High-temperature early warning method and system for welding device

Abstract

The invention relates to the technical field of data processing, and discloses a high-temperature early warning method and system for a welding device. The method comprises the following steps: carrying out multi-point temperature measurement by using a K-type thermocouple and an infrared thermal imager; fourier heat conduction calculation is conducted on the welding heat field; extracting temperature anomaly region features through thermal gradient analysis; carrying out multi-parameter risk assessment by combining material melting point parameters; a multi-channel alarm instruction is output through the industrial control bus; and current-voltage curve real-time adjustment is conducted on the welding power source, and closed-loop self-adaptive control is achieved. According to the method, a complete closed-loop system from multi-sensor temperature collection, thermal field analysis, abnormal feature extraction and risk assessment to multi-stage alarm and self-adaptive control is established, so that early recognition and accurate control of thermal abnormality in the welding process are realized, and the welding quality and the production safety are remarkably improved.

Description

Technical Field

[0001] This application relates to the technical field of data processing, and particularly to a high-temperature warning method and system for a welding device. Background Art

[0002] Temperature control during the welding process has always been a key factor in ensuring welding quality and safety. Traditional welding temperature monitoring mainly relies on infrared thermometers or thermocouples for single-point measurement, and manual judgment and adjustment are carried out in combination with a fixed welding process parameter table. With the development of automated welding technology, closed-loop temperature control systems based on PID control and molten pool monitoring systems based on image processing have emerged. These systems collect temperature information in the welding area, compare it with a preset threshold, and issue an alarm or automatically adjust welding parameters when an abnormality is detected. Among them, relatively advanced methods include dynamic monitoring technology of welding molten pools based on infrared thermal imagers, welding parameter adaptive adjustment systems using neural network algorithms, and intelligent monitoring systems for welding processes combining multi-sensor fusion technology, etc.

[0003] However, there are still many deficiencies in the existing technologies. Firstly, the temperature measurement method using a single sensor has problems of incomplete spatial coverage and unstable measurement accuracy, making it difficult to accurately capture the temperature distribution characteristics of the entire welding area. Secondly, the traditional threshold alarm method is too simple and cannot perform differential processing according to different material and process characteristics, resulting in frequent false alarms or missed alarms. Thirdly, the alarm and control systems are usually independent of each other, and the alarm information cannot be effectively converted into precise control strategies, delaying the system response time. Fourthly, most of the existing temperature monitoring systems are limited to static threshold judgment, lacking in-depth analysis of the dynamic change trend and spatial distribution characteristics of temperature, and unable to foresee potential thermal abnormality risks. Finally, the adjustment of welding parameters usually adopts a fixed adjustment amount, without considering changes in material properties and heat conduction characteristics, which easily leads to parameter oscillation or insufficient adjustment. These problems seriously restrict the stability of the welding process and the consistency of product quality. Summary of the Invention

[0004] This application provides a high-temperature warning method and system for a welding device, which is used to realize early identification and precise control of thermal abnormalities during the welding process by establishing a complete closed-loop system from multi-sensor temperature acquisition, thermal field analysis, abnormal feature extraction, risk assessment to multi-level alarm and adaptive control, and significantly improve welding quality and production safety.

[0005] In a first aspect, the present application provides a high-temperature warning method for a welding device. The high-temperature warning method for the welding device includes: performing multi-point temperature measurement on the surface of a welding workpiece through a K-type thermocouple temperature sensor and a non-cooled infrared thermal imager to obtain real-time temperature data with a sampling frequency of 10 Hz; performing Fourier heat conduction calculation on the welding thermal field according to the real-time temperature data to form a hot spot distribution grid map with a spatial resolution of 0.5 mm; using the hot spot distribution grid map to extract boundary features of a temperature anomaly region exceeding a preset threshold through thermal gradient difference analysis to generate an abnormal heat zone feature set including position coordinates and temperature rise rates; performing multi-parameter weighted calculation on the risk level according to the abnormal heat zone feature set in combination with the material melting point parameter to form a warning level matrix divided into four levels: low, medium, high, and emergency; transmitting differential signals to an acoustic-optic integrated alarm system according to the warning level matrix through an industrial control bus protocol to output multi-channel alarm instructions with frequency encoding; and performing real-time adjustment of the current-voltage curve of a welding power controller based on the multi-channel alarm instructions to achieve closed-loop adaptive control of welding parameters.

[0006] In the first implementation manner of the first aspect, the performing multi-point temperature measurement on the surface of a welding workpiece through a K-type thermocouple temperature sensor and a non-cooled infrared thermal imager to obtain real-time temperature data with a sampling frequency of 10 Hz includes: arranging multiple groups of K-type thermocouples around the welding device according to a grid topology structure to perform contact-type temperature acquisition on the surface of the workpiece to obtain a target temperature data set; performing thermal radiation imaging scanning on the welding area through a non-cooled infrared thermal imager to obtain a thermal map of the temperature distribution on the surface of the workpiece; performing noise filtering processing on the target temperature data set to generate a calibrated contact temperature value; performing pixel temperature mapping conversion on the thermal map of the temperature distribution on the surface of the workpiece to form a non-contact temperature matrix; performing data fusion on the calibrated contact temperature value and the non-contact temperature matrix to eliminate emissivity error and obtain a corrected temperature field; and packaging and transmitting the corrected temperature field through an industrial field bus at a sampling frequency of 10 Hz to form the real-time temperature data.

[0007] In the second implementation manner of the first aspect, the performing Fourier heat conduction calculation on the welding thermal field according to the real-time temperature data to form a hot spot distribution grid map with a spatial resolution of 0.5 mm includes: Convert the real-time temperature data into temperature spectrum domain data to obtain the frequency characteristics of the temperature spatial distribution; perform heat flux density expansion on the temperature spectrum domain data through the forward Fourier transform to obtain the basic heat conduction equation; construct a heat diffusion difference format according to the basic heat conduction equation to form a heat flux propagation vector field; perform grid subdivision processing on the heat flux propagation vector field to establish a discrete calculation grid with a precision of 0.5 mm; perform iterative solution on the discrete calculation grid through heat flux boundary condition constraints to obtain the steady-state and transient temperature distributions; perform visualization processing on the steady-state and transient temperature distributions through pseudo-color mapping technology to generate the hot spot distribution grid map.

[0008] In the third implementation manner of the first aspect, the boundary characteristics of the temperature anomaly region exceeding the preset threshold are extracted by using the hot spot distribution grid map through heat gradient difference analysis to generate an abnormal hot zone feature set including position coordinates and temperature rise rate, including: calculating the temperature difference gradient of the hot spot distribution grid map to generate a temperature change rate distribution map; performing binary processing on the temperature change rate distribution map according to the material safety threshold to obtain a temperature anomaly candidate area; performing connected domain marking on the temperature anomaly candidate area through an improved region growing algorithm to form a hot zone contour boundary; extracting center coordinates, area, and perimeter features from the hot zone contour boundary to construct a hot zone spatial feature table; performing temporal comparison on the same hot zone in multiple consecutive frames of the hot spot distribution grid map, calculating the temperature change speed of each hot zone to obtain the hot zone temperature rise rate data; combining and processing the hot zone spatial feature table and the hot zone temperature rise rate data to generate the abnormal hot zone feature set.

[0009] In the fourth implementation manner of the first aspect, the risk level is calculated by multi-parameter weighting according to the abnormal hot zone feature set combined with the material melting point parameter to form an early warning level matrix divided into four levels: low, medium, high, and urgent, including: querying the melting point, flash point, and safe working temperature of the welded workpiece corresponding to the material database to construct a material thermal characteristic threshold table; calculating the ratio of the temperature value in the abnormal hot zone feature set to the material thermal characteristic threshold table to obtain a temperature danger index; dividing the temperature rise rate in the abnormal hot zone feature set by the material safe temperature rise rate to obtain a temperature rise danger coefficient; calculating a hot zone diffusion index according to the proportional relationship between the hot zone area in the abnormal hot zone feature set and the standard area of the welding point; performing weighted summation on the temperature danger index, the temperature rise danger coefficient, and the hot zone diffusion index to generate a comprehensive risk score; mapping the comprehensive risk score into four levels: low, medium, high, and urgent through multi-level threshold segmentation to form the early warning level matrix.

[0010] In the fifth implementation of the first aspect, the differential signal transmission to the opto-acoustic-electronic integrated alarm system through the industrial control bus protocol according to the early warning level matrix, and the output of multi-channel alarm instructions with frequency encoding, includes: mapping the low-level early warning in the early warning level matrix to a green flashing signal encoding to generate a visual warning data packet; converting the medium-level early warning in the early warning level matrix into a medium-frequency beeping pulse sequence to form an audio warning signal; constructing a tactile feedback instruction with vibration intensity parameters according to the high-level early warning in the early warning level matrix to obtain a tactile alarm code; compiling an emergency response instruction containing the device ID and the fault type for the emergency early warning in the early warning level matrix to generate a system control command; packing the visual warning data packet, the audio warning signal, the tactile alarm code, and the system control command according to the priority to construct a multi-channel instruction data stream; and sending the multi-channel instruction data stream to each terminal device in a multicast manner through the industrial Ethernet to output the multi-channel alarm instructions with frequency encoding.

[0011] In the sixth implementation of the first aspect, the real-time adjustment of the current-voltage curve of the welding power controller based on the multi-channel alarm instructions to achieve closed-loop adaptive control of welding parameters, includes: parsing the risk level identifier and the hot spot position information from the multi-channel alarm instructions to generate a welding danger status table; querying the process parameter library according to the welding danger status table to obtain the corresponding current-voltage adjustment strategy; converting the current-voltage adjustment strategy into a digital signal control quantity to form a power control data stream; dynamically compensating the power control data stream according to the welding material characteristic curve to obtain a smoothly transitioning current-voltage adjustment curve; converting the current-voltage adjustment curve into an analog control signal through a digital-to-analog converter and outputting it to the welding power module; and the welding power module dynamically adjusting the welding parameters in real time according to the analog control signal to complete the closed-loop adaptive control of the welding parameters.

[0012] In the second aspect, the present application provides a high-temperature early warning system for a welding device, and the high-temperature early warning system for the welding device includes: An extraction module, configured to perform multi-point temperature measurement on the surface of the welding workpiece through a K-type thermocouple temperature sensor and a non-cooled infrared thermal imager to obtain real-time temperature data with a sampling frequency of 10 Hz; A calculation module, configured to perform Fourier heat conduction calculation on the welding thermal field according to the real-time temperature data to form a hot spot distribution grid map with a spatial resolution of 0.5 mm; An extraction module, configured to extract the boundary features of the temperature anomaly region exceeding the preset threshold through thermal gradient difference analysis by using the hot spot distribution grid map to generate an abnormal thermal zone feature set including position coordinates and temperature rise rate; A weighting module, configured to perform multi-parameter weighting calculation on the risk level according to the abnormal hot zone feature set in combination with the material melting point parameter, so as to form an early warning level matrix divided into four levels: low, medium, high, and emergency; A transmission module, configured to perform differential signal transmission on the sound-light-electricity integrated alarm system according to the early warning level matrix through the industrial control bus protocol, and output multi-channel alarm instructions with frequency encoding; An adjustment module, configured to perform real-time adjustment of the current-voltage curve of the welding power supply controller based on the multi-channel alarm instructions, so as to achieve closed-loop adaptive control of the welding parameters.

