Remote welding machine control system and method based on Internet of Things

Abstract

The invention discloses a remote welding machine control system and method based on the Internet of Things, and the method comprises the following steps: S1, collecting and preprocessing multi-source time sequence data, and generating a time sequence data set; s2, constructing a variational auto-encoder model, extracting potential features and reconstructing original data; s3, using a bird flock optimization algorithm to optimize hyper-parameters of the VAE model; s4, performing feature extraction and data reconstruction through the optimized VAE model; s5, model output and actual data are compared, the welding quality is evaluated, and control parameters are adjusted; s6, real-time welding operation is carried out according to the adjusted control parameters, and equipment identity binding and permission verification are carried out; s7, performing real-name fire approval through the platform, and remotely starting the equipment after the approval is passed; s8, the welding state is monitored, early warning is conducted on abnormity, and a maintenance plan and reminding are pushed; and S9, uniformly managing equipment states, task approval, early warning and operation and maintenance, and generating an analysis report. Intelligent control and safety management of the remote welding machine are achieved, and automation and efficiency of welding operation are improved.

Description

Technical Field

[0001] This invention relates to the field of Internet of Things (IoT) technology, and in particular to a remote welding machine control system and method based on IoT. Background Technology

[0002] With the deepening development of intelligent and digital manufacturing, welding, as one of the most critical connection processes in industrial production, is constantly evolving from traditional manual operations towards automation, remote operation, and intelligence. Driven by intelligent manufacturing, the Industrial Internet of Things (IIoT), and big data analytics, welding equipment is gradually realizing core capabilities such as status perception, remote control, predictive maintenance, and closed-loop feedback. However, most current welding management systems still have significant limitations in terms of overall architecture, functional coordination, intelligent perception, and security, making it difficult to meet the demands for high reliability, high security, and high efficiency in welding operation and maintenance management in complex industrial scenarios.

[0003] In traditional welding operations, the acquisition and control of equipment operating status rely on local manual inspections and operations, lacking real-time sensing and remote control capabilities. This easily leads to information lag and slow response, affecting welding quality and production efficiency. Especially when welding operations are carried out in high-temperature, high-risk, or hot-fire environments, the lack of effective approval and monitoring mechanisms will seriously threaten the safety of personnel and equipment. While some deployed remote monitoring systems have basic status display functions, they are mostly partially integrated and lack the data governance and control scheduling capabilities of a unified platform, failing to achieve intelligent collaboration among multiple devices, tasks, and users. Furthermore, current welding systems mostly utilize data only for "post-event review" rather than "real-time diagnosis" and "proactive optimization," failing to promptly identify potential anomalies and intervene in adjustments.

[0004] In welding safety management, traditional hot work permit processes rely heavily on paper forms or manual review, resulting in delays, opaque approval processes, and the accidental activation of equipment without being linked to an approval task. Furthermore, welding risks are often sudden and uncontrollable, such as abnormal current, excessively high temperature, and exceeding welding time limits. These risks rely on manual judgment and experience-based handling, lacking efficient and automated risk identification mechanisms. At the equipment level, the lack of unified identification and access control means frequently leads to untraceable work processes and uncontrollable equipment operation, creating a blind spot in safety management.

[0005] Furthermore, the maintenance and management of welding equipment generally relies on periodic inspections, lacking predictive maintenance strategies based on operational behavior and health status. This can easily lead to equipment operating with defects or undergoing excessive maintenance, reducing equipment utilization efficiency and operational resource utilization. At the platform level, many current welding management systems have limited functionality, fragmented interfaces, and simple permission structures. They cannot achieve centralized scheduling and unified data analysis across regions and equipment, nor do they support multi-role collaboration and hierarchical management, which is detrimental to the digital supervision and resource optimization of large-scale welding projects.

[0006] Regarding data application, while existing systems possess some parameter acquisition capabilities, they lack effective deep learning models for intelligent interpretation and dynamic prediction of welding data. For instance, time-series data such as current, temperature, and pressure during the welding process are often simply compared using threshold settings, ignoring the correlations and time-dependent characteristics between multiple factors. This lack of data modeling capability for the complex behaviors of the welding process makes it difficult to accurately assess welding quality and optimize parameters. Furthermore, most current systems do not incorporate advanced intelligent optimization algorithms, such as swarm intelligence algorithms and neural network evolution techniques, hindering efficient adaptive adjustment of model structure or hyperparameters, thus limiting the model's accuracy and generalization ability.

[0007] Therefore, how to provide a remote welding machine control system and method based on the Internet of Things is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0008] One objective of this invention is to propose a remote welding machine control system and method based on the Internet of Things (IoT). This invention fully utilizes IoT communication technology, variational autoencoder models, and bird flocking optimization algorithms to construct an intelligent sensing, remote control, safety approval, risk warning, and centralized management system for the entire welding operation process. It describes in detail how to model, predict, and adjust the welding state through an optimized deep learning model. It has the advantages of strong security, real-time response, precise control, and high platform integration, and can effectively improve the intelligence level and full-process supervision capability of welding operations.

[0009] A remote welding machine control method based on the Internet of Things according to an embodiment of the present invention includes the following steps: S1. Collect multi-source time-series data of remote welding machines through IoT devices, preprocess the multi-source time-series data, and generate a time-series dataset; S2. Based on the time series dataset, construct a variational autoencoder model. The encoder maps the time series dataset to the latent space, and the decoder reconstructs the latent variables into the original data. S3. The hyperparameters of the variational autoencoder model are optimized by applying the bird flock optimization algorithm. The variational autoencoder is then trained using the hyperparameters optimized by the bird flock optimization algorithm to obtain the optimized variational autoencoder model. S4. Based on the real-time sensor data of the welding machine, input the optimized variational autoencoder model, use the encoder to extract latent features, the decoder to reconstruct the data, and transmit the results in real time through the Internet of Things platform. S5. Based on the comparative analysis of the output of the optimized variational autoencoder model and the actual welding data, the welding quality is evaluated, and the evaluation results are fed back to the welding machine control system through the Internet of Things to adjust the welding control parameters. S6. Based on the adjusted welding control parameters, real-time welding operations are performed under the command drive of the IoT platform. Real-time data is continuously collected and status analysis is performed. At the same time, a unique identity binding and permission verification mechanism is combined with the one machine, one code identification mechanism to bind the welding machine to a unique identity and verify permissions. Only operation tasks that have been approved and authorized are allowed to issue control commands. S7. Establish an intelligent hot work approval mechanism. Before welding operations, real-name registration and approval must be carried out through the platform. Only after approval can the welding machine be remotely started. Unauthorized tasks must not activate the equipment. S8. Construct a welding condition monitoring and operation and maintenance management module to automatically push early warning information, generate maintenance plans and push reminders; S9. Construct a platform-based data and scheduling module to uniformly manage welding equipment status, task approval, early warning and maintenance information through the Internet of Things, generate analysis reports regularly, and support intelligent decision-making and operation and maintenance under multi-terminal hierarchical permissions.

