Multi-welding-head battery welding combined array detecting and positioning method

Abstract

The invention provides an array detection and positioning method combined with multi-welding-head battery welding, which comprises the following steps: monitoring the distribution of a welding head temperature field in real time by adopting an infrared thermal imaging technology, judging whether the welding head temperature field deviates or not by combining a preset temperature threshold value, and judging that the welding head temperature field deviates if detecting that the temperature of a certain welding head exceeds a threshold value range; according to the offset of the welding position, motion control parameters of the digital welding system are adjusted in a self-adaptive mode, and the offset error of a welding head is compensated in real time through a closed-loop feedback control algorithm; the welding spot quality online monitoring technology is adopted, the welding spot fusion depth and the welding spot diameter are extracted, whether the welding spot quality is qualified or not is judged in real time in combination with a pre-established welding spot quality evaluation model, and if the quality of a plurality of continuous welding spots is not qualified, it is judged that the welding spot quality is affected by welding head temperature field deviation.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to an array detection and positioning method combined with multi-welding head battery welding. Background Art

[0002] During the long-term operation of the multi-welding head array, the temperature field of the welding head is offset due to the thermal field coupling effect between the welding heads. The spacing of the welding head arrangement is a key factor affecting heat conduction. If the spacing is too small, the coupling transfer of heat between the welding heads will be aggravated, which will aggravate the temperature field offset of the welding head; if the spacing is too large, the thermal coupling between the welding heads will be weakened, but the size of the welding array will be increased. The offset of the welding head temperature field will cause the welding head base to produce non-uniform thermal expansion deformation, which will cause the welding position to shift. As the working time increases, the position offset will continue to accumulate, which will have a significant impact on the synchronization of the welding array. The inconsistent position offset of different welding heads will cause changes in the welding parameters and welding trajectories of each welding point, resulting in differences in the quality of the welding points, and ultimately affecting the overall welding quality of the workpiece. The thermal deformation of the welding head base is closely related to the welding head layout, welding process parameters, base structure design and heat dissipation conditions. It is necessary to take into account the overall considerations in the welding head layout design, optimize the welding process, improve heat dissipation, and reduce thermal deformation, so as to ensure the consistency of the quality of the welding points. Summary of the invention

[0003] The present invention provides an array detection and positioning method combined with multi-welding head battery welding, which mainly includes:

[0004] The arrangement spacing and material properties of the welding head array are obtained, and a mathematical model of heat conduction of the welding head array is established based on the arrangement spacing and material properties of the welding head array. The temperature field distribution of the welding heads under different arrangement spacings is simulated and calculated by the finite element analysis method to determine the influence of the welding head spacing on heat conduction.

[0005] Infrared thermal imaging technology is used to monitor the temperature field distribution of the welding head in real time, and the preset temperature threshold is combined to determine whether the temperature field of the welding head is offset. If it is detected that the temperature of a welding head exceeds the threshold range, it is determined that the temperature field of the welding head is offset;

[0006] After determining that the temperature field has shifted, the temperature field offset is calculated by the temperature difference before and after the welding head temperature field offset, and the thermal expansion coefficient of the welding head base material is obtained. Combined with the temperature field offset, the thermal expansion deformation of the welding head base under the condition of temperature field offset is calculated, and the displacement deformation of the welding head base is detected in real time through machine vision measurement technology;

[0007] The actual position of the welding head is obtained, and the thermal expansion deformation and temperature field offset are combined and correlated. Through theoretical derivation and experimental verification, a ternary mathematical model of welding position offset, thermal expansion deformation and temperature field offset is constructed to calculate the welding position offset of each welding head caused by thermal effect in real time;

[0008] According to the offset of welding position, the motion control parameters of digital welding system are adaptively adjusted, and the welding head offset error is compensated in real time through closed-loop feedback control algorithm;

[0009] Adopt the online monitoring technology of solder joint quality, extract the solder joint penetration depth and solder joint diameter, and combine with the pre-established solder joint quality evaluation model to judge whether the solder joint quality is qualified in real time. If the quality of multiple consecutive solder joints is unqualified, it is determined that the temperature field offset of the welding head affects the quality of the solder joint;

[0010] A machine learning algorithm is used to perform fusion calculation of the welding head temperature field offset, welding position offset and solder joint quality, and a multi-parameter coupling model of the welding process is constructed. The key parameters reflecting the temperature field offset law, welding position offset characteristics and solder joint defect mode are extracted. Through data association analysis and mechanism deduction, the influence of the welding head thermal field coupling effect on the welding quality is obtained, and the quantitative relationship between temperature field offset, welding position offset and solder joint quality is established.