[0013] The third aspect of the present invention provides a computer device, including: a memory and at least one processor, wherein instructions are stored in the memory; the at least one processor calls the instructions in the memory, so that the computer device executes the above-mentioned high-temperature early warning method for a welding device.

[0014] The fourth aspect of the present invention provides a computer-readable storage medium, wherein instructions are stored in the computer-readable storage medium, and when the instructions are run on a computer, the computer is made to execute the above-mentioned high-temperature early warning method for a welding device.

[0015] In the technical solution provided by this application, the surface of the welded workpiece is measured for multi-point temperature through a K-type thermocouple temperature sensor and a non-cooled infrared thermal imager, and real-time temperature data with a sampling frequency of 10 Hz is obtained, achieving a comprehensive coverage and high-precision acquisition of the temperature field in the welding area, solving the problems of measurement blind spots and large errors of a single sensor, and eliminating the emissivity error through data fusion technology, improving the accuracy of temperature measurement; Fourier heat conduction calculation is performed on the welding thermal field according to the real-time temperature data to form a hot spot distribution grid map with a spatial resolution of 0.5 mm, realizing the accurate calculation of the heat conduction path and temperature distribution, enabling the system to capture temperature anomalies at the millimeter level; using the hot spot distribution grid map, boundary features of the temperature anomaly area exceeding the preset threshold are extracted through thermal gradient difference analysis, generating an abnormal heat zone feature set containing position coordinates and temperature rise rates. An improved region growing algorithm is introduced to intelligently identify and extract the temperature anomaly area. This algorithm is optimized for the characteristics of the welding heat zone, can accurately distinguish normal thermal gradients and abnormal hot spots, significantly reduces the false alarm rate, and enhances the system's ability to identify different types of thermal anomalies; according to the abnormal heat zone feature set and combining with the material melting point parameters, multi-parameter weighted calculation is performed on the risk level to form an early warning level matrix divided into four levels: low, medium, high, and emergency. By introducing material characteristic parameters and multi-dimensional feature analysis, the accuracy and differentiation of risk assessment are realized, and the system can give a more scientific risk judgment according to the thermal sensitivity of different materials; according to the early warning level matrix, differential signal transmission is carried out on the sound-light integrated alarm system through the industrial control bus protocol, and multi-channel alarm instructions with frequency encoding are output, establishing a seamless connection mechanism from risk assessment to alarm response. The multi-channel alarm method ensures that information can be timely perceived in various working environments; based on the multi-channel alarm instructions, real-time adjustment of the current-voltage curve of the welding power controller is carried out to realize closed-loop adaptive control of welding parameters. This closed-loop control mechanism dynamically adjusts welding parameters according to the abnormal heat zone characteristics and material characteristics. The system considers the change of the resistance characteristics of the material at different temperatures and adopts a smooth transition parameter adjustment strategy to avoid fluctuations and instabilities during the parameter adjustment process, improving the welding quality and consistency, further improving the analysis accuracy and response speed of the system, and significantly enhancing the system's adaptability to complex welding scenarios, enabling the early warning system to handle various non-linear thermal anomaly situations, effectively reducing the welding defect rate and rework rate, and greatly improving the production efficiency and product quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0017] Figure 1 It is a schematic diagram of an embodiment of the high-temperature early warning method for a welding device in an embodiment of the present application; Figure 2 It is a schematic diagram of an embodiment of the high-temperature early warning system for a welding device in an embodiment of the present application; Figure 3 It is a structural schematic block diagram of a computer device in an embodiment of the present invention. Specific implementation manners

[0018] The embodiments of the present application provide a high-temperature early warning method and system for a welding device. Terms such as "first", "second", "third", "fourth", etc. (if any) in the specification, claims and above-mentioned drawings of the present application are used to distinguish similar objects, and do not have to be used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments described here can be implemented in an order other than that illustrated or described here. In addition, the terms "include" or "have" and any deformation thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device including a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0019] For ease of understanding, the specific process of the embodiments of the present application will be described below. Please refer to Figure 1 , an embodiment of the high-temperature early warning method for a welding device in an embodiment of the present application includes: Step S101: Perform multi-point temperature measurement on the surface of the welded workpiece through a K-type thermocouple temperature sensor and a non-cooled infrared thermal imager to obtain real-time temperature data with a sampling frequency of 10 Hz; Step S102: Perform Fourier heat conduction calculation on the welding thermal field according to the real-time temperature data to form a hot spot distribution grid map with a spatial resolution of 0.5 mm; Step S103: Use the hot spot distribution grid map to extract the boundary features of the temperature anomaly area exceeding the preset threshold through thermal gradient difference analysis, and generate an abnormal hot zone feature set including position coordinates and temperature rise rate; Step S104: Perform multi-parameter weighted calculation on the risk level according to the abnormal hot zone feature set combined with the material melting point parameter to form an early warning level matrix divided into four levels: low, medium, high, and emergency; Step S105: According to the early warning level matrix, perform differential signal transmission on the sound and light integrated alarm system through the industrial control bus protocol, and output multi-channel alarm instructions with frequency encoding; Step S106: Based on the multi-channel alarm instruction, the current-voltage curve of the welding power controller is adjusted in real time to achieve closed-loop adaptive control of welding parameters.

[0020] It can be understood that the execution subject of this application can be a high-temperature early warning system for welding devices, or a terminal or a server. Specifically, it is not limited here. In this embodiment of the application, the server is used as the execution subject for illustration.

[0021] Specifically, the surface of the welding workpiece is measured for temperature at multiple points by a K-type thermocouple temperature sensor and a non-cooled infrared thermal imager to obtain real-time temperature data with a sampling frequency of 10 Hz. When implementing this step, multiple groups of K-type thermocouples are arranged around the welding device according to a grid topology structure to perform contact-type temperature acquisition on the surface of the workpiece. The K-type thermocouple is a temperature sensor composed of chromium-nickel - nickel-silicon alloy, with a temperature measurement range of -200°C to 1300°C, suitable for monitoring the high-temperature environment of welding. At the same time, the non-cooled infrared thermal imager performs thermal radiation imaging scanning on the welding area to obtain a thermal image of the temperature distribution on the surface of the workpiece. The non-cooled infrared thermal imager uses the principle of thermal radiation to capture the temperature distribution without contacting the workpiece, solving the problem that the K-type thermocouple cannot cover all areas. The collected target temperature data set is processed by noise filtering to generate calibrated contact temperature values, and the thermal image of the temperature distribution on the surface of the workpiece is subjected to pixel temperature mapping conversion to form a non-contact temperature matrix. Then, the calibrated contact temperature values and the non-contact temperature matrix are subjected to data fusion to eliminate the emissivity error and obtain a corrected temperature field. The corrected temperature field is packaged and transmitted at a sampling frequency of 10 Hz through an industrial field bus to form real-time temperature data. According to the obtained real-time temperature data, Fourier heat conduction calculation is performed on the welding thermal field to form a hot spot distribution grid map with a spatial resolution of 0.5 mm. The real-time temperature data is converted into temperature spectral domain data to obtain the frequency characteristics of the temperature spatial distribution. The heat flux density is expanded for the temperature spectral domain data through Fourier forward transform to obtain the basic heat conduction equation. The Fourier heat conduction equation describes the variation law of temperature with time and space, and through this equation, the propagation path of heat in the welding workpiece can be predicted. According to the basic heat conduction equation, a thermal diffusion difference format is constructed to form a heat flux propagation vector field. The heat flux propagation vector field is subjected to grid subdivision processing to establish a discrete calculation grid with an accuracy of 0.5 mm. The discrete calculation grid is iteratively solved through heat flux boundary condition constraints to obtain the steady-state and transient temperature distributions. The steady-state and transient temperature distributions are visualized through pseudo-color mapping technology to generate a hot spot distribution grid map. The pseudo-color mapping technology maps different temperature values to different colors, making the temperature distribution intuitively visible.

[0022] Using the hotspot distribution grid map, boundary features of temperature anomaly regions exceeding the preset threshold are extracted through thermal gradient difference analysis to generate an abnormal hot zone feature set containing position coordinates and temperature rise rates. The temperature difference gradient of the hotspot distribution grid map is calculated to generate a temperature change rate distribution map. The temperature difference gradient calculation is achieved by dividing the temperature difference between adjacent grid points by the grid distance, which reflects the spatial change rate of temperature. The temperature change rate distribution map is binarized according to the material safety threshold to obtain candidate temperature anomaly regions. The binarization process marks the regions where the temperature change rate is greater than the safety threshold as abnormal regions, otherwise as normal regions. The candidate temperature anomaly regions are marked for connected components through an improved region growing algorithm to form the boundary of the hot zone contour. The improved region growing algorithm starts from the seed point and gradually expands to adjacent regions that meet the similarity conditions with the seed point until no further expansion is possible. The central coordinates, area, and perimeter features are extracted from the hot zone contour boundary to construct a hot zone spatial feature table. The same hot zones in consecutive frames of the hotspot distribution grid map are compared temporally, and the temperature change speed of each hot zone is calculated to obtain the hot zone temperature rise rate data. The hot zone spatial feature table and the hot zone temperature rise rate data are combined and processed to generate an abnormal hot zone feature set. Based on the abnormal hot zone feature set and the material melting point parameter, a multi-parameter weighted calculation of the risk level is performed to form an early warning level matrix divided into four levels: low, medium, high, and urgent. The melting point, flash point, and safe working temperature of the welded workpiece are queried from the material database to construct a material thermal property threshold table. The temperature values in the abnormal hot zone feature set are calculated with the material thermal property threshold table to obtain the temperature hazard index. The temperature rise rate in the abnormal hot zone feature set is divided by the material safe temperature rise rate to obtain the temperature rise hazard coefficient. According to the proportional relationship between the hot zone area in the abnormal hot zone feature set and the standard area of the welding point, the hot zone diffusion index is calculated. The temperature hazard index, temperature rise hazard coefficient, and hot zone diffusion index are weighted and summed to generate a comprehensive risk score. The comprehensive risk score is mapped to four levels: low, medium, high, and urgent through multi-level threshold segmentation to form an early warning level matrix.

[0023] Differentially transmit signals to the opto-acoustic-electronic integrated alarm system via the industrial control bus protocol according to the early warning level matrix, and output multi-channel alarm instructions with frequency encoding. Map the low-level early warning in the early warning level matrix to a green flashing signal encoding to generate a visual warning data packet. Convert the medium-level early warning in the early warning level matrix into a medium-frequency buzzer sound pulse sequence to form an audio warning signal. Construct a tactile feedback instruction with vibration intensity parameters according to the high-level early warning in the early warning level matrix to obtain a tactile alarm code. Compile an emergency response instruction containing the device ID and fault type for the emergency early warning in the early warning level matrix to generate a system control command. Pack the visual warning data packet, audio warning signal, tactile alarm code, and system control command according to the priority to construct a multi-channel instruction data stream. Send the multi-channel instruction data stream to each terminal device in a multicast manner via the industrial Ethernet to output multi-channel alarm instructions with frequency encoding.