[0010] Optionally, the multi-source time-series data specifically includes welding current, temperature, pressure, vibration, and welding quality data, which are used to analyze the welding process in real time and optimize welding control parameters.

[0011] Optionally, the preprocessing of the multi-source time series data specifically includes denoising, standardization, and data fusion, which are used to improve the quality of the multi-source time series data and provide accurate input for the variational autoencoder model.

[0012] Optionally, S2 specifically includes: S21. Receive the preprocessed time series dataset and pass it as input data to the variational autoencoder model. S22. Design and construct the structure of a variational autoencoder model, including an encoder and a decoder, where the encoder's role is to extract features from a time-series dataset and map the features to a latent space; S23. In the encoder part, a multi-layer fully connected neural network is used to process the time series dataset. The encoder maps the input time series dataset to the latent space through the multi-layer network to generate the mean and variance of the latent variables, which are used to describe the latent distribution of the time series dataset. S24. Apply the reparameterization technique to transform the mean and variance parameters of the latent variables output by the encoder into latent variables. S25. In the decoder section, a multi-layer fully connected neural network is used to map the latent variables obtained from the latent space back to the original data space. The output generated by the decoder is the reconstructed time series data. S26. By minimizing the reconstruction error and KL divergence, a variational autoencoder model is trained, and the parameters of the encoder and decoder are optimized. The variational autoencoder model can effectively reconstruct time series data and ensure that the data distribution in the latent space is consistent with the standard normal distribution.

[0013] Optionally, S3 specifically includes: S31. Initialize the bird flock. During initialization, each bird represents a combination of hyperparameters of the variational autoencoder model. Hyperparameters include the number of hidden layers, the number of neurons per layer, the dimension of the latent space, and the learning rate. The initial position of each bird represents a set of random hyperparameters. The initial velocity of each bird is determined according to the set initial search space range. The bird's initial position is ,in, The number of hidden layers in the encoder. The number of neurons per layer, For the dimension of the potential space, Let be the learning rate, and be the initial velocity. ,in, , , , This represents the components of the initial velocity in each hyperparameter dimension; S32. Real-time welding data during the welding process is collected via IoT devices. ,in, For welding current, For welding temperature, The welding pressure is used as the real-time welding data as feedback input to the bird flock optimization algorithm, which is used to dynamically adjust the search strategy and affect the flight mode and hyperparameter adjustment in real time. S33. Each bird coordinates its flight based on its historical best position, global best position, and the flight status of other birds in its neighborhood. The flight status includes speed, direction, and position. The influence of neighborhood information is dynamically adjusted based on feedback from real-time welding data, updating the speed of each bird. ; in, For the first Only one bird in the first The speed of generation For the first Only one bird in the first The position of the generation, For the first Only one bird in the first The position of the generation, For the first The best historical position for a bird This is the optimal position for the entire flock of birds. For the first The neighboring area of ​​a single bird For the first Only bird and the first The distance between the birds , , , For constant parameters, , , It is a random number; S34. Based on real-time welding data, optimize and adjust the hyperparameter combination for each bird in real time. Through real-time welding quality prediction and dynamic parameter adjustment, evaluate the hyperparameter combination corresponding to each bird and calculate the welding quality index. : ; in, , and They are the first The current, temperature, and pressure values ​​during the welding process under the corresponding hyperparameters for a single bird. , and The target welding current, temperature, and pressure are the ideal values. , , This is a weighting adjustment factor used to adjust the degree of influence of welding current, welding temperature, and welding pressure on welding quality assessment. S35. Based on feedback from real-time welding data, dynamically adjust the weights of the global and local searches for bird flocks. If temperature fluctuations are significant during the welding process, enhance the global search capability; conversely, enhance the local fine search when the welding process is stable. Introduce a welding process error feedback mechanism, adjusting the weights of the global and local searches based on the welding quality prediction error. When the welding quality deviates from expectations, strengthen the global search; conversely, strengthen the local search. The adjustment of the global and local search weights is as follows: ; ; in, This serves as the global search weight, used to dynamically adjust the breadth of the search strategy. These are local search weights used to dynamically adjust the precision of the search strategy. Welding current and reference current The difference is used to reflect changes in welding current. For welding temperature and reference temperature The difference is used to reflect fluctuations in welding temperature. Predicted error and target error for welding quality The difference is used to reflect the deviation in welding quality. and These represent the reference current and reference temperature during the welding process, respectively. The target welding quality error is calculated by real-time monitoring of the deviation between the welding quality and the set target. , , This is a weighting adjustment factor; S36. A dynamic flight mode adjustment mechanism is introduced. Based on real-time welding data, the flight mode of the bird flock can switch between broad search and fine search. The broad search phase uses global search weights, while the fine search phase uses local search weights. When the welding current fluctuates significantly, the global search intensity is increased. ; When welding conditions are stable, local search weights are used: ; in, For global search, the first Only one bird in the first The speed of generation For local search, the first Only one bird in the first The speed of generation and The acceleration constant, and It is a random number; S37. Update the positions and speeds of the flock members, calculate the new fitness values, and evaluate the effect of the new position and hyperparameter combination for each bird: ; in, For the first The fitness value of a single bird. For the first The welding quality assessment value of a single bird. For the first The welding temperature of a single bird under the current hyperparameter combination. For the first The welding material consumption for each bird This represents the total welding time. For the target welding time, To optimize the target temperature, For the target welding material consumption, For the target welding time, , and As a regulating factor; S38. Based on the updated fitness values, when the maximum number of iterations is reached or the fitness threshold is exceeded, the optimal hyperparameter configuration is output. The optimal hyperparameter configuration is then used to train the final variational autoencoder model. for: ; in, The optimal number of hidden layers for the encoder. To determine the optimal number of neurons per layer, For the optimal potential spatial dimension, The optimal learning rate; S39. Finally, the optimized variational autoencoder model is obtained. The optimized variational autoencoder model can perform data reconstruction and welding quality prediction more accurately.