[0011] The technical solution provided by the embodiment of the present invention may have the following beneficial effects:

[0012] The present invention discloses an array detection and positioning method for battery welding combined with multiple welding heads. By establishing a thermal conduction model of the welding head array, the temperature field distribution of the welding head is monitored in combination with infrared thermal imaging to determine the temperature field offset. Based on the temperature field offset and the thermal expansion coefficient, the thermal expansion deformation of the welding head base is calculated. Machine vision is used to measure the actual position of the welding head, and a mathematical model of welding position offset and thermal effect is constructed. The welding system parameters are adaptively adjusted according to the offset to compensate for the welding head error in real time. At the same time, the quality of the solder joint is monitored to determine the influence of the temperature field offset on the solder joint. The machine learning algorithm is used to fuse and analyze the temperature field, position offset and solder joint quality, and a multi-parameter coupling model is established to reveal the influence of the thermal field coupling effect on the welding quality. The present invention realizes the precise compensation of thermal effects during welding, and effectively improves the welding quality and precision. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 The present invention is a flow chart of an array detection and positioning method combined with multi-welding head battery welding.

[0014] Figure 2 It is a schematic diagram of an array detection and positioning method combined with multi-welding head battery welding of the present invention.

[0015] Figure 3It is another schematic diagram of an array detection and positioning method combined with multi-welding head battery welding of the present invention. DETAILED DESCRIPTION

[0016] The following will describe the technical solutions in the embodiments of the present invention in detail in conjunction with the accompanying drawings in the embodiments of the present invention. The described embodiments are only part of the embodiments of the present invention.

[0017] like Figure 1 -3. In this embodiment, an array detection and positioning method combined with multi-welding head battery welding may specifically include:

[0018] S101. Obtain the arrangement spacing and material properties of the welding head array, and establish a mathematical model of heat conduction of the welding head array based on the arrangement spacing and material properties of the welding head array. Use the finite element analysis method to simulate and calculate the temperature field distribution of the welding heads under different arrangement spacings, and determine the influence of the welding head spacing on heat conduction.

[0019] According to the arrangement mode of welding head array, the horizontal and vertical spacing values ​​of welding head array are obtained, and the initial value of temperature field on the surface of welding head array is collected by infrared thermal imager to obtain temperature distribution function; the thermal conductivity of welding head material is extracted according to the initial value of temperature field, and the heat diffusion function is established by using the thermal conductivity and thermal impedance, and the heat diffusion function includes temperature diffusion coefficient; the heat conduction calculation area of ​​welding head array is divided into square grid units, and the fourth-order Runge-Kutta algorithm is used to solve the heat conduction differential equation in the grid unit to obtain temperature-time variation function; the temperature distribution function is fitted with the temperature-time variation function, and the linear least squares method is used to calculate the correspondence between welding head spacing and heat diffusion to obtain the spatial distribution law of temperature field.

[0020] Exemplarily, the horizontal and vertical spacing values ​​of the welding head array are obtained according to the arrangement of the welding head array, the initial value of the temperature field on the surface of the welding head array is collected by an infrared thermal imager at a sampling interval of 0.1 seconds, the temperature field gradient value is calculated based on the initial value of the temperature field, and the temperature gradient value is expanded by Taylor series to obtain the temperature distribution function. The thermal conductivity of the welding head material is extracted from the initial value of the temperature field collected by the thermal imager, the thermal diffusion coefficient is calculated according to the density and specific heat capacity of the welding head material, and a heat diffusion function Q=λA(T2-T1) / L including the temperature diffusion coefficient and thermal impedance is established, where λ is the thermal conductivity, A is the cross-sectional area through which the heat flow passes, T2-T1 is the temperature difference between two points, and L is the heat transfer distance. The heat conduction calculation area of ​​the welding head array is divided into square grid units with a side length of 0.5 mm, and the temperature boundary conditions are set in the grid units. The fourth-order Runge-Kutta algorithm is used to solve the heat conduction differential equation, and the grid unit node temperature is iterated at a time interval of 0.1 seconds to obtain the temperature-time change function. The temperature distribution function and the temperature-time variation function are fitted, and the linear least square method is used to calculate the correspondence between the spacing between the welding heads and the amount of heat diffusion. The spatial distribution law of the temperature field is obtained from the corresponding relationship, and a quantitative characterization function of the spacing between the welding heads for heat conduction is established. The welding head array includes the horizontal and vertical arrangement spacing in the process parameters. The horizontal spacing represents the distance between adjacent welding heads in the horizontal direction, and the vertical spacing represents the distance between adjacent welding heads in the vertical direction. The arrangement spacing of the welding head array is usually in the range of 2-10 mm. In view of the temperature field characteristics of the welding head array, the infrared thermal imager obtains the temperature distribution by detecting the infrared radiation intensity on the surface of the welding head. When the sampling interval is set to 0.1 seconds, the transient change process of the temperature can be captured, and the temperature field gradient value reflects the rate of change of the temperature in space. The thermal conductivity of the welding head material describes the strength of the material's thermal conductivity. For the copper welding head, its thermal conductivity is 380 watts per meter Kelvin at 20 degrees Celsius. As the temperature rises, the thermal conductivity shows a downward trend, and drops to 340 watts per meter Kelvin at 200 degrees Celsius. The thermal diffusion coefficient characterizes how fast the temperature propagates in the material, and is related to the thermal conductivity, density and specific heat capacity of the material. The thermal diffusion coefficient of the copper welding head is about 1.1 square centimeters per second. In the calculation of heat conduction of the welding head array, the fineness of the grid division affects the calculation accuracy. The square grid unit with a side length of 0.5 mm can better reflect the spatial distribution characteristics of the temperature field. The temperature boundary conditions include the constant temperature boundary of the welding head heating surface and the convection heat transfer boundary of the outer surface. The temperature of the welding head heating surface is set to 350 degrees Celsius, and the heat transfer coefficient of the outer surface is 20 watts per square meter Kelvin. The effect of the welding head spacing on the temperature field distribution is reflected in the thermal interaction between adjacent welding heads. When the welding head spacing is small, the heat transfer between welding heads is more significant. When the spacing is 3 mm, the temperature difference between adjacent welding heads is less than 10 degrees Celsius, and the temperature field presents a relatively uniform distribution.As the spacing increases to 8 mm, the temperature difference between adjacent welding heads increases to 25 degrees Celsius, and the temperature field distribution shows obvious local high temperature areas. The spatial distribution law of the temperature field is quantitatively described by the corresponding relationship between the spacing between welding heads and the temperature gradient. For every 1 mm increase in spacing, the temperature gradient value increases by about 4 degrees Celsius per millimeter. In the process of heat conduction, the thermal conductivity and thermal impedance of the welding head material jointly determine the efficiency of heat transfer. The larger the thermal conductivity, the faster the heat transfer, while the thermal impedance reflects the resistance in the heat transfer process. For copper welding heads, at an operating temperature of 350 degrees Celsius, the thermal impedance value is about 0.15 Kelvin per watt. There is a difference in the transverse and longitudinal heat conduction rates, and the transverse conduction rate is 15% higher than the longitudinal conduction rate.