[0024] Based on the multi-channel alarm instructions, perform real-time adjustment of the current-voltage curve of the welding power supply controller to achieve closed-loop adaptive control of welding parameters. Parse the risk level identifier and hot spot location information from the multi-channel alarm instructions to generate a welding danger status table. Query the process parameter library according to the welding danger status table to obtain the corresponding current-voltage adjustment strategy. Convert the current-voltage adjustment strategy into a digital signal control quantity to form a power control data stream. Perform dynamic compensation on the power control data stream according to the welding material characteristic curve to obtain a smoothly transitioning current-voltage adjustment curve. Convert the current-voltage adjustment curve into an analog control signal through a digital-to-analog converter and output it to the welding power supply power module. The welding power supply power module adjusts the welding parameters in real time and dynamically according to the analog control signal to complete the closed-loop adaptive control of welding parameters.

[0025] For example, when welding steel materials, K-type thermocouples are distributed in a 4×4 grid around the welding point, and each thermocouple real-time collects temperature values. The uncooled infrared thermal imager scans the entire workpiece surface simultaneously. After data fusion processing, corrected temperature field data is obtained, and a hot spot distribution grid map is formed through Fourier heat conduction calculation. It is detected that the temperature in a certain area near the welding point reaches 1100°C and the temperature rise rate is 25°C / second. Through thermal gradient analysis, this area is determined as a temperature anomaly area. Querying from the material database, it is known that the melting point of this steel is 1450°C and the safe operating temperature is 800°C. Calculate that the temperature danger index is 0.76, the temperature rise danger coefficient is 1.25, the hot zone diffusion index is 1.5, and the comprehensive risk score is 3.51, corresponding to the high-level early warning level. The system immediately emits a yellow flashing signal and a high-frequency sound alarm, and at the same time sends an adjustment instruction to the welding power supply controller to reduce the welding current from 200A to 175A and adjust the voltage from 28V to 25V, effectively controlling the further expansion of the abnormal temperature area and preventing workpiece deformation and welding defects.

[0026] In the embodiment of the present application, multi-point temperature measurement is performed on the surface of the welded workpiece through a K-type thermocouple temperature sensor and a non-cooled infrared thermal imager, and real-time temperature data with a sampling frequency of 10 Hz is obtained, realizing the comprehensive coverage and high-precision acquisition of the temperature field in the welding area, solving the problems of measurement blind spots and large errors of a single sensor, and eliminating the emissivity error through data fusion technology, improving the accuracy of temperature measurement; Fourier heat conduction calculation is performed on the welding thermal field according to the real-time temperature data to form a hot spot distribution grid map with a spatial resolution of 0.5 mm, realizing the accurate calculation of the heat conduction path and temperature distribution, enabling the system to capture temperature anomalies at the millimeter level; the boundary characteristics of the temperature anomaly area exceeding the preset threshold are extracted through the analysis of the thermal gradient difference using the hot spot distribution grid map, generating an abnormal hot area feature set containing position coordinates and temperature rise rates, and introducing an improved region growing algorithm to intelligently identify and extract the temperature anomaly area. This algorithm is optimized for the characteristics of the welding hot area, can accurately distinguish normal thermal gradients and abnormal hot spots, significantly reduces the false alarm rate, and enhances the system's ability to identify different types of thermal anomalies; according to the abnormal hot area feature set and combined with the material melting point parameters, multi-parameter weighted calculation is performed on the risk level to form an early warning level matrix divided into four levels: low, medium, high, and emergency. By introducing material characteristic parameters and multi-dimensional feature analysis, the accuracy and differentiation of risk assessment are realized, and the system can give a more scientific risk judgment according to the thermal sensitivity of different materials; according to the early warning level matrix, differential signal transmission is performed on the sound-light integration alarm system through the industrial control bus protocol, and multi-channel alarm instructions with frequency encoding are output, establishing a seamless connection mechanism from risk assessment to alarm response. The multi-channel alarm method ensures that information can be sensed in a timely manner in various working environments; based on the multi-channel alarm instructions, real-time adjustment of the current-voltage curve of the welding power controller is performed to realize closed-loop adaptive control of welding parameters. This closed-loop control mechanism dynamically adjusts welding parameters according to the abnormal hot area characteristics and material characteristics. The system considers the change of the resistance characteristics of the material at different temperatures and adopts a smooth transition parameter adjustment strategy to avoid fluctuations and instabilities during the parameter adjustment process, improving the welding quality and consistency, further improving the analysis accuracy and response speed of the system, and significantly enhancing the system's adaptability to complex welding scenarios, enabling the early warning system to handle various non-linear thermal anomaly situations, effectively reducing the welding defect rate and rework rate, and greatly improving the production efficiency and product quality.

[0027] In a specific embodiment, the process of executing step S101 may specifically include the following steps: (1) Arrange multiple groups of K-type thermocouples around the welding device according to the grid topology structure, perform contact temperature acquisition on the surface of the workpiece, and obtain the target temperature data set; (2) Perform thermal radiation imaging scanning on the welding area through a non-cooled infrared thermal imager to obtain a thermal map of the workpiece surface temperature distribution; (3) Perform noise filtering on the target temperature data set to generate calibrated contact temperature values; (4) Perform pixel temperature mapping conversion on the thermal map of the workpiece surface temperature distribution to form a non-contact temperature matrix; (5) Perform data fusion on the calibrated contact temperature values and the non-contact temperature matrix to eliminate the emissivity error and obtain a corrected temperature field; (6) Package and transmit the corrected temperature field at a sampling frequency of 10 Hz through an industrial fieldbus to form real-time temperature data.

[0028] Specifically, multiple groups of K-type thermocouples are arranged around the welding device according to a grid topology structure to perform contact temperature acquisition on the workpiece surface and obtain a target temperature data set. The K-type thermocouple is a temperature sensor composed of chromel-nickel silicon alloy, with a temperature measurement range of -200°C to 1300°C, featuring a fast response speed and high accuracy, and is suitable for high-temperature monitoring during the welding process. The grid topology structure means that the thermocouples are evenly distributed around the welding area in a matrix form, usually using a grid layout of 4×4 or 8×8 and other specifications to ensure full coverage of the welding area. Contact temperature acquisition means that the thermocouple directly contacts the workpiece surface and uses the Seebeck Effect to generate an electromotive force proportional to the temperature, which is amplified by an amplifier and then converted into a standard electrical signal after that. The target temperature data set contains the coordinate positions of each sampling point and the corresponding temperature values, forming discrete temperature distribution data. Perform thermal radiation imaging scanning on the welding area through a non-cooled infrared thermal imager to obtain a thermal map of the workpiece surface temperature distribution. The non-cooled infrared thermal imager is a thermal imaging device based on the principle of microbolometer, without a cooling system such as liquid nitrogen, and obtains temperature information by detecting the infrared radiation emitted from the object surface. During the thermal radiation imaging scanning process, the infrared thermal imager scans the welding area at a speed of more than 10 frames per second, receives infrared radiation signals from different positions on the workpiece surface, and converts the radiation intensity into temperature information according to Planck's blackbody radiation law. The thermal map of the workpiece surface temperature distribution is a two-dimensional pseudo-color image, where different colors represent different temperature values, usually using a rainbow color spectrum (from blue to red) to represent the temperature change from low to high.

[0029] Perform noise filtering on the target temperature dataset to generate calibrated contact temperature values. Noise filtering is a data processing process to eliminate random disturbances and systematic errors in temperature measurement, mainly using algorithms such as median filtering, wavelet transform, and Kalman filtering. The median filtering algorithm performs a time-domain window processing on the temperature data of each sampling point, sorts the temperature values within each window by size and takes the median value, effectively removing impulse noise. The wavelet transform algorithm decomposes the temperature signal into different frequency components, removes the high-frequency noise components through threshold processing and then reconstructs the signal. The Kalman filtering algorithm is based on the statistical characteristics of temperature changes and realizes the optimal estimation of temperature data through a prediction-correction mechanism. After the filtering process, the obtained calibrated contact temperature values have higher signal-to-noise ratio and reliability.

[0030] Perform pixel temperature mapping conversion on the workpiece surface temperature distribution heat map to form a non-contact temperature matrix. Pixel temperature mapping conversion refers to the process of converting the gray value or color value of each pixel point in the thermal imaging image into the corresponding temperature value. Calibrate the thermal imager with a blackbody to establish the correspondence between radiation intensity and temperature, and then map the image pixel values to the temperature space through linear or non-linear interpolation algorithms. Considering the emissivity change of the welded workpiece, emissivity correction is also required, and the temperature values are corrected according to the emissivity parameters of different materials and surface states. The formed non-contact temperature matrix is a two-dimensional array, and each element represents the temperature value at the corresponding spatial position. Perform data fusion on the calibrated contact temperature values and the non-contact temperature matrix to eliminate the emissivity error and obtain a corrected temperature field. Data fusion uses algorithms such as weighted average method, Bayesian estimation method, or Kalman filtering method to complement the advantages of the two temperature measurement methods. The weighted average method assigns weight coefficients according to the reliability of the two measurement methods and calculates the weighted average temperature value. The Bayesian estimation method takes the contact measurement result as prior information, combines the non-contact measurement result to update the posterior probability distribution, and obtains the temperature estimation with the maximum posterior probability. The Kalman filtering method constructs a state space model, takes the two measurement data as observation values, and iteratively calculates the optimal temperature estimation through prediction and update steps. Through data fusion, the limitations of point measurement of K-type thermocouples and the influence of emissivity error of infrared thermal imagers are overcome, and a corrected temperature field with high spatial resolution and high measurement accuracy is obtained.

[0031] The corrected temperature field is sampled and transmitted in packets at a sampling frequency of 10 Hz via an industrial fieldbus, forming real-time temperature data. An industrial fieldbus refers to a real-time communication network used in the field of industrial automation, such as Profibus, DeviceNet, or EtherCAT, etc., which has high reliability, deterministic latency, and anti-interference capabilities. During the data packet transmission process, the corrected temperature field data is indexed according to timestamps and spatial coordinates to construct the packet header; then the temperature values are stored in the packet body in the order of grid positions, and a checksum is added to ensure data integrity; the packet is encapsulated according to the requirements of the bus protocol and sent to the data processing unit at a frequency of 10 Hz. Real-time temperature data refers to the temperature information flow that can reflect the current thermal state of the welding process, with a delay time not exceeding 100 milliseconds.

[0032] Taking the welding process of a steel structure as an example, a K-type thermocouple array with a 4×4 grid is arranged around the welding area, with a spacing of 10 mm, covering a monitoring area of 40 mm×40 mm. At the same time, an uncooled infrared thermal imager with a resolution of 384×288 pixels takes pictures of the welding area from a distance of 500 mm, and the spatial resolution is about 0.5 mm / pixel. When the welding arc is started, the K-type thermocouples collect temperature data at 16 points in real time, and the temperature fluctuations caused by electromagnetic interference are eliminated through median filtering. The thermal image collected by the infrared thermal imager is corrected for emissivity (the emissivity of steel is set to 0.7) and converted into a temperature matrix of 110,592 points. Through the Bayesian estimation method, with the thermocouple measurement values as the reference points, interpolation calculations are performed in combination with the thermal imaging data to construct a high-precision corrected temperature field. The corrected temperature field data is transmitted to the control unit via the EtherCAT bus at a frequency of 10 Hz, and each data packet contains temperature distribution information and timestamps for thermal field analysis and early warning judgment.