[0014] Optionally, S4 specifically includes: S41. Real-time welding data during the welding process is collected through IoT sensors, including welding current, welding temperature, and welding pressure. S42. The collected real-time welding data is transmitted to the optimized variational autoencoder model. The encoder part maps the real-time welding data to the latent space and extracts the latent features related to welding quality and production efficiency during the welding process. The optimized variational autoencoder model can automatically adjust the feature extraction strategy based on historical data and real-time feedback. S43. Input the latent space representation extracted by the encoder into the decoder. The decoder reconstructs the welding parameters based on the latent variables and generates optimized welding data output. S44. By comparing the error between the reconstructed data generated by the decoder and the actual real-time acquired welding data, an adaptive error correction mechanism is introduced. The error feedback will correct the deviation in the welding process by dynamically adjusting the weights. ; in, For the first Reconstruction error of each welding parameter and The first The values ​​of each welding parameter in the actual data and the reconstructed data. It is a dynamic adjustment factor. It is an error feedback item used to adjust the error correction strength for each welding parameter; S45. Based on the calculated reconstruction error and error feedback, dynamically optimize the learning rate and hyperparameters of the variational autoencoder model. By introducing a multi-stage adaptive learning rate adjustment mechanism, the learning rate is adjusted according to real-time errors and environmental fluctuations at different stages of the welding process to optimize welding quality. ; in, For the updated learning rate, To optimize the learning rate, let represent the learning rate under ideal conditions. and This is an adjustment factor used to control the degree to which error affects the learning rate adjustment. As an adaptive factor, Indicates welding fluctuations; S46. The optimized welding data is transmitted to the remote control system in real time through the Internet of Things (IoT) platform, so that the operator can monitor and adjust it in real time. The IoT platform automatically adjusts the control parameters in the welding process based on real-time feedback and the prediction of the optimized variational autoencoder model.

[0015] Optionally, S5 specifically includes: S51. Obtain the output results of the optimized variational autoencoder model. The output results include the prediction data and reconstruction data of the welding process by the optimized variational autoencoder model. S52. Acquire real-time welding data. The real-time welding data is collected in real time by IoT sensors, including the actual welding current, welding temperature and welding pressure. S53. Compare and analyze the reconstructed data output by the optimized variational autoencoder model with the real-time welding data acquired in real time to identify deviations or abnormalities during the welding process. S54. Based on the results of comparative analysis, evaluate the welding quality. The evaluation criteria include the stability and accuracy of the welding parameters and whether they meet the preset quality standards. S55. Feed the evaluation results back to the Internet of Things (IoT) platform. The IoT platform will transmit the welding quality evaluation results to the welding machine control system through real-time data transmission. S56. Based on the evaluation results, the welding machine control system triggers an early warning mechanism based on the specific evaluation data. If the evaluation results indicate that the welding quality is substandard, the system will take measures, including suspending welding, sending an alarm, or further analyzing the data.

[0016] Optionally, the welding status monitoring and operation and maintenance management module is built using IoT sensing devices, and the platform-based data and scheduling module is built using an IoT communication architecture.

[0017] A remote welding machine control system based on the Internet of Things according to an embodiment of the present invention includes the following modules: The remote control module is used to receive welding machine control commands in real time through the Internet of Things platform, and control the start, stop and change the operating status of the welding machine according to the commands; The real-time monitoring module is used to collect the welding machine's operating status data in real time and monitor key parameters during the welding process; The fire monitoring and approval module is used to detect fire risks in real time during the welding process and to approve them. The device identification and binding module is used to assign a unique identifier to each welding machine and bind it. The data analysis and risk warning module is used to analyze various data in the welding process and conduct risk assessments, issuing warnings when potential problems are discovered. The report generation and optimization module is used to generate a report on the welding process based on real-time collected data, and to optimize and adjust the welding parameters based on the evaluation results. The welding status monitoring and operation and maintenance management module is used to identify welding anomalies in real time and generate maintenance plans, and to complete welding safety early warning and equipment health management. The platform-based data and scheduling module is used to uniformly schedule welding data and control tasks, supporting intelligent decision-making and collaborative operation and maintenance under multi-terminal hierarchical permissions.

[0018] The beneficial effects of this invention are: This invention proposes an IoT-based remote welding machine control system and method, which effectively solves key problems in existing welding operations, such as insufficient safety supervision, uncontrollable equipment operation, lagging parameter adjustment, severe data silos, and lack of intelligent optimization models. By constructing a welding behavior modeling mechanism based on multi-source time-series data, and combining variational autoencoders and bird flocking optimization algorithms, deep learning, predictive reconstruction, and dynamic parameter adjustment of the welding state are achieved. This improves the data perception capability and modeling accuracy of the welding process, ensuring that the system still has good generalization and adaptability when facing complex welding conditions.

[0019] This invention enables remote command control and status acquisition of welding equipment through an IoT platform. It not only supports remote start / stop and permission binding of equipment but also introduces a "one machine, one code" mechanism to achieve precise correspondence between equipment and approved tasks and traceability of the operation process, effectively preventing unauthorized operations. The system's built-in intelligent hot work approval module ensures that welding tasks undergo full-process real-name verification before execution, significantly improving the standardization of operations and risk control capabilities. Simultaneously, this invention constructs a real-time monitoring and early warning mechanism for the welding process, capable of identifying abnormal behaviors and triggering alarms based on real-time collected multi-dimensional parameters such as current, temperature, and pressure, enabling early detection and response to welding safety issues.

[0020] In terms of equipment maintenance and management, this invention establishes an equipment health status assessment model by collecting indicators such as runtime, fault records, and start-up / shutdown frequency. This model generates precise maintenance plans and repair reminders, breaking away from the traditional passive, fixed-cycle maintenance mode and significantly improving equipment utilization efficiency and system reliability. Furthermore, this invention integrates various data resources from welding operations through a platform architecture, enabling centralized scheduling and management of welding equipment status, approval processes, early warning events, and maintenance plans. Combined with data statistics and report generation functions, it provides intelligent support for management decisions across multiple roles and terminals.