[0021] S102, using infrared thermal imaging technology to monitor the temperature field distribution of the welding head in real time, and judging whether the temperature field of the welding head is offset in combination with a preset temperature threshold. If it is detected that the temperature of a welding head exceeds the threshold range, it is judged that the temperature field of the welding head is offset.

[0022] The temperature field image of the welding head array surface is collected by an infrared thermal imager, and the temperature field image is processed by Gaussian filtering to obtain a temperature field filtered image; the temperature color scale value of the pixel point in the welding head area is obtained from the temperature field filtered image, and the temperature gradient value is calculated according to the temperature color scale value to obtain temperature time series data; the deviation amplitude between the temperature value of the welding head area and the reference temperature is calculated according to the temperature time series data, and the welding head position coordinates and the temperature deviation value are recorded when the temperature deviation amplitude exceeds the preset range; the temperature deviation value of the welding head position is accumulated, and if the temperature deviation accumulated value exceeds the preset temperature fluctuation threshold, it is determined that the welding head temperature field has an abnormal deviation, and the welding head center point coordinates and the temperature deviation accumulated value are obtained as temperature anomaly records.

[0023] Exemplarily, the temperature field image of the welding head array surface is collected by an infrared thermal imager at a sampling interval of 0.1 seconds, and the image resolution is set to 640x480 pixels. The temperature field image is subjected to Gaussian filtering denoising with a kernel size of 5x5 pixels and a standard deviation of 1.5 to obtain a temperature field filter image, and the temperature value and temperature field distribution data of each welding head area are extracted from the temperature field filter image. The temperature color scale value of the pixel point in the welding head array area is obtained from the temperature field filter image, and the temperature color scale value is divided into 256 gray levels within the range of 200 to 400 degrees Celsius. The temperature gradient value is calculated by scanning the 8 neighboring pixel points around each welding head area, and a temperature time series data record table is constructed for the temperature gradient value. The temperature time series data is statistically analyzed for the numerical changes within the 0.1 second interval to obtain a temperature fluctuation curve. The deviation between the temperature value of each welding head area and the 350 degrees Celsius reference temperature is calculated according to the temperature fluctuation curve. When the temperature deviation exceeds the range of plus or minus 5 degrees Celsius, the welding head position coordinates and temperature deviation value are recorded, and the temperature deviation value of the welding head position is accumulated in real time in the subsequent 0.5 second time window. The accumulated temperature deviation value of each welding head position is compared with the preset temperature fluctuation threshold. If the accumulated temperature deviation value of the welding head position in the 0.5 second time window exceeds the range of plus or minus 20 degrees Celsius, it is determined that the welding head temperature field has an abnormal deviation, and the coordinates of the welding head center point and the accumulated temperature deviation value are extracted as temperature anomaly records. The infrared thermal imager obtains temperature distribution data by detecting the infrared radiation intensity on the welding head surface. The 640x480 pixel resolution can achieve a spatial resolution accuracy of 0.2 mm in the 10 cm x 8 cm welding head array area, and each welding head area contains about 400 effective pixels. The sampling interval of 0.1 seconds corresponds to a sampling frequency of 10 Hz, which meets the monitoring requirements of the rapid change of the welding head temperature. During the heating stage of the welding head, a temperature change rate of 20 degrees Celsius per second can be captured. The temperature field image denoising process uses a 5x5 pixel Gaussian filter kernel. The standard deviation of 1.5 effectively suppresses thermal noise while retaining the edge features of the temperature field. For images at a temperature of 350 degrees Celsius, the noise amplitude after filtering is reduced from the original plus or minus 2 degrees Celsius to plus or minus 0.5 degrees Celsius. The welding head area extraction is based on temperature gradient boundary detection. The difference between the center temperature of the welding head and the background temperature is usually more than 100 degrees Celsius, forming a clear temperature boundary. The temperature color scale mapping uses 256 grayscale values ​​to represent the temperature range of 200 to 400 degrees Celsius, with a resolution of 0.8 degrees Celsius per grayscale level. The temperature field gradient calculation of the welding head is obtained through the temperature difference of 8 neighboring pixels. Under normal working conditions, the temperature gradient value from the center to the edge of the welding head is in the range of 25 to 35 degrees Celsius per millimeter. The temperature fluctuation curve records the temperature change within 0.1 second intervals. When working stably, the temperature fluctuation range should be kept within plus or minus 1 degree Celsius. The temperature deviation value of the welding head is calculated based on the reference temperature of 350 degrees Celsius. The deviation threshold of plus or minus 5 degrees Celsius takes into account the temperature measurement accuracy and process allowable range.The accumulated temperature deviation within the 0.5 second time window reflects the persistence of the temperature anomaly. The instantaneous deviation caused by normal fluctuations will offset each other after integration, while the continuous deviation will cause the accumulated value to increase. When the accumulated temperature deviation within 0.5 seconds exceeds plus or minus 20 degrees Celsius, it indicates that the welding head temperature has continuously deviated from the normal range. At this time, the recorded welding head coordinates and the accumulated temperature deviation value can be used to locate the faulty welding head and determine the degree of temperature deviation. In actual application scenarios, the abnormal temperature of the welding head is usually manifested as low temperature caused by insufficient heating power, or high temperature caused by heat accumulation. When the temperature is low, the temperature will continue to be lower than 345 degrees Celsius within 0.5 seconds, and the accumulated deviation value will reach minus 25 degrees Celsius. When the temperature is high, it is difficult for local heat to dissipate, resulting in a temperature that is continuously higher than 355 degrees Celsius, and the accumulated deviation value exceeds plus 25 degrees Celsius. By real-time monitoring of the accumulated temperature deviation value, the problem welding head can be discovered and located in time before the temperature anomaly develops to affect the process quality.