[0033] In a specific embodiment, the process of performing step S102 may specifically include the following steps: (1) Convert the real-time temperature data into temperature spectral domain data to obtain the frequency characteristics of the temperature spatial distribution; (2) Expand the heat flux density of the temperature spectral domain data through the forward Fourier transform to obtain the basic heat conduction equation; (3) Construct a thermal diffusion difference format according to the basic heat conduction equation to form a heat flux propagation vector field; (4) Perform grid subdivision processing on the heat flux propagation vector field to establish a discrete calculation grid with an accuracy of 0.5 mm; (5) Iteratively solve through the heat flux boundary condition constraints of the discrete calculation grid to obtain the steady-state and transient temperature distributions; (6) Visualize the steady-state and transient temperature distributions through pseudo-color mapping technology to generate a hot spot distribution grid map.

[0034] Specifically, the real-time temperature data is the spatial temperature distribution information obtained by a K-type thermocouple and a non-cooled infrared thermal imager at a sampling frequency of 10 Hz. The temperature spectrum domain conversion uses the two-dimensional discrete Fourier transform (2D-DFT) method to convert the temperature data in the spatial domain to the frequency domain. For a two-dimensional temperature field T(x, y), where x and y are spatial coordinates, the following calculation is performed through the fast Fourier transform (FFT) algorithm to obtain the frequency domain representation F(u, v), where u and v are spatial frequency variables. The spectrum domain data reflects the distribution characteristics of the temperature field at different spatial frequencies. The low-frequency components represent the overall trend of the temperature field, and the high-frequency components correspond to the local details and boundary information of the temperature field. This conversion enables subsequent heat conduction analysis to be carried out in the frequency domain, simplifying the computational complexity.

[0035] The heat flux density expansion is performed on the temperature spectrum domain data through the forward Fourier transform to obtain the basic heat conduction equation. The forward Fourier transform is a mathematical method for converting a signal in the spatial domain to a signal in the frequency domain. The heat flux density expansion is a process of correlating the temperature field with the heat flux density based on Fourier heat conduction theory. For the temperature spectrum domain data during the welding process, the heat flux density expansion is performed through the following formula:

[0036] where, represents the heat flux density distribution function in the frequency domain; represents the nth-order thermal conductivity coefficient, with the unit of W / (m·K); represents the gradient of the nth-order temperature mode; , , respectively represent the dimensionless x, y, and z spatial coordinates; represents the dimensionless time variable; represents the nth-order attenuation coefficient, which is related to the thermal physical properties of the material. This formula establishes the correlation between the spectral components of the temperature field and the corresponding heat flux density. Through the inverse transform, the basic heat conduction equation can be obtained:

[0037] where, represents the mass heat capacity of the material; represents the temperature field function; represents the generalized spatial coordinate; represents the directional thermal conductivity; Represents the internal heat source function. This equation describes the basic law of heat transfer during welding. According to the basic equation of heat conduction, the heat diffusion difference format is constructed to form a heat flux propagation vector field. The heat diffusion difference format is the process of discretizing continuous partial differential equations into numerical equations suitable for computer solving. Using the finite difference method, the derivatives in space and time are replaced by the difference quotient of the function value of discrete points. For spatial derivatives, the central difference format is used; for time derivatives, the forward difference format is used. Taking three-dimensional space as an example, for the grid point (i, j, k), its discretized equation is expressed as follows, thus forming a vector field describing the direction and intensity of heat propagation.

[0038] The heat flux propagation vector field is meshed and processed to establish a discrete computational grid with an accuracy of 0.5 mm. Mesh subdivision is an important means to improve computational accuracy, and adaptive mesh refinement technology is used. The mesh density is determined according to the size of the temperature gradient. A minimum mesh size of 0.5 mm is used in areas with drastic temperature changes (such as near the welding pool), and a larger mesh size is used in areas with gentle temperature changes. In the specific implementation, the temperature gradient of each mesh unit is calculated, and when the gradient exceeds the preset threshold, the unit is divided into four (two-dimensional) or eight (three-dimensional) sub-units. The mesh subdivision process continues until the target accuracy of 0.5 mm is reached or other termination conditions are met. The refined mesh structure is stored using a quadtree (two-dimensional) or octree (three-dimensional) data structure to efficiently manage mesh units of different sizes.

[0039] The discrete computational grid is iteratively solved through the heat flow boundary condition constraints to obtain the steady-state and transient temperature distribution. The heat flow boundary condition constraints include the welding heat source boundary, the heat dissipation boundary of the workpiece outer surface, etc. The iterative solution adopts implicit or explicit numerical methods. For the welding high temperature warning system, the following implicit iteration format is adopted:

[0040] in, represents the temperature value of the pth grid point in the s+1th iteration; Represents the heat transfer coefficient matrix, describing the heat transfer relationship between grid points p and q; represents the coefficient of inertia, which is related to the heat capacity of the material and the time step; represents the temperature value of the qth grid point in the sth iteration; represents the heat source influence coefficient matrix; Represents the heat input of the rth heat source point in the sth iteration. Through repeated iterative calculations, when the temperature field difference between two adjacent iterations is less than the preset threshold, the calculation is considered to have converged and the steady-state temperature distribution is obtained. If time changes are considered, the temperature field at each moment is calculated by gradually advancing the time step to obtain the transient temperature distribution.

[0041] The steady-state and transient temperature distributions are visualized using pseudo-color mapping technology to generate a hotspot distribution grid map. Pseudo-color mapping is a visualization technology that maps scalar data to colors. A rainbow color spectrum (from blue to red) is usually used to represent temperature changes from low to high. In specific implementation, the minimum and maximum values ​​of the temperature are determined, and a linear or nonlinear mapping relationship from temperature value to color value is established. The temperature value of each grid point is then converted to the corresponding RGB color value to generate a color image of the entire temperature field. The hotspot distribution grid map intuitively displays the temperature distribution during the welding process. The high-temperature area is represented by red or white, and the low-temperature area is represented by blue or green, which is convenient for quickly identifying temperature abnormalities.

[0042] For example, the real-time temperature data of the welding area is obtained by using a K-type thermocouple array and an infrared thermal imager to form a 100×100 spatial temperature matrix. The temperature matrix is ​​converted into a frequency domain representation through a two-dimensional FFT algorithm to obtain spectral domain data. The low-frequency component indicates that the welding area as a whole presents a bell-shaped temperature distribution, with the highest temperature in the center; the high-frequency component reflects the temperature mutation characteristics at the edge of the weld. Based on the spectral domain data, the heat flux density distribution is calculated through the aforementioned heat flux density expansion formula, and the heat conduction equation is constructed. Considering the mobile characteristics of the arc heat source during welding, the mobile heat source term is introduced into the equation. The equation is discretized using the implicit finite difference method, and the initial grid size is set to 2mm. According to the calculated temperature gradient, the grid near the molten pool is refined, and the minimum grid size reaches 0.5mm, accurately capturing the temperature change at the boundary of the molten pool. Appropriate boundary conditions (such as convection heat dissipation and radiation heat dissipation on the workpiece surface) are set, and the temperature distribution during welding is obtained through iterative calculation. The results show that the temperature in the center of the arc can reach more than 2000°C, the temperature at the edge of the molten pool is about the melting point of the material (about 1500°C), and the temperature of the heat-affected zone of the welding is between 800-1200°C. These temperature data are converted into a hot spot distribution grid map through pseudo-color mapping, in which the molten pool area is displayed in red and white, the heat-affected zone is displayed in yellow and orange, and the area away from the weld is green and blue, which intuitively reflects the temperature distribution characteristics during the welding process.

[0043] In a specific embodiment, the process of executing step S103 may specifically include the following steps: (1) Calculate the temperature gradient of the hotspot distribution grid map and generate a temperature change rate distribution map; (2) Binarize the temperature change rate distribution map according to the material safety threshold to obtain the temperature anomaly candidate area; (3) The temperature anomaly candidate areas are marked with connected domains using an improved region growing algorithm to form the contour boundary of the hot zone; (4) Extract the center coordinates, area and perimeter features from the hot zone contour boundary and construct the hot zone spatial feature table; (5) Perform temporal comparison on the same hot zone in a series of consecutive hot spot distribution grid maps, calculate the temperature change rate of each hot zone, and obtain the hot zone temperature rise rate data; (6) Combine and process the hot zone spatial feature table and the hot zone temperature rise rate data to generate an abnormal hot zone feature set.

[0044] Specifically, calculating the temperature difference gradient of the hot spot distribution grid map to generate a temperature change rate distribution map is the primary step in identifying abnormal hot zones. The hot spot distribution grid map is a visualization result of the temperature field generated by the pseudo-color mapping technology, containing matrix data of spatial positions and corresponding temperature values. The temperature difference gradient calculation uses the Sobel operator or the Laplacian operator to perform spatial differential operations on the temperature field. The Sobel operator approximately obtains the temperature gradient magnitude and direction by calculating the temperature differences in the horizontal and vertical directions. When calculating the temperature gradient, for each grid point, calculate the temperature difference between it and the adjacent point, and then divide by the distance between the two points to obtain the temperature change rate in that direction. Calculate the temperature change rates in both the horizontal and vertical directions, and then obtain the total temperature gradient value through vector synthesis. The Laplacian operator directly calculates the second derivative of the temperature field to detect the local change rate of the temperature. The calculated temperature gradient values form a temperature change rate distribution map, which reflects the severity of the spatial change of the surface temperature of the welded workpiece. The larger the gradient value, the faster the temperature change, and the more likely there are potential thermal anomalies. Perform binarization processing on the temperature change rate distribution map according to the material safety threshold. The material safety threshold refers to the maximum temperature gradient or change rate that the material can withstand during the welding process. Exceeding this threshold may cause excessive internal stress, deformation, or cracking of the material. The material safety threshold is determined according to the physical properties of different materials. For example, for steel, it is usually 30 - 50 °C / mm, and for aluminum alloy, it is 15 - 25 °C / mm. The binarization process is to compare each pixel point in the temperature change rate distribution map with a preset threshold. Points greater than the threshold are marked as 1 (abnormal), and points less than the threshold are marked as 0 (normal). The binarization process uses the global threshold method or the local adaptive threshold method. The global threshold method uses a fixed threshold to process the entire image. If the temperature gradient value is greater than the threshold, it is marked as an abnormal point; otherwise, it is marked as a normal point. The local adaptive threshold method dynamically adjusts the threshold according to the statistical characteristics of the local area around the pixel point, and is suitable for situations with uneven illumination or uneven temperature distribution. After binarization processing, the temperature anomaly candidate areas are clearly identified, laying a foundation for regional analysis.