[0021] In summary, the remote welding machine control system proposed in this invention not only achieves breakthroughs in welding data perception and intelligent modeling, but also forms effective synergy in multiple key aspects such as remote equipment control, safety approval, risk warning, maintenance management, and platform scheduling. It has significant advantages such as reasonable structure, complete functions, flexible deployment, and strong scalability, and can significantly improve the safety, intelligence level, and operation and maintenance efficiency of welding operations, meeting the actual needs of industrial sites for high-safety, high-efficiency, and visualized welding management. Attached Figure Description

[0022] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart of a remote welding machine control method based on the Internet of Things proposed in this invention; Figure 2 This is a schematic diagram of the structure of a remote welding machine control system based on the Internet of Things proposed in this invention. Detailed Implementation

[0023] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0024] refer to Figure 1 A remote welding machine control method based on the Internet of Things includes the following steps: S1. Collect multi-source time-series data of remote welding machines through IoT devices, preprocess the multi-source time-series data, and generate a time-series dataset; S2. Based on the time series dataset, construct a variational autoencoder model. The encoder maps the time series dataset to the latent space, and the decoder reconstructs the latent variables into the original data. S3. The hyperparameters of the variational autoencoder model are optimized by applying the bird flock optimization algorithm. The variational autoencoder is then trained using the hyperparameters optimized by the bird flock optimization algorithm to obtain the optimized variational autoencoder model. S4. Based on the real-time sensor data of the welding machine, input the optimized variational autoencoder model, use the encoder to extract latent features, the decoder to reconstruct the data, and transmit the results in real time through the Internet of Things platform. S5. Based on the comparative analysis of the output of the optimized variational autoencoder model and the actual welding data, the welding quality is evaluated, and the evaluation results are fed back to the welding machine control system through the Internet of Things to adjust the welding control parameters. S6. Based on the adjusted welding control parameters, real-time welding operations are performed under the command drive of the IoT platform. Real-time data is continuously collected and status analysis is performed. At the same time, a unique identity binding and permission verification mechanism is combined with the one machine, one code identification mechanism to bind the welding machine to a unique identity and verify permissions. Only operation tasks that have been approved and authorized are allowed to issue control commands. S7. Establish an intelligent hot work approval mechanism. Before welding operations, real-name registration and approval must be carried out through the platform. Only after approval can the welding machine be remotely started. Unauthorized tasks must not activate the equipment. S8. Construct a welding condition monitoring and operation and maintenance management module to automatically push early warning information, generate maintenance plans and push reminders; S9. Construct a platform-based data and scheduling module to uniformly manage welding equipment status, task approval, early warning and maintenance information through the Internet of Things, generate analysis reports regularly, and support intelligent decision-making and operation and maintenance under multi-terminal hierarchical permissions.

[0025] In this embodiment, the multi-source time-series data specifically includes welding current, temperature, pressure, vibration, and welding quality data, which are used to analyze the welding process in real time and optimize welding control parameters.

[0026] In this embodiment, the preprocessing of the multi-source time series data specifically includes denoising, standardization, and data fusion, which are used to improve the quality of the multi-source time series data and provide accurate input for the variational autoencoder model.

[0027] In this embodiment, S2 specifically includes: S21. Receive the preprocessed time series dataset and pass it as input data to the variational autoencoder model. S22. Design and construct the structure of a variational autoencoder model, including an encoder and a decoder, where the encoder's role is to extract features from a time-series dataset and map the features to a latent space; S23. In the encoder part, a multi-layer fully connected neural network is used to process the time series dataset. The encoder maps the input time series dataset to the latent space through the multi-layer network to generate the mean and variance of the latent variables, which are used to describe the latent distribution of the time series dataset. S24. Apply the reparameterization technique to transform the mean and variance parameters of the latent variables output by the encoder into latent variables. S25. In the decoder section, a multi-layer fully connected neural network is used to map the latent variables obtained from the latent space back to the original data space. The output generated by the decoder is the reconstructed time series data. S26. By minimizing the reconstruction error and KL divergence, a variational autoencoder model is trained, and the parameters of the encoder and decoder are optimized. The variational autoencoder model can effectively reconstruct time series data and ensure that the data distribution in the latent space is consistent with the standard normal distribution.