[0024] S103. After determining that the temperature field has shifted, the temperature field offset is calculated by the temperature difference before and after the welding head temperature field offset, the thermal expansion coefficient of the welding head base material is obtained, and the thermal expansion deformation of the welding head base under the condition of temperature field offset is calculated in combination with the temperature field offset, and the displacement deformation of the welding head base is detected in real time by machine vision measurement technology.

[0025] The temperature field value of the welding head is collected by the infrared thermal imager, and the temperature difference is calculated by the temperature value of the central area of ​​the welding head in the time window before and after the temperature field anomaly judgment moment, so as to obtain the temperature field offset of the welding head base; the linear thermal expansion coefficient value in the physical property database of the welding head base copper material and the temperature field offset are used to calculate the theoretical thermal expansion displacement according to the three-dimensional size parameters of the base; the corner feature mark coordinates are extracted from the base surface image collected by the industrial camera, and the feature mark coordinates are tracked by the sub-pixel edge detection algorithm to obtain the actual deformation of the base surface; the Kalman filter algorithm is used to perform filtering processing on the theoretical displacement of the base thermal expansion and the actual deformation to obtain the three-dimensional space deformation vector of the base, and if the deformation vector exceeds the positive and negative threshold range, it is determined that the thermal strain of the base is abnormal.

[0026] For example, the temperature field value of the welding head is collected by an infrared thermal imager at a time interval of 0.1 seconds, and the temperature value of the 10x10 pixel range in the central area of ​​the welding head is extracted in a time window of 0.5 seconds before and after the temperature field abnormality is determined. The temperature value before and after the temperature field abnormality is time-series registered and the temperature difference is calculated, and a three-dimensional temperature distribution function is constructed from the temperature difference.

[0027]