[0045] The connected component labeling of the temperature anomaly candidate areas is performed by an improved region growing algorithm to form the contour boundary of the hot zone. The region growing algorithm is a segmentation method that starts from a seed point and gradually merges adjacent pixels that meet the similarity conditions into the region. The improved region growing algorithm adds direction constraints and temperature gradient information on the basis of the traditional algorithm, which is suitable for the characteristics of the welding hot zone. The specific steps include: selecting the points with a value of 1 in the binary image as the initial seed points; then, checking the 8-neighborhood or 4-neighborhood pixels of the seed points. If the value of the neighborhood pixel is 1 and it has not been labeled, add it to the current region and label it; then, take the newly added pixel as the new seed point and repeat the above process until no further expansion is possible; uniquely identify all connected regions to form different connected components. To improve the algorithm efficiency, breadth-first search or stack structure is used to implement the region growing. After the connected component labeling, the edge pixels of each connected component are extracted to form the contour boundary of the hot zone. Edge extraction uses the boundary tracking algorithm or the contour detection algorithm to obtain a closed boundary curve. The center coordinates, area, and perimeter features are extracted from the contour boundary of the hot zone to construct the spatial feature table of the hot zone. The center coordinates are calculated using the centroid method or the center method of the bounding box. When calculating the centroid, the sum of the coordinate values of all points in the region is divided by the number of points to obtain the center coordinates of the region. The area calculation method is to count the number of pixel points in the connected component and multiply it by the actual area unit (square millimeter) represented by each pixel point. The perimeter calculation method is to count the number of boundary pixels or calculate the boundary length using the chain code representation method. In addition to the basic geometric features, shape descriptors such as shape factor, circularity, and rectangularity are also extracted to distinguish different types of thermal anomaly regions. The shape factor is defined as the ratio of the square of the perimeter to the area, which is used to describe the regularity of the region. Circularity refers to the similarity degree of the region to a circle with the same area. Rectangularity is the ratio of the area of the region to its minimum bounding rectangle. The characteristic values of each hot zone are organized into a structured data table, including the hot zone ID, center coordinates, area, perimeter, and other shape features, to form the spatial feature table of the hot zone. This table provides the basis for spatial features for subsequent time series analysis and risk assessment.

[0046] Perform temporal comparison on the same hot zone in a series of consecutive hot spot distribution grid maps, calculate the temperature change rate of each hot zone, and obtain the hot zone temperature rise rate data. Temporal comparison needs to solve the inter-frame matching problem of hot zones, that is, determine which hot zones in different time frames correspond to the same physical area. Inter-frame matching adopts a matching algorithm based on spatial position and shape features, usually implemented using the Hungarian algorithm or the greedy matching algorithm. When matching, factors such as the Euclidean distance of the center position of the hot zone, the ratio of area change, and the shape similarity are considered, and the hot zone pair with the highest comprehensive score is considered to be the performance of the same hot zone in different time frames. For the successfully matched hot zone pairs, extract their average temperature or maximum temperature, and calculate the rate of change of temperature over time. The calculation method of the temperature rise rate is the temperature difference between two frames divided by the time interval, with the unit of °C / second. To reduce the influence of random fluctuations, a multi-frame sliding window average or linear regression method can be used to calculate a more stable temperature rise rate. For each hot zone, record its temperature rise rate and historical change trend to form the hot zone temperature rise rate data.

[0047] Merge and process the hot zone spatial feature table and the hot zone temperature rise rate data to generate an abnormal hot zone feature set. The merge process is a process of correlating and integrating spatial and temporal features. The specific method is as follows: establish an association table indexed by the hot zone ID; then, merge the spatial features (center coordinates, area, perimeter, etc.) and temporal features (average temperature, maximum temperature, temperature rise rate, etc.) of each hot zone into the same data structure; add additional derived features, such as the ratio of temperature to area, the ratio of temperature rise rate to perimeter, etc., to enhance the description ability of the features. The abnormal hot zone feature set is a multi-dimensional data set, with each hot zone corresponding to a record, and the record contains spatial-temporal features. This feature set provides comprehensive data support for risk assessment and warning level judgment.

[0048] For example, temperature data during the welding process is collected by a K-type thermocouple and an infrared thermal imager, and a hot spot distribution grid map is obtained through Fourier heat conduction calculation. The temperature difference gradient is calculated for the hot spot distribution grid map, and it is found that there is an obvious temperature gradient near the weld, especially at the junction of the weld and the base metal, where the temperature gradient value is as high as 80 °C / mm. According to the safety threshold of the stainless steel material (about 40 °C / mm), the temperature change rate distribution map is binarized, and the area greater than 40 °C / mm is marked as the temperature anomaly candidate area. The improved region growing algorithm is applied to the anomaly candidate area to identify three main connected domains, which are located near the starting point, the middle section, and the ending point of the weld respectively. The spatial features of these three hot areas are extracted, including the center coordinates, area, and perimeter: the area of the hot area at the starting point of the weld is about 50 square millimeters, the center coordinates are (25, 30), and the perimeter is about 28 millimeters; the area of the hot area in the middle section is 80 square millimeters, the center coordinates are (150, 30), and the perimeter is 36 millimeters; the area of the hot area at the ending point is 60 square millimeters, the center coordinates are (275, 30), and the perimeter is 32 millimeters. Through the temporal comparison of 10 consecutive frames (equivalent to 1 second because the sampling frequency is 10 Hz) of the hot spot distribution grid map, the temperature rise rates of the three hot areas are calculated: the temperature rise rate of the hot area at the starting point is 15 °C / second, the temperature rise rate of the hot area in the middle section is 8 °C / second, and the temperature rise rate of the hot area at the ending point is 20 °C / second. The spatial features and temperature rise rate data are combined to form an abnormal hot area feature set. Based on this feature set, the welding control system can accurately identify the potential risks existing in the hot area at the ending point because although the area of this area is not the largest, the temperature rise rate is significantly higher than other areas, which may lead to rapid thermal stress accumulation and the formation of welding defects.

[0049] In a specific embodiment, the process of executing step S104 may specifically include the following steps: (1) Query the melting point, flash point, and safe working temperature corresponding to the welded workpiece from the material database to construct a material thermal property threshold table; (2) Calculate the ratio of the temperature values in the abnormal hot area feature set to the material thermal property threshold table to obtain the temperature hazard index; (3) Divide the temperature rise rate in the abnormal hot area feature set by the material safe temperature rise rate to obtain the temperature rise hazard coefficient; (4) Calculate the hot area diffusion index according to the proportional relationship between the hot area area in the abnormal hot area feature set and the standard area of the welding point; (5) Perform weighted summation on the temperature hazard index, temperature rise hazard coefficient, and hot area diffusion index to generate a comprehensive risk score; (6) Map the comprehensive risk score into four levels of low, medium, high, and emergency through multi-level threshold segmentation to form a warning level matrix.

[0050] Specifically, the material database is a structured data set containing the thermophysical property parameters of various metal and non-metal materials, and the storage format is usually a relational database or a JSON format file. The melting point refers to the temperature at which the material changes from a solid state to a liquid state, the flash point refers to the lowest temperature required to ignite the material, and the safe operating temperature refers to the highest temperature at which the material can operate for a long time without significant performance degradation. The query process uses the material code or name as an index, and extracts relevant parameters from the database through SQL query statements or key-value pair matching. The extracted parameters are organized into a material thermal property threshold table, where each material in the table corresponds to a record, and the record contains fields such as material name, material code, melting point, flash point, and safe operating temperature. For composite materials or cases involving multiple materials in welding, the parameter with the lowest critical value is selected as the threshold to ensure a safety margin. In addition, the threshold table also contains the safe heating rate of the material, that is, the maximum temperature rise rate allowed for the material without thermal stress damage, usually in units of °C / second. The temperature values in the abnormal heat zone feature set are used to calculate the ratio with the material thermal property threshold table to obtain the temperature hazard index. The abnormal heat zone feature set is a data set generated through previous steps, containing the position coordinates, area, perimeter, and temperature characteristics of each heat zone. The temperature value refers to the highest temperature or average temperature of the heat zone, and the specific selection depends on the risk assessment strategy. The ratio calculation method is as follows: divide the heat zone temperature by the safe operating temperature of the material to obtain a preliminary ratio; then convert this ratio into a standardized temperature hazard index through a non-linear mapping function (such as an exponential function or a piecewise linear function), and the value range is usually from 0 to 10. Non-linear mapping can more accurately reflect the accelerating growth characteristics of risk when the temperature approaches the critical threshold. For example, when the temperature is close to but lower than the safe operating temperature, the hazard index may be between 1 and 3; when the temperature exceeds the safe operating temperature but is lower than the melting point, the hazard index may be between 4 and 7; when the temperature is close to or exceeds the melting point, the hazard index may be between 8 and 10. The temperature hazard index directly reflects the threat degree of the heat zone temperature to the material safety.

[0051] The temperature rise risk coefficient is obtained by dividing the temperature rise rate in the abnormal hot zone feature set by the material's safe temperature rise rate. The temperature rise rate refers to the speed at which the hot zone temperature changes over time and is calculated through the sequential comparison of consecutive multi-frame hot spot distribution grid maps. The material's safe temperature rise rate refers to the maximum temperature change rate that the material can withstand. Exceeding this rate may cause internal stress concentration, deformation, or cracks in the material. The method for calculating the temperature rise risk coefficient is to divide the actual temperature rise rate of the hot zone by the material's safe temperature rise rate, resulting in a dimensionless ratio. When the ratio is less than 1, it indicates that the temperature rise rate is within the safe range; when the ratio is greater than 1, it indicates that the temperature rise rate exceeds the safety limit, and there is a risk of thermal stress damage. To standardize the temperature rise risk coefficients of different hot zones, a piecewise mapping function is used to convert the original ratio into a standard score between 0 and 10. The temperature rise risk coefficient reflects the speed of thermal stress accumulation and is an important indicator for evaluating the dynamic risk of the welding process. According to the proportional relationship between the hot zone area and the standard area of the welding point in the abnormal hot zone feature set, the hot zone diffusion index is calculated. The standard area of the welding point refers to the ideal area of the welding molten pool or heat-affected zone expected to be formed under normal welding conditions, usually determined by process specifications. The hot zone area refers to the area of the abnormal hot zone formed during the actual welding process and is obtained through connected component analysis. The method for calculating the hot zone diffusion index is as follows: divide the actual hot zone area by the standard area of the welding point to obtain an area ratio; then, according to the ratio value, convert it into a standardized hot zone diffusion index through table lookup or function mapping, with a value range of 0 to 10. The larger the area ratio, the more the heat-affected range exceeds the expectation, and the higher the hot zone diffusion index. The hot zone diffusion index reflects the degree of heat diffusion in the material. An excessively high diffusion index indicates poor control of welding heat, which may affect the welding quality or result in an overly large heat-affected zone.