[0028] In this embodiment, S3 specifically includes: S31. Initialize the bird flock. During initialization, each bird represents a combination of hyperparameters of the variational autoencoder model. Hyperparameters include the number of hidden layers, the number of neurons per layer, the dimension of the latent space, and the learning rate. The initial position of each bird represents a set of random hyperparameters. The initial velocity of each bird is determined according to the set initial search space range. The bird's initial position is ,in, The number of hidden layers in the encoder. The number of neurons per layer, For the dimension of the potential space, Let be the learning rate, and be the initial velocity. ,in, , , , This represents the components of the initial velocity in each hyperparameter dimension; S32. Real-time welding data during the welding process is collected via IoT devices. ,in, For welding current, For welding temperature, The welding pressure is used as the real-time welding data as feedback input to the bird flock optimization algorithm, which is used to dynamically adjust the search strategy and affect the flight mode and hyperparameter adjustment in real time. S33. Each bird coordinates its flight based on its historical best position, global best position, and the flight status of other birds in its neighborhood. The flight status includes speed, direction, and position. The influence of neighborhood information is dynamically adjusted based on feedback from real-time welding data, updating the speed of each bird. ; in, For the first Only one bird in the first The speed of generation For the first Only one bird in the first The position of the generation, For the first Only one bird in the first The position of the generation, For the first The best historical position for a bird This is the optimal position for the entire flock of birds. For the first The neighboring area of ​​a single bird For the first Only bird and the first The distance between the birds , , , For constant parameters, , , It is a random number; S34. Based on real-time welding data, optimize and adjust the hyperparameter combination for each bird in real time. Through real-time welding quality prediction and dynamic parameter adjustment, evaluate the hyperparameter combination corresponding to each bird and calculate the welding quality index. : ; in, , and They are the first The current, temperature, and pressure values ​​during the welding process under the corresponding hyperparameters for a single bird. , and The target welding current, temperature, and pressure are the ideal values. , , This is a weighting adjustment factor used to adjust the degree of influence of welding current, welding temperature, and welding pressure on welding quality assessment. S35. Based on feedback from real-time welding data, dynamically adjust the weights of the global and local searches for bird flocks. If temperature fluctuations are significant during the welding process, enhance the global search capability; conversely, enhance the local fine search when the welding process is stable. Introduce a welding process error feedback mechanism, adjusting the weights of the global and local searches based on the welding quality prediction error. When the welding quality deviates from expectations, strengthen the global search; conversely, strengthen the local search. The adjustment of the global and local search weights is as follows: ; ; in, This serves as the global search weight, used to dynamically adjust the breadth of the search strategy. These are local search weights used to dynamically adjust the precision of the search strategy. Welding current and reference current The difference is used to reflect changes in welding current. For welding temperature and reference temperature The difference is used to reflect fluctuations in welding temperature. For welding quality prediction error and target error The difference is used to reflect the deviation in welding quality. and These represent the reference current and reference temperature during the welding process, respectively. The target welding quality error is calculated by real-time monitoring of the deviation between the welding quality and the set target. , , This is a weighting adjustment factor; S36. A dynamic flight mode adjustment mechanism is introduced. Based on real-time welding data, the flight mode of the bird flock can switch between broad search and fine search. The broad search phase uses global search weights, while the fine search phase uses local search weights. When the welding current fluctuates significantly, the global search intensity is increased. ; When welding conditions are stable, local search weights are used: ; in, For global search, the first Only one bird in the first The speed of generation For local search, the first Only one bird in the first The speed of generation and The acceleration constant, and It is a random number; S37. Update the positions and speeds of the flock members, calculate the new fitness values, and evaluate the effect of the new position and hyperparameter combination for each bird: ; in, For the first The fitness value of a single bird. For the first The welding quality assessment value of a single bird. For the first The welding temperature of a single bird under the current hyperparameter combination. For the first The welding material consumption for each bird This represents the total welding time. For the target welding time, To optimize the target temperature, For the target welding material consumption, For the target welding time, , and As a regulating factor; S38. Based on the updated fitness values, when the maximum number of iterations is reached or the fitness threshold is exceeded, the optimal hyperparameter configuration is output. The optimal hyperparameter configuration is then used to train the final variational autoencoder model. for: ; in, The optimal number of hidden layers for the encoder. To determine the optimal number of neurons per layer, For the optimal potential spatial dimension, The optimal learning rate; S39. Finally, the optimized variational autoencoder model is obtained. The optimized variational autoencoder model can perform data reconstruction and welding quality prediction more accurately.

[0029] In this embodiment, S4 specifically includes: S41. Real-time welding data during the welding process is collected through IoT sensors, including welding current, welding temperature, and welding pressure. S42. The collected real-time welding data is transmitted to the optimized variational autoencoder model. The encoder part maps the real-time welding data to the latent space and extracts the latent features related to welding quality and production efficiency during the welding process. The optimized variational autoencoder model can automatically adjust the feature extraction strategy based on historical data and real-time feedback. S43. Input the latent space representation extracted by the encoder into the decoder. The decoder reconstructs the welding parameters based on the latent variables and generates optimized welding data output. S44. By comparing the error between the reconstructed data generated by the decoder and the actual real-time acquired welding data, an adaptive error correction mechanism is introduced. The error feedback will correct the deviation in the welding process by dynamically adjusting the weights. ; in, For the first Reconstruction error of each welding parameter and The first The values ​​of each welding parameter in the actual data and the reconstructed data. It is a dynamic adjustment factor. It is an error feedback item used to adjust the error correction strength for each welding parameter; S45. Based on the calculated reconstruction error and error feedback, dynamically optimize the learning rate and hyperparameters of the variational autoencoder model. By introducing a multi-stage adaptive learning rate adjustment mechanism, the learning rate is adjusted according to real-time errors and environmental fluctuations at different stages of the welding process to optimize welding quality. ; in, For the updated learning rate, To optimize the learning rate, let represent the learning rate under ideal conditions. and This is an adjustment factor used to control the degree to which error affects the learning rate adjustment. As an adaptive factor, Indicates welding fluctuations; S46. The optimized welding data is transmitted to the remote control system in real time through the Internet of Things (IoT) platform, so that the operator can monitor and adjust it in real time. The IoT platform automatically adjusts the control parameters in the welding process based on real-time feedback and the prediction of the optimized variational autoencoder model.

[0030] In this embodiment, S5 specifically includes: S51. Obtain the output results of the optimized variational autoencoder model. The output results include the prediction data and reconstruction data of the welding process by the optimized variational autoencoder model. S52. Acquire real-time welding data. The real-time welding data is collected in real time by IoT sensors, including the actual welding current, welding temperature and welding pressure. S53. Compare and analyze the reconstructed data output by the optimized variational autoencoder model with the real-time welding data acquired in real time to identify deviations or abnormalities during the welding process. S54. Based on the results of comparative analysis, evaluate the welding quality. The evaluation criteria include the stability and accuracy of the welding parameters and whether they meet the preset quality standards. S55. Feed the evaluation results back to the Internet of Things (IoT) platform. The IoT platform will transmit the welding quality evaluation results to the welding machine control system through real-time data transmission. S56. Based on the evaluation results, the welding machine control system triggers an early warning mechanism based on the specific evaluation data. If the evaluation results indicate that the welding quality is substandard, the system will take measures, including suspending welding, sending an alarm, or further analyzing the data.

[0031] In this embodiment, the welding status monitoring and operation and maintenance management module is built through IoT sensing devices, and the platform-based data and scheduling module is built through an IoT communication architecture.

[0032] refer to Figure 2 A remote welding machine control system based on the Internet of Things includes the following modules: The remote control module is used to receive welding machine control commands in real time through the Internet of Things platform, and control the start, stop and change the operating status of the welding machine according to the commands; The real-time monitoring module is used to collect the welding machine's operating status data in real time and monitor key parameters during the welding process; The fire monitoring and approval module is used to detect fire risks in real time during the welding process and to approve them. The device identification and binding module is used to assign a unique identifier to each welding machine and bind it. The data analysis and risk warning module is used to analyze various data in the welding process and conduct risk assessments, issuing warnings when potential problems are discovered. The report generation and optimization module is used to generate a report on the welding process based on real-time collected data, and to optimize and adjust the welding parameters based on the evaluation results. The welding status monitoring and operation and maintenance management module is used to identify welding anomalies in real time and generate maintenance plans, and to complete welding safety early warning and equipment health management. The platform-based data and scheduling module is used to uniformly schedule welding data and control tasks, supporting intelligent decision-making and collaborative operation and maintenance under multi-terminal hierarchical permissions.