[0028] f(x, y, z) represents the temperature distribution function in three-dimensional space, A is the amplitude, (x0, y0, z0) is the center position of the temperature distribution, and σ is the standard deviation of the distribution, which controls the width of the distribution. The temperature field offset of each area of ​​the welding head base is obtained through the three-dimensional temperature distribution function. The linear thermal expansion coefficient values ​​in the range of 20 to 400 degrees Celsius are obtained from the physical property database of the copper material of the welding head base. According to the three-dimensional size parameters of the length, width and height of the welding head base and the temperature field offset, the linear thermal expansion formula is used to calculate the theoretical thermal expansion displacement of the base in three-dimensional space, and the theoretical thermal expansion displacement is updated in real time at intervals of 0.1 seconds. The coordinates of the feature markers of 9 corner points are extracted from the 1920x1080 pixel image of the base surface collected by the industrial camera, and the sub-pixel edge detection algorithm is used to track the feature marker coordinates in real time at intervals of 0.1 seconds, and the displacement increment of the feature marker coordinates in three-dimensional space is calculated, and the actual deformation of the base surface is obtained through the displacement increment. For the data sequence of the theoretical displacement and actual deformation of the thermal expansion of the base within the 0.5 second time window, the Kalman filter algorithm is used to perform real-time filtering on the displacement data, and the deformation vector of the base in three-dimensional space is extracted from the filtering result. If the deformation vector exceeds the range of plus or minus 0.1 mm, the thermal strain of the base is judged to be abnormal. The abnormal judgment of the temperature field of the welding head is realized by real-time monitoring of the temperature difference. The temperature field image collected by the infrared thermal imager has high temporal and spatial resolution characteristics. The sampling interval of 0.1 seconds ensures the capture of transient temperature changes. The 10x10 pixel welding head center area contains 100 temperature measurement points, providing a stable and reliable temperature data foundation. When the welding head is heated abnormally, the temperature field will show obvious changes within the 0.5 second time window, and the temperature difference will increase from plus or minus 2 degrees Celsius in the normal state to more than 10 degrees Celsius. The welding head base is made of high-purity copper material. Its linear thermal expansion coefficient shows nonlinear characteristics with temperature changes. It is 16.5 microns per meter per degree Celsius at 20 degrees Celsius, and gradually increases with the increase of temperature, reaching 18.5 microns per meter per degree Celsius at 400 degrees Celsius. The typical size of the base is 50 mm long, 30 mm wide, and 20 mm high. When the temperature shifts by 50 degrees Celsius, the theoretical thermal expansion in the length direction reaches 41 microns and in the width direction reaches 25 microns. In the base surface image collected by the industrial camera, 9 characteristic markers are distributed in a 3x3 matrix, with a spacing of 15 mm between markers. The displacement measurement accuracy of 0.5 microns is achieved through the sub-pixel edge detection algorithm. When the base surface temperature rises and causes thermal expansion deformation, the coordinates of the markers change, and the distance between adjacent markers increases, forming a non-uniform deformation field. In the early stage of temperature field offset, the deformation field shows the characteristics of central expansion and peripheral contraction. As the temperature continues to rise, the deformation gradually transitions to overall expansion. The Kalman filter algorithm processes the displacement data in real time, and the filter window is set to 0.5 seconds, including 5 continuous sampling points. Under normal working conditions, the thermal expansion deformation of the base remains stable, and the displacement vector after filtering fluctuates within the range of plus or minus 0.02 mm.When the temperature field is abnormal, the thermal expansion deformation intensifies, and the displacement vector increases rapidly within 0.5 seconds. In the thermal expansion anomaly caused by the high temperature of the welding head, the top area of ​​the base displaces upward by 0.15 mm, and the side expands outward by 0.12 mm, forming an uneven three-dimensional deformation. The thermal contraction caused by the low temperature causes the overall size of the base to decrease, the top sinks by 0.08 mm, and the side shrinks by 0.06 mm. The thermal deformation monitoring data shows that the base material will produce cumulative strain during the cyclic heating process, which is manifested as the displacement of the mark point increases with the number of heating times. In the initial 10 heating cycles, the maximum displacement gradually increased from 0.1 mm to 0.13 mm, and the displacement direction also changed from perpendicular to the surface to 45 degrees, reflecting the evolution of the stress distribution inside the base. By real-time tracking the size and direction of the displacement vector, the early characteristics of the base thermal strain anomaly can be identified.

[0029] S104, obtaining the actual position of the welding head, and combining the thermal expansion deformation and the temperature field offset, and performing correlation analysis, constructing a ternary mathematical model of the welding position offset, the thermal expansion deformation and the temperature field offset through theoretical derivation and experimental verification, and calculating the welding position offset of each welding head caused by the thermal effect in real time.

[0030] The welding head surface image is collected by an industrial camera, the center coordinates of the calibration pattern are extracted from the welding head surface image, and the camera parameters are calibrated using the center coordinates to obtain the welding head position deviation vector; the temperature field data of the welding head area is obtained by an infrared thermal imager, the temperature deviation value is calculated according to the temperature field data, and the thermal expansion displacement vector is obtained by combining the temperature deviation value with the thermal expansion coefficient; a Bayesian network model is constructed according to the welding head position deviation vector and the thermal expansion displacement vector, and a position compensation vector is extracted from the Bayesian network model; a recursive filtering operation is performed on the position compensation vector, and a temperature compensation value is obtained through the recursive filtering operation, and a three-dimensional space compensation function is constructed according to the temperature compensation value.

