel information knowledge base, and generate a personalized intelligent agent tool based on the basic model and the provided knowledge base. According to the corpus, generate a standardized warning dialogue mode and content including the cause of the warning, the warning level, the relevant personnel, etc.
[0088] S503, develop an outbound calling system combined with a large model. When an early warning is generated, the early warning AI agent retrieves the mobile phone number of the relevant person, calls the outbound calling system, calls the mobile phone number, and informs the relevant person of the standardized early warning content.
[0089] The above describes the main technical features and basic principles of the present invention and the related advantages. It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the concept or basic features of the present invention. Therefore, no matter from which point of view, the above specific embodiments should be regarded as exemplary and non-restrictive. The scope of the present invention is defined by the attached claims rather than the above description, and it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present invention.
[0090] In addition, it should be understood that although the present specification is described according to various implementation methods, not every implementation method contains only one independent technical solution. This narrative method of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each implementation method may also be appropriately combined to form other implementation methods that can be understood by those skilled in the art.
Claims
1. A digital intelligent supervision platform for electric welding operations, characterized in that: include: User management module, which has four system identities: hot work personnel, enterprise managers, fire emergency management personnel and fire law enforcement personnel. Users use it according to their own identities. Hot work personnel are responsible for the actual operation of electric welding operations. Enterprises are registered units of the platform and have management rights of the platform. Enterprise managers are responsible for managing users and related operations within the enterprise. Fire emergency management personnel are professionals responsible for supervising and managing the safety of electric welding operations. Fire law enforcement personnel are personnel from relevant departments responsible for law enforcement supervision and inspection of electric welding operations; Filing management module, which is responsible for receiving qualification documents of hot work personnel for filing and providing approval process for filing applications; Approval management module, which includes four submodules: face verification submodule, data receiving submodule, environment detection submodule and review authorization submodule. The face verification submodule is responsible for verifying whether the information of the personnel applying for hot work is consistent with the information filed in the system. The data receiving submodule is responsible for receiving and verifying the location, photos and video data of the construction environment submitted by the personnel doing hot work. The environment detection submodule is responsible for verifying whether the location submitted by the personnel doing hot work is in the specified construction area, and whether there are combustibles and flammable and explosive risks in the photos and videos. The review authorization submodule is responsible for reviewing the application submitted by the personnel doing hot work; The equipment management module includes a positioning chip, a charging chip and a light sensing chip installed in the electric welding equipment. The module obtains the location-related information of the electric welding equipment in real time through the positioning chip and GNSS satellite technology, monitors in real time whether the electric welding equipment deviates from the predetermined area, monitors in real time the battery charging status and working time of the equipment through the charging chip, detects whether there is any illegal operation, and monitors the light perception of the equipment through the light sensing chip to detect whether the equipment is disassembled; The early warning management module includes working environment early warning, equipment status early warning and intelligent early warning. It is used to monitor abnormal situations that occur during the application and process of electric welding operations, and issue early warnings in time to remind relevant personnel to pay attention and take measures.
2. According to claim 1, a digital intelligent monitoring platform for electric welding operations is characterized in that: The environmental detection submodule in the approval management module adopts a target detection algorithm to detect combustibles. By extracting the features of various objects in the construction environment, it identifies and marks potential combustible targets in the construction environment, evaluates the potential fire and explosion hazards, and generates a hazard assessment report.
3. According to claim 2, a digital intelligent monitoring platform for electric welding operations is characterized in that: The target detection algorithm adds a CBAM (Convolutional Block Attention Module) attention mechanism that combines channel attention and spatial attention. The specific steps for detecting combustibles are as follows: S301, standardizing the images collected from the working environment, including scaling to a size required for model input, and normalizing brightness and contrast to ensure consistency of the model input; S302, input the preprocessed image into the YOLOv8 model, extract features through the backbone network of the model, and then use the feature maps of three different scales, P3, P4, and P5, to perform preliminary target detection to obtain candidate combustible target areas; S303, using the channel attention module in CBAM, through the two branches of global average pooling and global maximum pooling, capturing the statistical information between channels, inputting the captured statistical information into two MLP (multi-layer perceptron) networks, obtaining the importance weight of each channel, multiplying the channel weight by the original feature channel, strengthening the feature representation of important channels, and suppressing unimportant channels; S304, using global average pooling and global maximum pooling in the spatial dimension to capture spatial position information, and converting the spatial information into a spatial attention map through 1x1 convolution and ReLU activation function, and finally applying the spatial attention map to each channel of the feature map to enhance the feature response of the key area; S305, fusing the feature maps processed by channel attention and spatial attention to obtain optimized feature representation, and inputting the fused feature maps into the detection head of the YOLOv8 model to perform more refined object detection and bounding box regression; S306, encapsulating the final detection result, including the bounding box coordinates, the predicted category and the confidence information; S307: According to the feedback results, the model structure, attention mechanism parameters and training strategy are adjusted to further improve the detection performance.