[0033] Example 1: To verify the feasibility of this invention in practice, it was applied to a smart manufacturing enterprise. Welding, as one of the core processes, is widely used in the manufacturing of metal structural components and automated equipment. Before introducing the system of this invention, the factory had long relied on traditional welding methods, mainly depending on manual equipment startup, manual record-keeping and approval, regular manual inspections, and a lack of systematic monitoring and scheduling. Especially during periods of intensive welding tasks, there were problems such as startup delays, loopholes in safety approvals, and low equipment utilization efficiency, resulting in large fluctuations in work efficiency and welding quality, and making it difficult to predict and control the risk of welding accidents.

[0034] To address the aforementioned issues, in early February 2025, the company deployed an IoT-based remote welding machine control system at five typical welding stations and conducted comparative tests with traditional welding methods in areas where the system was not deployed. The system binds the welding machine to the platform using a unique code mechanism. All operators initiate real-name approval through the platform, and after approval, remote start / stop commands can be issued. The platform runs a variational autoencoder model based on collected real-time welding data, dynamically predicts the welding status using a bird flocking optimization algorithm, automatically identifies and issues warnings for abnormal behavior, and pushes maintenance suggestions based on equipment operating parameters.

[0035] In practical applications, traditional welding machine startup takes an average of about 8.5 seconds, and the lack of approval authority poses a risk of accidental startup. This system, however, controls the startup response time to between 1.8 and 2.2 seconds, ensuring platform controllability, safety, and efficiency. Traditional methods lack real-time monitoring, while this system controls data latency to 150–180 milliseconds, providing strong real-time performance, identifying over 95% of abnormal behaviors, and keeping the false alarm rate below 5%, significantly reducing reliance on manual intervention. For example, traditional methods often result in current fluctuations within ±1.5A, while the optimized system achieves current stability of ±0.3 to 0.6A, effectively improving welding quality.

[0036] The approval process has also been significantly optimized. Traditional paper-based processes take an average of 10-15 minutes and are prone to omissions and errors. The system platform's average approval time is 2-4 minutes, and approvals automatically take effect upon device binding, preventing any unauthorized operations. More importantly, regarding equipment maintenance, this system can predictively remind users of maintenance 4-6 hours in advance based on actual operating data, while the traditional model relies entirely on manual inspections, posing a risk of equipment operating with defects. Furthermore, after the system's deployment, the average uptime of the five welding machines increased by over 13%, demonstrating significant value in optimizing production efficiency.

[0037] By deploying the system of this invention, enterprises have achieved, for the first time, full-process digital monitoring and closed-loop parameter control of welding equipment status, greatly improving operational safety, operational standardization, and management efficiency. From remote platform operation to predictive equipment maintenance, from intelligent approval binding to data report generation, the system breaks down multiple "information silos" in welding operations, realizing collaborative linkage between the perception layer, platform layer, and application layer. Practice has proven that the system of this invention is particularly suitable for industrial scenarios with high welding task intensity, diverse equipment, and high process risks, bringing a new model of visible, controllable, and predictable welding operations.

[0038] Table 1 Comparison of the effects of traditional welding methods and the system of this invention ; As shown in Table 1, traditional welding methods generally employ manual initiation, resulting in an average response time as high as 8.5 seconds. This is not only inefficient but also prone to issues such as duplicate commands and misoperation in multi-device parallel operation environments. In contrast, the system of this invention issues control commands through an IoT platform, significantly reducing the welding machine response time to between 1.8 and 2.2 seconds. This greatly improves control accuracy and operational efficiency, making it particularly suitable for welding tasks with frequent starts and stops and high scheduling density.

[0039] In terms of real-time monitoring capabilities, traditional welding operations rely heavily on manual inspections or post-process data collection, lacking process-level real-time data support. Information delays often exceed 500 milliseconds or even longer, failing to meet the dynamic assurance requirements for welding quality. This system, however, can control data latency between 150 and 180 milliseconds, enabling stable millisecond-level real-time monitoring and providing reliable basic data support for subsequent anomaly identification and parameter optimization.

[0040] In terms of intelligent identification and early warning, this system combines variational autoencoders and bird flocking optimization algorithms to model welding behavior, achieving an anomaly recognition rate of over 95%. It successfully identified potential risks in multiple test samples. Traditional methods, on the other hand, rely primarily on welder experience or manual observation, resulting in limited identification capabilities and a high false alarm rate approaching 20%. This system significantly reduces the false alarm rate to below 5%, greatly minimizing unintended interventions and improving the stability of safety monitoring.

[0041] Regarding the control of welding parameters, the system of this invention can achieve current stability control between ±0.3 and ±0.6 Amperes. Compared with the traditional method, which generally fluctuates within a range of ±1.5 Amperes, the fluctuation is small and the precision is high, which significantly improves the quality and consistency of the weld. This is especially important in manufacturing processes where welding strength and appearance requirements are high.

[0042] Regarding hot work permits, traditional approval methods typically involve paper-based or separate electronic processes, with approval cycles of 10 to 15 minutes. Furthermore, information is difficult to synchronize with the equipment level, posing risks such as "approval completed but equipment not authorized" or "unapproved equipment already started." This system integrates the approval process into the platform operation, reducing approval time to 2 to 4 minutes. Approval results are automatically linked to equipment, eliminating unauthorized operations and effectively improving operational standardization and process closure capabilities.

[0043] In terms of equipment operation and maintenance, traditional methods rely on regular manual inspections, lacking health status assessments, which easily leads to untimely maintenance or over-maintenance. This system can automatically assess the health level based on the welding machine's operating status and send reminders 4 to 6 hours before potential failures occur, enabling predictive maintenance. The proactiveness and accuracy of maintenance response are superior to traditional methods, significantly reducing the risk of unplanned downtime.

[0044] Finally, regarding overall equipment utilization efficiency, through the synergistic effects of accelerated approval, intelligent parameter adjustment, automatic early warning, and predictive maintenance, this system has achieved a significant improvement in equipment uptime. The average uptime of the five welding machines increased by more than 13%, fully validating the system's application value and potential for widespread adoption in actual industrial production.

[0045] In summary, the comparison clearly shows that the system of the present invention is significantly superior to traditional welding methods in many key performance indicators. It not only improves control efficiency, welding quality and operational safety, but also realizes the transformation from "passive response" to "proactive prediction" operation and maintenance mode, demonstrating the practicality and advancement of modern IoT intelligent welding systems in industrial manufacturing environments.