[0031] For example, based on the 1920x1080 pixel welding head surface image collected by the industrial camera at 0.1 second intervals, the center coordinates of the 9-point calibration pattern on the welding head surface are extracted, the internal and external parameters of the camera are calibrated using the Zhang Zhengyou calibration algorithm, and the 3D spatial position of the characteristic points on the welding head surface in the world coordinate system is calculated by 3D reconstruction, and the 3D spatial position is compared with the coordinates of the machining reference point to obtain the welding head position deviation vector. The temperature field distribution data within the 10x10 pixel range of the welding head area is obtained from the infrared thermal imager, and the temperature deviation value of each pixel point is calculated for the reference temperature of 350 degrees Celsius, and the temperature field space offset function is constructed. Combined with the thermal expansion coefficient change curve of copper material in the range of 20 to 400 degrees Celsius, the 3D thermal expansion displacement vector of the welding head base is calculated, and a real-time mapping relationship between the temperature deviation value and the thermal expansion displacement vector is established. According to the change data of the welding head position deviation vector and the thermal expansion displacement vector in the 0.5 second time window, the fifth-order Bayesian network algorithm is used to train the coupling model of temperature field offset and spatial displacement. The Bayesian network contains position nodes, temperature nodes and thermal expansion nodes. The position compensation vector is extracted from the coupling model and updated at intervals of 0.1 seconds in three-dimensional space. A three-dimensional spatial recursive filter is used to perform real-time calculations on the position compensation vector. A 0.5 second sliding time window is set. The position compensation vector is forward recursively calculated in the sliding window. The temperature compensation value of the actual position of the welding head is obtained from the recursive calculation results, and a three-dimensional spatial compensation function including temperature field changes and thermal expansion deformation is constructed. The industrial camera uses a 9-point calibration pattern when monitoring the welding head position online. The calibration pattern consists of a 3x3 matrix of circular marking points with a diameter of 2 mm. The spacing between marking points is 15 mm. The sub-pixel center coordinates are extracted through circular markings. The camera calibration uses Zhang Zhengyou algorithm. The size of the calibration plate is 100x100 mm. The calibration images are collected at 5 different viewing angles. The camera focal length is calculated to be 8 mm, the principal point coordinate offset is less than 0.1 mm, and the radial distortion coefficient is within 0.01. The temperature field distribution on the surface of the welding head shows the characteristics of high temperature in the center and low temperature at the edge. The 10x10 pixel range covers an area of ​​100 square millimeters in the center area of ​​the welding head, and a single pixel corresponds to an actual size of 1 square millimeter. When 350 degrees Celsius is used as the reference temperature, the temperature fluctuation range of the center area is within plus or minus 2 degrees Celsius under normal working conditions, and the temperature gradient of the edge area reaches 15 degrees Celsius per millimeter. When the welding head is heated abnormally, the center temperature rises to 380 degrees Celsius, and the edge temperature rises to 365 degrees Celsius, and the temperature gradient decreases to 8 degrees Celsius per millimeter. The thermal expansion coefficient of copper material has a significant temperature dependence, which is 16.5 microns per meter per degree Celsius at 20 degrees Celsius, increases to 17.5 microns per meter per degree Celsius at 200 degrees Celsius, and reaches 18.5 microns per meter per degree Celsius at 400 degrees Celsius. The thermal expansion displacement of the welding head base in three-dimensional space is anisotropic, with the largest longitudinal thermal expansion amount reaching 45 microns, followed by 30 microns in the transverse direction, and the smallest vertical direction is 20 microns.The mapping relationship between temperature field offset and thermal expansion displacement shows that for every 10 degrees Celsius increase in temperature, the longitudinal displacement increases by 9 microns and the lateral displacement increases by 6 microns. The temperature field offset and spatial displacement coupling model constructed by the Bayesian network contains a three-layer network structure. The input layer receives temperature field data, the middle layer processes thermal expansion calculations, and the output layer generates position compensation. The model training uses 1,000 sets of temperature field and displacement data, and the prediction accuracy reaches plus or minus 2 microns after convergence. The position compensation vector is updated 50 times within a 0.5-second sliding window to achieve a compensation frequency of 10 Hz. The three-dimensional spatial recursive filter uses a 5-point filter kernel to smooth the compensation vector, and the position fluctuation after filtering is controlled within plus or minus 0.5 microns. Under the condition of rapid temperature change of the welding head, it takes 0.3 seconds for the temperature field to rise from 350 degrees Celsius to 380 degrees Celsius, the thermal expansion displacement response lags by 0.1 seconds, and the position compensation is updated within 0.5 seconds. During the compensation process, the maximum position deviation reached 55 microns, which was reduced to 35 microns after recursive filtering, and the deviation from the theoretically calculated thermal expansion displacement was less than 5 microns. The delay time of the compensation function to the temperature field change was kept within 0.2 seconds, meeting the real-time compensation requirements for the welding head position.

[0032] S105, adaptively adjusting the motion control parameters of the digital welding system according to the welding position offset, and compensating the welding head offset error in real time through a closed-loop feedback control algorithm.

[0033] According to the numerical comparison between the welding head position offset and the preset compensation threshold, the compensation gain coefficient is calculated by using the particle swarm optimization algorithm; a three-dimensional space compensation vector is constructed by the product operation of the compensation gain coefficient and the position offset, and a compensation response curve containing position component, velocity component and acceleration component is generated from the compensation vector; the compensation amount is extracted from the compensation response curve to construct a third-order state space equation, and the state space equation is iteratively calculated by using the Kalman state observer to obtain the compensation control amount; the position loop gain parameter, the velocity loop gain parameter and the current loop gain parameter are updated according to the compensation control amount, the actual compensation displacement is obtained by adjusting the gain parameter, the actual compensation displacement is numerically compared with the theoretical compensation amount to obtain the compensation tracking error, and the compensation tracking error is closed-loop regulated by a proportional-integral-differential regulator.