[0052] The temperature risk index, temperature rise risk coefficient, and heat zone diffusion index are weighted and summed to generate a comprehensive risk score. Weighted summation is a multi-index comprehensive evaluation method that reflects the contribution degree of each factor to the overall risk by assigning different weights to each index. The weight assignment is based on the experience of welding experts and the results of statistical analysis. Usually, the weight of the temperature risk index is the highest because temperature is the most direct risk indicator. The specific calculation method is as follows: multiply the temperature risk index by its weight (such as 0.5), multiply the temperature rise risk coefficient by its weight (such as 0.3), multiply the heat zone diffusion index by its weight (such as 0.2), and then sum the three to obtain the comprehensive risk score. The value range of the comprehensive risk score is from 0 to 10, and the higher the score, the greater the risk. The weight settings can be dynamically adjusted according to different welding materials and process characteristics to adapt to different application scenarios. The comprehensive risk score provides a single numerical indicator for quickly judging the safety status of the welding process. The comprehensive risk score is mapped into four levels: low, medium, high, and emergency through multi-level threshold segmentation to form a warning level matrix. Multi-level threshold segmentation is a method of dividing continuous score values into discrete levels. Usually, the comprehensive risk score is mapped into a low-level warning when it is between 0 and 3, a medium-level warning when it is between 3 and 5, a high-level warning when it is between 5 and 7, and an emergency warning when it is between 7 and 10. The warning level matrix is a two-dimensional data structure. The rows represent the spatial division of the welding area, and the columns represent different time points. The matrix element values are the warning levels corresponding to the spatio-temporal positions. The warning level matrix is visually displayed through color coding. For example, green represents a low-level warning, yellow represents a medium-level warning, orange represents a high-level warning, and red represents an emergency warning. This multi-level classification mechanism not only avoids resource waste caused by over-warning but also ensures that real dangerous situations can be responded to in a timely manner.

[0053] For example, query the key parameters of this superalloy from the material database: the melting point is 1380 °C, the flash point is not applicable (for metal materials), the safe operating temperature is 950 °C, and the safe heating rate is 45 °C per second. After constructing the material thermal characteristic threshold table, a risk assessment is performed on the identified abnormal thermal areas. Suppose the highest temperature of a certain thermal area is 1050 °C. Calculate its ratio with the safe operating temperature of 950 °C to obtain an initial ratio of 1.11, which is converted into a temperature hazard index of 5.8 through a non-linear mapping function. At the same time, the temperature rise rate of this thermal area is 60 °C per second, and the ratio with the safe heating rate of 45 °C per second is 1.33, which is mapped to a temperature rise hazard coefficient of 6.5. The area of this thermal area is 180 square millimeters, while the standard area of the welding point is 120 square millimeters, and the area ratio is 1.5, corresponding to a thermal area diffusion index of 7.2. According to the preset weights (temperature 0.5, temperature rise 0.3, diffusion 0.2), calculate the comprehensive risk score: 5.8×0.5 + 6.5×0.3 + 7.2×0.2 = 6.29. This score is between 5 and 7, corresponding to the high-level warning level. This indicates that there are obvious risks in this area during the welding process and the welding parameters need to be adjusted in a timely manner. At this time, the warning system marks this area as orange in the warning level matrix and triggers the corresponding alarm mechanism, such as emitting a high-frequency sound alarm and sending a parameter adjustment instruction to the welding power controller to reduce the welding current and voltage by 10% respectively.

[0054] In a specific embodiment, the process of executing step S105 may specifically include the following steps: (1) Map the low-level warning in the warning level matrix to a green flashing signal code to generate a visual warning data packet; (2) Convert the medium-level warning in the warning level matrix into a medium-frequency beeping pulse sequence to form an audio warning signal; (3) Construct a tactile feedback instruction with a vibration intensity parameter according to the high-level warning in the warning level matrix to obtain a tactile alarm code; (4) Compile an emergency response instruction containing the device ID and fault type for the emergency warning in the warning level matrix to generate a system control command; (5) Package the visual warning data packet, audio warning signal, tactile alarm code, and system control command according to the priority to construct a multi-channel instruction data stream; (6) Send the multi-channel instruction data stream to each terminal device in a multicast manner through the industrial Ethernet to output a multi-channel alarm instruction with a frequency code.

[0055] Specifically, mapping the low-level warnings in the warning level matrix to green flashing signal codes and generating visual warning data packets is the first-level response mechanism of the alarm system. The warning level matrix is a two-dimensional data structure generated during the risk assessment phase, which records the risk levels at different positions in the welding area. Low-level warnings correspond to areas with a comprehensive risk score between 0 and 3, indicating minor anomalies but no safety threats. The green flashing signal code is a visual cue method, which is displayed using LED lights. The flashing signal code includes three parameters: flashing frequency, duration, and brightness. Low-level warnings are usually set to a green light signal with a frequency of 0.5 Hz (flashing once every 2 seconds) and a brightness of 50%. The visual warning data packet is a structured data that contains information such as warning type identifier, RGB color value (#00FF00 for green), flashing frequency value, duration value, and display area coordinates. It is packed in binary format for easy transmission in the control network. After being generated by the microprocessor, the visual warning data packet is sent to the display unit to achieve visual cues for low-level risks. Converting the medium-level warnings in the warning level matrix into medium-frequency beeping pulse sequences to form an audio warning signal. Medium-level warnings correspond to areas with a comprehensive risk score between 3 and 5, indicating abnormal conditions that need attention. Medium-frequency beeping refers to a sound signal with a frequency between 2 kHz and 4 kHz. The human ear is more sensitive in this frequency range and is easily perceivable. The pulse sequence refers to the on-off pattern of the sound. Medium-level warnings use an intermittent beeping mode, such as beeping for 1 second and then pausing for 2 seconds, repeating in a cycle. The generation process of the audio warning signal includes: determining the sound frequency (usually 3 kHz); then setting the volume level (usually 60-70 decibels, loud enough to attract attention but not overly disturbing); arranging the timing pattern of the sound (on duration, interval time, and number of repetitions). These parameters are encoded into an audio warning signal data packet, which contains fields such as frequency value, volume value, pulse pattern description, and priority flag. The audio warning signal is played through a speaker or buzzer to remind the operator to pay attention to potential problems during the welding process.

[0056] Construct a haptic feedback instruction with vibration intensity parameters based on the high-level warning in the warning level matrix to obtain a haptic alarm code. The high-level warning corresponds to the area where the comprehensive risk score is between 5 and 7, indicating a relatively serious abnormal situation that requires timely intervention. Haptic feedback is a way to transmit information through mechanical vibration and is often used for alarm transmission in noisy environments. The vibration intensity parameter refers to the operating intensity of the vibration motor, usually expressed as a percentage, and the high-level warning is set to a vibration intensity of 70%-80%. The process of constructing the haptic feedback instruction includes: setting the vibration mode (such as short-short-long or crescendo mode); determining the vibration intensity parameter; specifying the duration and interval of the vibration; adding the target ID of the haptic device. These pieces of information are combined to form a haptic alarm code, which is a binary coding format and includes fields such as device type field, vibration mode field, intensity field, and time parameter field. The haptic alarm code is sent to the vibration device worn by the operator or the seat vibration unit through the control bus, enabling the operator to perceive the welding danger without relying on vision and hearing. Compile an emergency response instruction containing the device ID and fault type for the emergency warning in the warning level matrix to generate a system control command. The emergency warning corresponds to the area where the comprehensive risk score is between 7 and 10, indicating an emergency danger that requires immediate protective measures. The device ID is the unique identifier of each device in the welding system, such as the power controller ID, feed system ID, etc. The fault type is the classification of abnormal situations, such as too high temperature, too fast heat zone diffusion, etc. The process of compiling the emergency response instruction includes: looking up the ID and communication protocol of the relevant device in the device registry; selecting the corresponding fault code according to the warning type; adding a timestamp and a priority mark; calculating the instruction checksum to ensure data integrity. The system control command is a directly executable instruction format and includes fields such as command type (such as stop, reduce power, emergency cooling, etc.), target device ID, parameter value, and execution priority. When generating a system control command at the emergency level, the safety protection mechanism of the device is usually automatically triggered, such as actively reducing the welding current, starting the auxiliary cooling system, or completely interrupting the welding process.

[0057] Pack the visual warning data packet, audio warning signal, tactile alarm code, and system control command according to priority to construct a multi-channel instruction data stream. Priority packing is a data processing method that organizes multiple different types of signals according to their importance. The priorities from high to low are: system control command (highest priority), tactile alarm code, audio warning signal, and visual warning data packet (lowest priority). The packing process uses a hierarchical encapsulation technique to construct a unified data header that contains information such as packet identification, total length, priority, source address, and destination address; then add various instruction data in priority order; and append check information. The multi-channel instruction data stream is a composite data stream that allows multiple alarm signals to be transmitted simultaneously to meet the requirements of different sensing channels. The data stream adopts a structured format, such as XML or JSON, for easy cross-platform parsing. The multi-channel design ensures that information can be conveyed to the operator through multiple sensing methods in a complex working environment, improving the reliability and effectiveness of the alarm. Send the multi-channel instruction data stream to each terminal device in a multicast manner through industrial Ethernet, and output a frequency-encoded multi-channel alarm instruction. Industrial Ethernet is a communication network designed for industrial automation environments, such as PROFINET, EtherNet / IP, or EtherCAT, with characteristics of deterministic delay, high reliability, and real-time performance. The multicast method is a network transmission technology that allows a single sender to transmit the same data to multiple receivers simultaneously, improving network utilization. The process of transmitting the multi-channel instruction data stream through industrial Ethernet includes: packet segmentation (if it exceeds the maximum transmission unit size); adding an Ethernet header and trailer; calculating the cyclic redundancy check code (CRC) to ensure data integrity; setting the multicast address so that all devices subscribed to this address can receive the information. Frequency encoding is a signal processing technique that associates different frequencies or frequency combinations with specific meanings to improve anti-interference ability. In the welding high-temperature warning system, frequency encoding is used to distinguish different types and levels of alarms, such as a low-frequency range (0.1 - 1 Hz) corresponding to a low-level warning and a high-frequency range (5 - 10 Hz) corresponding to an emergency warning. Each terminal device executes corresponding display, sound, vibration, or control actions according to the received frequency-encoded signal to achieve multi-channel and multi-level alarm responses.

[0058] For example, when the welding monitoring system detects abnormal high-temperature diffusion in the edge area of the welding molten pool, the risk assessment module calculates a comprehensive risk score of 4.8 based on the temperature, temperature rise rate, and heat zone area, which belongs to a medium-level warning. The warning system then extracts the warning information for this area from the warning level matrix and generates corresponding multi-channel alarm instructions. The system generates a visual warning data packet, sets a yellow flashing signal with a frequency of 1 Hz and a brightness of 70%, and designates to display a warning sign in the welding molten pool monitoring window of the operation interface. At the same time, the audio processing unit generates a medium-frequency beeping pulse sequence, sets the sound frequency to 3.5 kHz and the volume to 65 decibels, and adopts an intermittent mode of sounding for 0.5 seconds and pausing for 1 second. These alarm data, together with equipment control instructions (such as the command to reduce the welding current by 5%), are packaged into a multi-channel instruction data stream according to the priority. The system sends this data stream to terminal devices such as the console display, workshop alarm, and welding power controller through the industrial Ethernet in a multicast manner. The operator receives the alarm information visually and auditorily at the same time, quickly identifies the problem area, and observes that the system has automatically adjusted the welding parameters to control the heat input. This multi-channel and differentiated alarm mechanism significantly improves the effectiveness of the warning system, enabling the operator to intervene in a timely manner before the problem expands and avoiding the formation of welding defects and thermal damage to the workpiece.