[0046] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A remote welding machine control method based on the Internet of Things, characterized in that, Includes the following steps: S1. Collect multi-source time-series data of remote welding machines through IoT devices, preprocess the multi-source time-series data, and generate a time-series dataset; S2. Based on the time series dataset, construct a variational autoencoder model. The encoder maps the time series dataset to the latent space, and the decoder reconstructs the latent variables into the original data. S3. The hyperparameters of the variational autoencoder model are optimized by applying the bird flock optimization algorithm. The variational autoencoder is then trained using the hyperparameters optimized by the bird flock optimization algorithm to obtain the optimized variational autoencoder model. S4. Based on the real-time sensor data of the welding machine, input the optimized variational autoencoder model, use the encoder to extract latent features, the decoder to reconstruct the data, and transmit the results in real time through the Internet of Things platform. S5. Based on the comparative analysis of the output of the optimized variational autoencoder model and the actual welding data, the welding quality is evaluated, and the evaluation results are fed back to the welding machine control system through the Internet of Things to adjust the welding control parameters. S6. Based on the adjusted welding control parameters, real-time welding operations are performed under the command drive of the IoT platform. Real-time data is continuously collected and status analysis is performed. At the same time, a unique identity binding and permission verification mechanism is combined with the one machine, one code identification mechanism to bind the welding machine to a unique identity and verify permissions. Only operation tasks that have been approved and authorized are allowed to issue control commands. S7. Establish an intelligent hot work approval mechanism. Before welding operations, real-name registration and approval must be carried out through the platform. Only after approval can the welding machine be remotely started. Unauthorized tasks must not activate the equipment. S8. Construct a welding condition monitoring and operation and maintenance management module to automatically push early warning information, generate maintenance plans and push reminders; S9. Construct a platform-based data and scheduling module to uniformly manage welding equipment status, task approval, early warning and maintenance information through the Internet of Things, generate analysis reports regularly, and support intelligent decision-making and operation and maintenance under multi-terminal hierarchical permissions.

2. The remote welding machine control method based on the Internet of Things according to claim 1, characterized in that, The multi-source time-series data specifically includes welding current, temperature, pressure, vibration, and welding quality data, which are used to analyze the welding process in real time and optimize welding control parameters.

3. The remote welding machine control method based on the Internet of Things according to claim 1, characterized in that, The preprocessing of the multi-source time series data specifically includes denoising, standardization, and data fusion, which are used to improve the quality of the multi-source time series data and provide accurate input for the variational autoencoder model.

4. The remote welding machine control method based on the Internet of Things according to claim 1, characterized in that, S2 specifically includes: S21. Receive the preprocessed time series dataset and pass it as input data to the variational autoencoder model. S22. Design and construct the structure of a variational autoencoder model, including an encoder and a decoder, where the encoder's role is to extract features from a time-series dataset and map the features to a latent space; S23. In the encoder part, a multi-layer fully connected neural network is used to process the time series dataset. The encoder maps the input time series dataset to the latent space through the multi-layer network to generate the mean and variance of the latent variables, which are used to describe the latent distribution of the time series dataset. S24. Apply the reparameterization technique to transform the mean and variance parameters of the latent variables output by the encoder into latent variables. S25. In the decoder section, a multi-layer fully connected neural network is used to map the latent variables obtained from the latent space back to the original data space. The output generated by the decoder is the reconstructed time series data. S26. By minimizing the reconstruction error and KL divergence, a variational autoencoder model is trained, and the parameters of the encoder and decoder are optimized. The variational autoencoder model can effectively reconstruct time series data and ensure that the data distribution in the latent space is consistent with the standard normal distribution.

5. The remote welding machine control method based on the Internet of Things according to claim 1, characterized in that, S3 specifically includes: S31. Initialize the bird flock. During initialization, each bird represents a combination of hyperparameters of the variational autoencoder model. Hyperparameters include the number of hidden layers, the number of neurons per layer, the dimension of the latent space, and the learning rate. The initial position of each bird represents a set of random hyperparameters. The initial velocity of each bird is determined according to the set initial search space range. The bird's initial position is ,in, The number of hidden layers in the encoder. The number of neurons per layer, For the dimension of the potential space, Let be the learning rate, and be the initial velocity. ,in, , , , This represents the components of the initial velocity in each hyperparameter dimension; S32. Real-time welding data during the welding process is collected via IoT devices. ,in, For welding current, For welding temperature, The welding pressure is used as the real-time welding data as feedback input to the bird flock optimization algorithm, which is used to dynamically adjust the search strategy and affect the flight mode and hyperparameter adjustment in real time. S33. Each bird coordinates its flight based on its historical best position, global best position, and the flight status of other birds in its neighborhood. The flight status includes speed, direction, and position. The influence of neighborhood information is dynamically adjusted based on feedback from real-time welding data, updating the speed of each bird. ； in, For the first Only one bird in the first The speed of generation For the first Only one bird in the first The position of the generation, For the first Only one bird in the first The position of the generation, For the first The best historical position for a bird This is the optimal position for the entire flock of birds. For the first The neighboring area of ​​a single bird For the first Only bird and the first The distance between the birds , , , For constant parameters, , , It is a random number; S34. Based on real-time welding data, optimize and adjust the hyperparameter combination for each bird in real time. Through real-time welding quality prediction and dynamic parameter adjustment, evaluate the hyperparameter combination corresponding to each bird and calculate the welding quality index. : ； in, , and They are the first The current, temperature, and pressure values ​​during the welding process under the corresponding hyperparameters for a single bird. , and The target welding current, temperature, and pressure are the ideal values. , , This is a weighting adjustment factor used to adjust the degree of influence of welding current, welding temperature, and welding pressure on welding quality assessment. S35. Based on feedback from real-time welding data, dynamically adjust the weights of the global and local searches for bird flocks. If temperature fluctuations are significant during the welding process, enhance the global search capability; conversely, enhance the local fine search when the welding process is stable. Introduce a welding process error feedback mechanism, adjusting the weights of the global and local searches based on the welding quality prediction error. When the welding quality deviates from expectations, strengthen the global search; conversely, strengthen the local search. The adjustment of the global and local search weights is as follows: ； ； in, This serves as the global search weight, used to dynamically adjust the breadth of the search strategy. These are local search weights used to dynamically adjust the precision of the search strategy. Welding current and reference current The difference is used to reflect changes in welding current. For welding temperature and reference temperature The difference is used to reflect fluctuations in welding temperature. For welding quality prediction error and target error The difference is used to reflect the deviation in welding quality. and These represent the reference current and reference temperature during the welding process, respectively. The target welding quality error is calculated by real-time monitoring of the deviation between the welding quality and the set target. , , This is a weighting adjustment factor; S36. A dynamic flight mode adjustment mechanism is introduced. Based on real-time welding data, the flight mode of the bird flock can switch between broad search and fine search. The broad search phase uses global search weights, while the fine search phase uses local search weights. When the welding current fluctuates significantly, the global search intensity is increased. ； When welding conditions are stable, local search weights are used: ； in, For global search, the first Only one bird in the first The speed of generation For local search, the first Only one bird in the first The speed of generation and The acceleration constant, and It is a random number; S37. Update the positions and speeds of the flock members, calculate the new fitness values, and evaluate the effect of the new position and hyperparameter combination for each bird: ； in, For the first The fitness value of a single bird. For the first The welding quality assessment value of a single bird. For the first The welding temperature of a single bird under the current hyperparameter combination. For the first The welding material consumption for each bird This represents the total welding time. For the target welding time, To optimize the target temperature, For the target welding material consumption, For the target welding time, , and As a regulating factor; S38. Based on the updated fitness values, when the maximum number of iterations is reached or the fitness threshold is exceeded, the optimal hyperparameter configuration is output. The optimal hyperparameter configuration is then used to train the final variational autoencoder model. for: ； in, The optimal number of hidden layers for the encoder. To determine the optimal number of neurons per layer, For the optimal potential spatial dimension, The optimal learning rate; S39. Finally, the optimized variational autoencoder model is obtained. The optimized variational autoencoder model can perform data reconstruction and welding quality prediction more accurately.