[0034] Exemplarily, the compensation gain coefficient is calculated by comparing the welding head position offset with the preset compensation threshold of 0.1 mm, and the particle swarm optimization algorithm is used to calculate the compensation gain coefficient. The particle swarm size is set to 50, the number of iterations is 100, and the three-dimensional space compensation vector is constructed by multiplying the compensation gain coefficient with the position offset, and a compensation response curve containing position, velocity, and acceleration components is generated from the compensation vector. The position compensation, velocity compensation, and acceleration compensation are extracted from the compensation response curve, and a third-order state space equation is constructed by the compensation. The state space equation is iterated at intervals of 0.001 seconds using the Kalman state observer, and the compensation control amount is obtained from the iterative calculation result, and the compensation control amount is converted into a servo control instruction. According to the servo control instruction, the position loop gain, velocity loop gain, and current loop gain parameters are updated in real time, with the position loop gain range of 0.5 to 2, the velocity loop gain range of 2 to 8, and the current loop gain range of 5 to 20. The compensation control amount is dynamically tracked within a sampling interval of 0.001 seconds by adjusting the gain parameters to obtain the actual compensation displacement. The actual compensation displacement is compared with the theoretical compensation, the compensation tracking error is calculated, and the proportional integral differential regulator is used to adjust the compensation tracking error in a closed loop, with a proportional coefficient of 4, an integral time of 0.02 seconds, and a differential time of 0.005 seconds. The adjustment output result is fed back to the compensation gain coefficient calculation unit to realize the iterative optimization of the welding head position compensation. The welding head position compensation control involves a multi-layer feedback adjustment mechanism. The preset compensation threshold of 0.1 mm is based on the requirements of the welding process for the joint position accuracy. When it is higher than the threshold, the compensation control is activated. Particle swarm optimization selects 50 particles, each of which contains weight coefficients of three dimensions: position compensation, velocity compensation, and acceleration compensation. The optimal compensation gain combination is found in 100 iterations. When the position offset is 0.15 mm, the optimized compensation gain coefficient is 1.2 in the horizontal direction and 0.8 in the vertical direction, ensuring the linear correspondence between the compensation amount and the offset amount. The state space equation describes the dynamic change process of the compensation amount, and the position compensation, velocity compensation, and acceleration compensation constitute the third-order state variables. The Kalman observer updates the estimated state variables with a frequency of 0.001 seconds, the initial value of the prediction covariance matrix is ​​set to 0.01, and the measurement noise covariance is set to 0.001, so as to achieve fast tracking of the compensation state. When the welding head position suddenly changes by 0.2 mm, the observer converges to the steady-state estimate within 0.005 seconds, and the position estimation error is less than 0.01 mm. The dynamic adjustment of the servo control parameters realizes the precise execution of the compensation instruction. The position loop gain starts from 0.5 and gradually increases to 2 as the compensation amount increases, thereby improving the position tracking bandwidth. The speed loop gain is adjusted in the range of 2 to 8. When the compensation amount is large, the gain value is increased to speed up the response speed. The current loop gain ranges from 5 to 20 to ensure smooth torque output. In the 0.2 mm step displacement compensation, the position loop gain is adjusted to 1.5, the speed loop gain is 6, and the current loop gain is 15, achieving an adjustment time of 0.015 seconds.The proportional-integral-differential regulator constitutes the outermost compensation control loop. The proportional coefficient 4 provides the basic compensation gain, the 0.02 second integral time eliminates the steady-state error, and the 0.005 second differential time improves the dynamic response characteristics. The compensation tracking error is obtained by comparing the actual displacement with the theoretical compensation amount. During the forward 0.1 mm displacement compensation process, the overshoot is controlled within 15%, and the steady-state error is less than 0.005 mm. The regulator output and particle swarm optimization form a dual compensation loop. The fitness function of the particle swarm during the optimization iteration process includes three evaluation indicators: overshoot, adjustment time, and steady-state error. In the position offset compensation caused by the temperature change of the welding head, a displacement deviation of 0.05 mm is generated for every 10 degrees Celsius increase in temperature. The compensation control completes the response within 0.05 seconds, and the position compensation error is controlled within 0.008 mm. During the compensation process, the servo parameters transition smoothly to avoid mechanical vibration. The maximum change rate of the position loop gain is limited to 5 per second, and the change rate of the speed loop gain is limited to 20 per second to ensure the dynamic stability of the compensation process.

[0035] S106. Use the online monitoring technology of solder joint quality to extract the solder joint penetration depth and solder joint diameter, and combine it with the pre-established solder joint quality evaluation model to judge in real time whether the solder joint quality is qualified. If the quality of multiple consecutive solder joints is unqualified, it is determined that the temperature field offset of the welding head affects the solder joint quality.