[0059] In a specific embodiment, the process of executing step S106 may specifically include the following steps: (1) Parse the risk level identifier and hot spot location information from the multi-channel alarm instructions to generate a welding danger status table; (2) Query the process parameter library according to the welding danger status table to obtain the corresponding current-voltage adjustment strategy; (3) Convert the current-voltage adjustment strategy into a digital signal control quantity to form a power control data stream; (4) Perform dynamic compensation on the power control data stream according to the welding material characteristic curve to obtain a smoothly transitioning current-voltage adjustment curve; (5) Convert the current-voltage adjustment curve into an analog control signal through a digital-to-analog converter and output it to the welding power module; (6) The welding power module performs real-time dynamic adjustment of the welding parameters according to the analog control signal to complete the closed-loop adaptive control of the welding parameters.

[0060] Specifically, the multi-channel alarm instruction is a composite data stream, which contains multiple information such as visual warnings, audio warnings, tactile feedback and system control commands. The parsing process performs header recognition on the received data stream to confirm the data type and format; then extracts the risk level identifier, which represents the four risk levels of low, medium, high and emergency in the form of digital code; then extracts the hotspot location information, including the spatial coordinates, range size and temperature value of the hotspot. The parsing adopts data unpacking technology to read and convert field by field according to the predefined data structure into a standard format. The risk level identifier and hotspot location information are combined to form a welding hazard state table, which is a structured data set. Each record contains fields such as timestamp, area identifier, risk level, main hazard parameters (such as excessive temperature, rapid temperature rise, etc.) and location coordinates. The welding hazard state table provides accurate hazard area identification and classification information for subsequent parameter adjustment. According to the welding hazard state table, the process parameter library is queried to obtain the corresponding current-voltage adjustment strategy. The process parameter library is a database containing the optimal process parameters corresponding to various welding materials, thicknesses and joint types, and stores a large amount of empirical data and test results. The query process uses a multi-condition matching algorithm to locate the basic parameter range according to the material type, thickness and joint form of the current welding workpiece; then match the corresponding parameter adjustment strategy according to the risk level and hazard type in the welding hazard status table. The current-voltage adjustment strategy is a set of rules that defines how the welding current and voltage should be adjusted for different hazardous situations. The strategy includes elements such as adjustment direction (increase or decrease), adjustment amplitude (percentage or absolute value), adjustment rate (fast or gradual) and adjustment duration. For example, for the high-level risk of excessive hot spot temperature, the strategy may be "reduce current by 15%, reduce voltage by 5%, respond quickly, and continue until the risk drops to intermediate level." These strategies are extracted from the process parameter library in a structured data format to provide a decision basis for parameter adjustment.

[0061] Convert the current-voltage adjustment strategy into a digital signal control quantity to form a power supply control data stream. The digital signal control quantity refers to discrete numerical values that can be processed by a digital control system and is used to precisely control the output parameters of the welding power supply. The conversion process parses the text description of the adjustment strategy into numerical operation instructions; then calculates the adjusted target values based on the current welding current and voltage values; then determines the type of adjustment curve (such as linear, exponential, or S-shaped curve); generates a series of current-voltage value pairs corresponding to time points. The digital signal control quantity is usually represented by 16-bit or 32-bit integers and is encoded in percentages or actual physical units (amperes, volts). The power supply control data stream is a time series data that contains time points, current values, voltage values, and other control parameters (such as pulse frequency, duty cycle, etc.). This data stream is organized in a specific format (such as CSV or binary format) for easy subsequent processing and transmission. The power supply control data stream provides precise parameter adjustment instructions for the welding power supply and is the core data basis for realizing adaptive control. Dynamically compensate the power supply control data stream according to the welding material characteristic curve to obtain a smoothly transitioning current-voltage adjustment curve. The welding material characteristic curve is a curve that describes the resistance change law of the material under different temperature and current density conditions, and different materials have significantly different characteristic curves. Dynamic compensation is the process of correcting the original control data according to the material characteristic curve, aiming to avoid welding instability caused by parameter mutations. The compensation process includes: reading the characteristic curve data of the current welding material from the material database; then determining the current resistance characteristic of the material according to the measured real-time temperature; then calculating the actual power absorption of the material under the target current-voltage; generating corrected current-voltage values considering the material characteristic changes. Dynamic compensation also considers the time characteristics of the current-voltage changes and avoids jumps in control parameters by adding delays and smoothing factors. The smoothly transitioning current-voltage adjustment curve is an optimized control curve that takes into account the material characteristics, system response characteristics, and safety margins, and can avoid welding defects caused by parameter mutations while ensuring effective adjustment.

[0062] The current-voltage regulation curve is converted into an analog control signal through a digital-to-analog converter and output to the welding power supply power module. A digital-to-analog converter is an electronic device that converts digital signals into analog signals. In welding control systems, high-precision 16-bit or 24-bit DAC (digital-to-analog converter) chips are usually adopted. The conversion process includes: loading the digital control quantity into the converter buffer at a predetermined sampling rate; then triggering the DAC chip through an internal clock to output the corresponding analog voltage signal in sequence; then processing through a signal conditioning circuit (including filtering, amplification, and isolation, etc.) to filter out high-frequency noise and spike interference; connecting to the control end of the welding power supply power module through an analog interface. The analog control signal is usually a 0-10V voltage signal or a 4-20mA current signal, and its value has a linear or specific non-linear relationship with the output parameters of the welding power supply. This digital-to-analog conversion mechanism converts the decisions of advanced control algorithms into actual physical control signals and is a bridge between theoretical control strategies and actual hardware execution. The welding power supply power module adjusts the welding parameters in real-time and dynamically according to the analog control signal to complete the closed-loop adaptive control of the welding parameters. The welding power supply power module is the hardware unit that actually generates the welding current and voltage and is usually composed of a power conversion circuit formed by IGBT (insulated gate bipolar transistor) or MOSFET (metal oxide semiconductor field effect transistor). The real-time dynamic adjustment process includes: the control circuit of the power module receives the analog control signal; then the internal PID (proportional-integral-derivative) controller compares the control signal with the actual output parameters and calculates the adjustment amount; then by adjusting the switching frequency, duty cycle, or conduction angle of the power device, the output current and voltage are changed; the actual output parameters are detected through current and voltage sensors, and the measurement results are fed back to the control system to form a closed-loop control. Closed-loop adaptive control is an advanced control method that can automatically adjust control parameters according to the system state and external conditions and has the characteristics of real-time response, precise control, and strong anti-interference ability. In the welding high-temperature warning system, this control method can actively adjust the welding parameters according to the detected thermal anomaly to prevent overheating damage and welding defects.

[0063] For example, when the welding monitoring system detects an abnormal temperature rise in the end region of the weld seam, the multi-channel alarm system generates and sends an alarm instruction containing the risk information of this region. The control system analyzes these instructions and extracts the region coordinates (250 mm from the starting point of the weld seam), the risk level (high-level warning), and the type of danger (the temperature rise rate is too high, reaching 25 °C / second, far higher than the safe temperature rise rate of 15 °C / second). Based on this information, the system generates a welding danger status table, recording the risk region, degree, and nature. Then, the system queries the process parameter library to obtain the adjustment strategy for the case of excessive temperature rise at the end of the butt weld for high-strength steel HT700 with a thickness of 8 mm: "Reduce the welding current by 20%, reduce the voltage by 10%, use an exponential decay curve, and return to normal after 5 seconds." The control system converts this strategy into digital control quantities, calculates the target value of 144 A from the current welding current of 180 A, calculates the target value of 23.4 V from the current voltage of 26 V, and generates time-series data that smoothly transitions from the current value to the target value within 5 seconds. Considering the change in the resistance characteristics of high-strength steel at high temperatures, the system compensates the control data according to the material characteristic curve to obtain an optimized current-voltage adjustment curve. This curve is converted into an analog control signal of 0-10 V through a 24-bit high-precision DAC and output to the welding power module. After receiving the control signal, the internal control circuit of the welding power adjusts the switching parameters of the IGBT power device to gradually reduce the output current and voltage. At the same time, the current and voltage sensors continuously monitor the output parameters and feedback the measured values to the control system to form a closed-loop control to ensure the accuracy of parameter adjustment. Through this adaptive control, the system effectively suppresses the temperature rise rate at the end of the weld seam, reduces it to the safe range, and avoids quality problems such as weld cracks and hardening of the heat-affected zone caused by excessive heat input.

[0064] The above describes the high-temperature warning method for a welding device in an embodiment of the present application. Next, a high-temperature warning system for a welding device in an embodiment of the present application will be described. Please refer to Figure 2 , an embodiment of the high-temperature warning system for a welding device in an embodiment of the present application includes: An extraction module for performing multi-point temperature measurement on the surface of the welded workpiece through a K-type thermocouple temperature sensor and a non-cooled infrared thermal imager to obtain real-time temperature data with a sampling frequency of 10 Hz; A calculation module for performing Fourier heat conduction calculation on the welding thermal field according to the real-time temperature data to form a hot spot distribution grid map with a spatial resolution of 0.5 mm; An extraction module for using the hot spot distribution grid map to extract the boundary characteristics of the temperature abnormal region exceeding the preset threshold through thermal gradient difference analysis, and generating an abnormal hot zone feature set containing position coordinates and temperature rise rate; A weighted module, which is used to perform multi-parameter weighted calculation on the risk level according to the abnormal hot zone feature set combined with the material melting point parameter, and form an early warning level matrix divided into four levels: low, medium, high, and emergency; A transmission module, which is used to perform differential signal transmission on the opto-acoustic-electronic integrated alarm system according to the early warning level matrix through the industrial control bus protocol, and output multi-channel alarm instructions with frequency encoding; An adjustment module, which is used to perform real-time adjustment of the current-voltage curve of the welding power supply controller based on the multi-channel alarm instructions, and realize closed-loop adaptive control of the welding parameters.