6. The remote welding machine control method based on the Internet of Things according to claim 1, characterized in that, S4 specifically includes: S41. Real-time welding data during the welding process is collected through IoT sensors, including welding current, welding temperature, and welding pressure. S42. The collected real-time welding data is transmitted to the optimized variational autoencoder model. The encoder part maps the real-time welding data to the latent space and extracts the latent features related to welding quality and production efficiency during the welding process. The optimized variational autoencoder model can automatically adjust the feature extraction strategy based on historical data and real-time feedback. S43. Input the latent space representation extracted by the encoder into the decoder. The decoder reconstructs the welding parameters based on the latent variables and generates optimized welding data output. S44. By comparing the error between the reconstructed data generated by the decoder and the actual real-time acquired welding data, an adaptive error correction mechanism is introduced. The error feedback will correct the deviation in the welding process by dynamically adjusting the weights. ； in, For the first Reconstruction error of each welding parameter and The first The values ​​of each welding parameter in the actual data and the reconstructed data. It is a dynamic adjustment factor. It is an error feedback item used to adjust the error correction strength for each welding parameter; S45. Based on the calculated reconstruction error and error feedback, dynamically optimize the learning rate and hyperparameters of the variational autoencoder model. By introducing a multi-stage adaptive learning rate adjustment mechanism, the learning rate is adjusted according to real-time errors and environmental fluctuations at different stages of the welding process to optimize welding quality. ； in, For the updated learning rate, To optimize the learning rate, let represent the learning rate under ideal conditions. and This is an adjustment factor used to control the degree to which error affects the learning rate adjustment. As an adaptive factor, Indicates welding fluctuations; S46. The optimized welding data is transmitted to the remote control system in real time through the Internet of Things (IoT) platform, so that the operator can monitor and adjust it in real time. The IoT platform automatically adjusts the control parameters in the welding process based on real-time feedback and the prediction of the optimized variational autoencoder model.

7. The remote welding machine control method based on the Internet of Things according to claim 1, characterized in that, S5 specifically includes: S51. Obtain the output results of the optimized variational autoencoder model. The output results include the prediction data and reconstruction data of the welding process by the optimized variational autoencoder model. S52. Acquire real-time welding data. The real-time welding data is collected in real time by IoT sensors, including the actual welding current, welding temperature and welding pressure. S53. Compare and analyze the reconstructed data output by the optimized variational autoencoder model with the real-time welding data acquired in real time to identify deviations or abnormalities during the welding process. S54. Based on the results of comparative analysis, evaluate the welding quality. The evaluation criteria include the stability and accuracy of the welding parameters and whether they meet the preset quality standards. S55. Feed the evaluation results back to the Internet of Things (IoT) platform. The IoT platform will transmit the welding quality evaluation results to the welding machine control system through real-time data transmission. S56. Based on the evaluation results, the welding machine control system triggers an early warning mechanism based on the specific evaluation data. If the evaluation results indicate that the welding quality is substandard, the system will take measures, including suspending welding, sending an alarm, or further analyzing the data.

8. The remote welding machine control method based on the Internet of Things according to claim 1, characterized in that, The welding status monitoring and operation and maintenance management module is built using IoT sensing devices, and the platform-based data and scheduling module is built using an IoT communication architecture.

9. A remote welding machine control system based on the Internet of Things (IoT), comprising the remote welding machine control method based on the IoT as described in any one of claims 1 to 8, characterized in that, Includes the following modules: The remote control module is used to receive welding machine control commands in real time through the Internet of Things platform, and control the start, stop and change the operating status of the welding machine according to the commands; The real-time monitoring module is used to collect the welding machine's operating status data in real time and monitor key parameters during the welding process; The fire monitoring and approval module is used to detect fire risks in real time during the welding process and to approve them. The device identification and binding module is used to assign a unique identifier to each welding machine and bind it. The data analysis and risk warning module is used to analyze various data in the welding process and conduct risk assessments, issuing warnings when potential problems are discovered. The report generation and optimization module is used to generate a report on the welding process based on real-time collected data, and to optimize and adjust the welding parameters based on the evaluation results. The welding status monitoring and operation and maintenance management module is used to identify welding anomalies in real time and generate maintenance plans, and to complete welding safety early warning and equipment health management. The platform-based data and scheduling module is used to uniformly schedule welding data and control tasks, supporting intelligent decision-making and collaborative operation and maintenance under multi-terminal hierarchical permissions.

Images