4. According to claim 1, a digital intelligent monitoring platform for electric welding operations is characterized in that: In the user management module, enterprise managers are responsible for managing users within the enterprise and related operations including adding, deleting, and modifying user information, viewing the electric welding operations within the enterprise through the platform, and performing data analysis and report generation operations; fire emergency management personnel view the details of electric welding operations within their jurisdiction, regional data overviews, review of declaration-related information, and alarm records of illegal operations through the platform; fire law enforcement personnel view the filing status and approval records of electric welding operations through the platform, conduct on-site inspections and assessments, and handle and punish illegal behaviors.
5. The digital intelligent monitoring platform for electric welding operations according to claim 1 is characterized in that: The filing management module also provides real-time tracking and management functions for the filing application status, so that hot work personnel and management personnel can understand the processing progress and results of the filing application at any time.
6. The digital intelligent monitoring platform for electric welding operations according to claim 1 is characterized in that: The face verification submodule in the approval management module adopts face recognition technology and liveness detection technology based on convolutional neural network (CNN) to analyze the feature points and facial contour information of the face, and compare and verify the information with the pre-established hot work personnel information database. If the verification is successful, the system will immediately feedback to the person applying for the hot work and allow the person applying for the hot work to further operate. If the verification fails, the person applying for the hot work will be prompted to re-upload the identity information or contact the relevant management personnel for processing.
7. The digital intelligent monitoring platform for electric welding operations according to claim 1 is characterized in that: The data receiving submodule in the approval management module is connected to the communication interface of the platform. Once the hot work personnel upload data through the platform, the data receiving submodule responds and receives the data immediately. The data receiving submodule is compatible with different image and video formats, and is equipped with an exception handling mechanism, which can detect uploaded data anomalies in time and notify the personnel applying for the hot work to handle it.
8. The digital intelligent monitoring platform for electric welding operations according to claim 1 is characterized in that: In the device management module, the acquired location-related information is calibrated and updated through convolutional neural network and differential positioning technology.
9. A digital intelligent monitoring platform for electric welding operations according to claim 8, characterized in that: The calibrating of the acquired location related information specifically comprises the following steps: S401, using the Beidou positioning system, obtain the specific positioning of the electric welding construction area and multiple measurements of each electric welding equipment, and obtain the initial GNSS data set required by the network model. The data includes the equipment longitude, latitude, altitude, pseudorange, and deviation information. The data is processed by least squares and loaded into the input layer of the model in sequence; S402, setting a convolution kernel of appropriate size and using a convolutional neural network to extract features from the GNSS data; S403, through the loss layer and the fully connected layer, the model is helped to converge, the output result is obtained, the model training is completed, and the appropriate noise parameters are obtained; S404, using Kalman filtering combined with noise parameters output by the network to correct the predicted value of the electric welding equipment position; S405, calculating the error between the predicted value and the true value of the filter output, using the error evaluation score to back-propagate to the network, adjusting the network weights, optimizing the network model, and obtaining the optimal solution after repeated adjustments and updates; S406, obtaining appropriate noise parameters, and inferring the position information according to the noise parameters to obtain corrected position information.
10. The digital intelligent monitoring platform for electric welding operations according to claim 1 is characterized in that: In the working environment warning, when combustible materials are detected in the working environment, a warning is issued to the approval personnel to remind them to reject the application for hot work, and the information is fed back to the person applying for the hot work; in the equipment status warning, based on the monitoring of the equipment by the equipment management module, a warning is issued for not moving the electric welding equipment according to regulations, turning on the electric welding equipment, and disassembling the electric welding equipment; in the intelligent warning, big data and artificial intelligence technology are combined to perform intelligent approval and intelligent warning.
Citation Information
Cited By
Digital intelligent supervision platform for electro-gas welding operation
CN120551649A
Special operation safety early warning method and system based on multi-source data fusion and edge visual analysis
CN122220963A