[0065] Through the collaborative cooperation of the above-mentioned various components, multi-point temperature measurements are carried out on the surface of the welded workpiece by a K-type thermocouple temperature sensor and an uncooled infrared thermal imager, and real-time temperature data with a sampling frequency of 10 Hz are obtained, realizing the comprehensive coverage and high-precision acquisition of the temperature field in the welding area, solving the problems of measurement blind spots and large errors of a single sensor, and eliminating the emissivity error through data fusion technology, improving the accuracy of temperature measurement; Fourier heat conduction calculation is performed on the welding thermal field according to the real-time temperature data to form a hot spot distribution grid map with a spatial resolution of 0.5 mm, realizing the accurate calculation of the heat conduction path and temperature distribution, enabling the system to capture temperature anomalies at the millimeter level; using the hot spot distribution grid map, boundary features of the temperature anomaly area exceeding the preset threshold are extracted through thermal gradient difference analysis, generating an abnormal hot area feature set containing position coordinates and temperature rise rates. An improved region growing algorithm is introduced to intelligently identify and extract the temperature anomaly area. This algorithm is optimized for the characteristics of the welding hot area, can accurately distinguish normal thermal gradients and abnormal hot spots, significantly reduces the false alarm rate, and enhances the system's ability to identify different types of thermal anomalies; according to the abnormal hot area feature set and combined with the material melting point parameters, multi-parameter weighted calculation of the risk level is carried out to form an early warning level matrix divided into four levels: low, medium, high, and emergency. By introducing material characteristic parameters and multi-dimensional feature analysis, the accuracy and differentiation of risk assessment are realized. The system can give a more scientific risk judgment according to the thermal sensitivity of different materials; according to the early warning level matrix, differential signal transmission is carried out on the sound-light-electricity integrated alarm system through the industrial control bus protocol, and multi-channel alarm instructions with frequency encoding are output, establishing a seamless connection mechanism from risk assessment to alarm response. The multi-channel alarm method ensures that information can be timely perceived in various working environments; based on the multi-channel alarm instructions, real-time adjustment of the current-voltage curve of the welding power controller is carried out to realize closed-loop adaptive control of welding parameters. This closed-loop control mechanism dynamically adjusts welding parameters according to abnormal hot area characteristics and material characteristics. The system considers the change of the resistance characteristics of the material at different temperatures and adopts a smooth transition parameter adjustment strategy, avoiding fluctuations and instabilities during the parameter adjustment process, improving the welding quality and consistency, further improving the analysis accuracy and response speed of the system, and significantly enhancing the system's adaptability to complex welding scenarios, enabling the early warning system to handle various non-linear thermal anomaly situations, effectively reducing the welding defect rate and rework rate, and greatly improving the production efficiency and product quality.

[0066] Refer to Figure 3 , in the embodiment of the present invention, a computer device is further provided. This computer device can be a server, and its internal structure can be as Figure 3As shown in the figure. The computer device includes a processor, a memory, a display screen, an input device, a network interface, and a database connected via a system bus. Among them, the processor of the computer design is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the corresponding data in this embodiment. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, the above method is implemented.

[0067] Those skilled in the art can understand that Figure 3 the structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied.

[0068] An embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the above method is implemented. It can be understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0069] Those of ordinary skill in the art can understand that all or part of the processes in the above embodiment methods can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the above method embodiments. Among them, any reference to a memory, storage, database, or other medium provided by the present invention and used in the embodiments can include non-volatile and / or volatile memories. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or an external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM, etc.

[0070] Those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working processes of the systems, systems, and units described above can refer to the corresponding processes in the foregoing method embodiments, and will not be elaborated herein.

[0071] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs that can store program codes.

[0072] The above is the case. The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the various embodiments of the present application.

Claims

1. A high-temperature warning method for a welding device, characterized in that, The high-temperature warning method for the welding device includes: Performing multi-point temperature measurement on the surface of the welded workpiece through a K-type thermocouple temperature sensor and a non-cooled infrared thermal imager to obtain real-time temperature data with a sampling frequency of 10 Hz; Performing Fourier heat conduction calculation on the welding thermal field according to the real-time temperature data to form a hot spot distribution grid map with a spatial resolution of 0.5 mm; Using the hot spot distribution grid map to extract the boundary characteristics of the temperature anomaly area exceeding the preset threshold through thermal gradient difference analysis, and generating an abnormal hot zone feature set including position coordinates and temperature rise rate; Performing multi-parameter weighted calculation on the risk level according to the abnormal hot zone feature set combined with the material melting point parameters to form a warning level matrix divided into four levels: low, medium, high, and emergency; Transmitting differential signals to the sound-light-electronic integrated alarm system according to the warning level matrix through the industrial control bus protocol, and outputting multi-channel alarm instructions with frequency encoding; Based on the multi-channel alarm instructions, performing real-time adjustment of the current-voltage curve of the welding power controller to achieve closed-loop adaptive control of welding parameters.

2. The high-temperature warning method for a welding device according to claim 1, characterized in that, The performing multi-point temperature measurement on the surface of the welded workpiece through a K-type thermocouple temperature sensor and a non-cooled infrared thermal imager to obtain real-time temperature data with a sampling frequency of 10 Hz includes: Arranging multiple groups of K-type thermocouples around the welding device according to the grid topology structure, performing contact-type temperature acquisition on the surface of the workpiece, and obtaining a target temperature data set; Performing thermal radiation imaging scanning on the welding area through a non-cooled infrared thermal imager to obtain a thermal image of the temperature distribution on the surface of the workpiece; Performing noise filtering processing on the target temperature data set to generate a calibrated contact-type temperature value; Performing pixel temperature mapping conversion on the thermal image of the temperature distribution on the surface of the workpiece to form a non-contact temperature matrix; Performing data fusion on the calibrated contact-type temperature value and the non-contact temperature matrix to eliminate the emissivity error and obtain a corrected temperature field; Packaging and transmitting the corrected temperature field through an industrial field bus at a sampling frequency of 10 Hz to form the real-time temperature data.

3. The high-temperature warning method for a welding device according to claim 1, characterized in that, The performing Fourier heat conduction calculation on the welding thermal field according to the real-time temperature data to form a hot spot distribution grid map with a spatial resolution of 0.5 mm includes: Converting the real-time temperature data into temperature spectrum domain data to obtain the frequency characteristics of the temperature spatial distribution; Performing heat flux density expansion on the temperature spectrum domain data through Fourier forward transformation to obtain the basic heat conduction equation; Constructing a thermal diffusion difference format according to the basic heat conduction equation to form a heat flux propagation vector field; Performing grid subdivision processing on the heat flux propagation vector field to establish a discrete calculation grid with an accuracy of 0.5 mm; Performing iterative solution on the discrete calculation grid through heat flux boundary condition constraints to obtain the steady-state and transient temperature distributions; Visualizing the steady-state and transient temperature distributions through pseudo-color mapping technology to generate the hot spot distribution grid map.

4. The high-temperature warning method for a welding device according to claim 1, characterized in that, The using the hot spot distribution grid map to extract the boundary characteristics of the temperature anomaly area exceeding the preset threshold through thermal gradient difference analysis, and generating an abnormal hot zone feature set including position coordinates and temperature rise rate includes: Calculate the temperature difference gradient of the hot spot distribution grid map to generate a temperature change rate distribution map; Perform binary processing on the temperature change rate distribution map according to the material safety threshold to obtain a temperature anomaly candidate area; Perform connected domain labeling on the temperature anomaly candidate area through an improved region growing algorithm to form the boundary of the hot spot contour; Extract the central coordinates, area, and perimeter features from the hot spot contour boundary to construct a hot spot spatial feature table; Perform temporal comparison on the same hot spot in multiple consecutive frames of the hot spot distribution grid map, calculate the temperature change speed of each hot spot, and obtain the hot spot temperature rise rate data; Merge and process the hot spot spatial feature table and the hot spot temperature rise rate data to generate the abnormal hot spot feature set.

5. The high-temperature warning method for a welding device according to claim 1, characterized in that, Perform multi-parameter weighted calculation on the risk level according to the abnormal hot spot feature set combined with the material melting point parameter to form a warning level matrix divided into four levels: low, medium, high, and emergency, including: Query the melting point, flash point, and safe working temperature corresponding to the welded workpiece from the material database to construct a material thermal property threshold table; Perform ratio calculation on the temperature values in the abnormal hot spot feature set and the material thermal property threshold table to obtain the temperature risk index; Divide the temperature rise rate in the abnormal hot spot feature set by the material safe temperature rise rate to obtain the temperature rise risk coefficient; Calculate the hot spot diffusion index according to the proportional relationship between the hot spot area in the abnormal hot spot feature set and the standard area of the welding point; Perform weighted summation on the temperature risk index, the temperature rise risk coefficient, and the hot spot diffusion index to generate a comprehensive risk score; Map the comprehensive risk score to four levels: low, medium, high, and emergency through multi-level threshold segmentation to form the warning level matrix.

6. The high-temperature warning method for a welding device according to claim 1, characterized in that Transmit differential signals to the sound-light-visual integrated alarm system according to the warning level matrix through the industrial control bus protocol, and output multi-channel alarm instructions with frequency encoding, including: Map the low-level warning in the warning level matrix to a green flashing signal code to generate a visual warning data packet; Convert the medium-level warning in the warning level matrix into a medium-frequency beeping pulse sequence to form an audio warning signal; Construct a tactile feedback instruction with vibration intensity parameters according to the high-level warning in the warning level matrix to obtain a tactile alarm code; Compile an emergency response instruction containing the equipment ID and fault type for the emergency warning in the warning level matrix to generate a system control command; Pack the visual warning data packet, the audio warning signal, the tactile alarm code, and the system control command according to the priority to construct a multi-channel instruction data stream; Send the multi-channel instruction data stream to each terminal device in a multicast manner through the industrial Ethernet to output the multi-channel alarm instructions with frequency encoding.

7. The high-temperature warning method for a welding device according to claim 1, characterized in that Based on the multi-channel alarm instructions, perform real-time adjustment of the current-voltage curve of the welding power controller to achieve closed-loop adaptive control of welding parameters, including: Parse the risk level identifier and hot spot position information from the multi-channel alarm instructions to generate a welding danger status table; Query the process parameter library according to the welding danger status table to obtain the corresponding current-voltage adjustment strategy; Convert the current-voltage adjustment strategy into a digital signal control quantity to form a power control data stream; Dynamically compensate the power control data stream according to the welding material characteristic curve to obtain a smoothly transitioning current-voltage adjustment curve; Convert the current-voltage adjustment curve into an analog control signal through a digital-to-analog converter and output it to the welding power module; The welding power module dynamically adjusts the welding parameters in real time according to the analog control signal to complete the closed-loop adaptive control of the welding parameters.

8. A high-temperature warning system for a welding device, used to implement the high-temperature warning method for a welding device according to any one of claims 1-7, characterized in that The high-temperature warning system for the welding device includes: An extraction module for performing multi-point temperature measurement on the surface of the welded workpiece through a K-type thermocouple temperature sensor and a non-cooled infrared thermal imager to obtain real-time temperature data with a sampling frequency of 10 Hz; A calculation module for performing Fourier heat conduction calculation on the welding thermal field according to the real-time temperature data to form a hot spot distribution grid map with a spatial resolution of 0.5 mm; An extraction module for using the hot spot distribution grid map to extract the boundary features of the temperature anomaly region exceeding the preset threshold through thermal gradient difference analysis to generate an abnormal hot zone feature set including position coordinates and temperature rise rate; A weighting module for performing multi-parameter weighted calculation on the risk level according to the abnormal hot zone feature set combined with the material melting point parameters to form a warning level matrix divided into four levels: low, medium, high, and emergency; A transmission module for differentially transmitting signals to the sound-light integration alarm system according to the warning level matrix through the industrial control bus protocol and outputting multi-channel alarm instructions with frequency encoding; An adjustment module for performing real-time adjustment of the current-voltage curve on the welding power controller based on the multi-channel alarm instructions to achieve closed-loop adaptive control of the welding parameters.

9. A computer device, characterized in that It includes a memory and a processor. The memory stores a computer program that can run on the processor. When the processor executes the computer program, it implements the high-temperature warning method for the welding device according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that A computer program is stored thereon. When the computer program is run by the processor, the processor is caused to execute the high-temperature warning method for the welding device according to any one of claims 1 to 7.

Images