[0036] The solder joint surface image acquired by the industrial camera is received, and a grayscale image is obtained by grayscale quantization. The edge contour of the solder joint is extracted by the Sobel operator according to the grayscale image, and the sub-pixel coordinates of the solder joint contour are obtained by Gaussian curve fitting to obtain a feature vector including the solder joint diameter value, the penetration value, the grayscale mean, and the contour roundness. According to the feature vector, a deep convolutional neural network is used to identify and calculate the solder joint quality score, and the deep convolutional neural network is composed of three convolutional layers and two fully connected layers to obtain a standardized quality score. For the standardized quality score, a sliding time window is used to scan the score sequence. If the solder joint quality scores in the time window are all lower than the preset threshold, it is determined that the temperature field of the welding head corresponding to the time window has an abnormal offset. According to the feature vector and the quality score data in the temperature field offset time window, a recursive least squares algorithm is used to establish a mapping function from the feature vector to the quality score, and the temperature field offset law is judged from the convergence trend of the mapping function.

[0037] For example, an industrial camera collects 1920x1080 pixel images of the solder joint surface at 0.1 second intervals, generates a 256-level grayscale image through 8-bit grayscale quantization, extracts the solder joint edge contour using the Sobel edge detection operator, obtains the sub-pixel coordinates of the solder joint contour using Gaussian curve fitting, calculates the solder joint diameter, extracts the penetration feature from the grayscale value change curve, and constructs a four-dimensional feature vector including the solder joint diameter value, penetration value, grayscale mean, and contour roundness. The solder joint quality score is calculated for the four-dimensional feature vector, the solder joint diameter threshold interval is set in the range of 0.8 to 1.2 mm, the penetration threshold interval is set in the range of 0.3 to 0.6 mm, and the contour roundness deviation is limited to within 0.05. A deep convolutional neural network is used to identify the feature vector, and the neural network includes 3 convolutional layers and 2 fully connected layers, and outputs a standardized quality score of 0 to 100. The solder joint quality score is recorded into the score sequence at 0.1 second time intervals, and the score sequence is scanned in real time using a sliding time window. The time window length is set to 0.5 seconds and includes 5 consecutive solder joints. If the solder joint quality score in the time window is less than 80 points, it is determined that the temperature field of the welding head corresponding to the time window has an abnormal offset. According to the solder joint feature vector and quality score data in the temperature field offset time window, a recursive least squares algorithm with a forgetting factor of 0.95 is used to establish a mapping function from feature vector to quality score. The function parameters are updated every 0.1 second, and the quantitative effect of temperature field offset on solder joint quality is determined from the convergence trend of the mapping function. A high-resolution industrial camera is used for solder joint appearance inspection. The 1920x1080 resolution provides an image accuracy of 10 microns within a 10 mm x 8 mm field of view, and the 0.1 second sampling interval meets the real-time monitoring requirements of the solder joint forming process. 8-bit grayscale quantization converts the original image into 256 grayscale values. The solder joint area usually occupies a grayscale range of 180 to 220, and the background area is in the grayscale range of 50 to 80, forming a clear grayscale contrast. Sobel edge detection calculates the grayscale gradient in the horizontal and vertical directions respectively. The gradient value at the edge of the solder joint reaches 50 grayscales per pixel, which is much higher than the 5 grayscales per pixel of the background noise. Gaussian curve fitting performs sub-pixel interpolation on the edge contour to achieve an edge positioning accuracy of 0.5 microns. The diameter of a normal solder joint is distributed around 1 mm. The penetration feature is extracted from the half-height width of the grayscale profile curve. The standard penetration is about 0.4 mm, and the contour roundness deviation is less than 0.02. The deep convolutional neural network is trained using 1000 sets of standard solder joint samples. The first convolution layer uses 16 3x3 convolution kernels to extract edge features, the second convolution layer uses 32 3x3 convolution kernels to extract texture features, and the third convolution layer uses 64 3x3 convolution kernels to extract shape features. The fully connected layer converts the feature map into a quality score. The normal solder joint score is above 90 points, minor defects are reduced to 80 to 90 points, and serious defects are below 80 points. The sliding time window monitors the quality trend of the solder joints in real time. The 0.5 second window length covers 5 consecutive solder joints, reflecting the stability of the welding process.When the temperature field shifts, the diameter of the solder joint changes systematically, from the standard 1 mm to 0.85 mm, the penetration depth drops from 0.4 mm to 0.25 mm, and the quality score drops below 75 points. When the average score of 5 consecutive solder joints drops to 70 points, the temperature field anomaly is determined to persist. The recursive least squares algorithm establishes the mapping relationship between the temperature field offset and the solder joint quality through continuous updating, and the forgetting factor of 0.95 ensures that the algorithm is more sensitive to the latest data. In the early stage of the temperature field offset, the quality score drops by 15 points for every 0.1 mm decrease in the solder joint diameter, and the quality score drops by 20 points for every 0.1 mm decrease in the penetration depth. As the offset continues, the rate of change of the quality score gradually slows down, showing the nonlinear response characteristics of the solder joint quality to the temperature field offset. The solder joint quality evaluation model has differentiated response characteristics to different types of defects. Overmelting caused by excessive temperature is manifested as the solder joint diameter increases to 1.3 mm, the penetration depth exceeds 0.7 mm, and the quality score drops to 65 points. A cold weld caused by too low a temperature is characterized by a weld diameter less than 0.7 mm, a penetration depth less than 0.2 mm, and a quality score down to 50. This differentiated response helps identify the specific direction and degree of temperature field deviation.

[0038] S107. Use machine learning algorithms to perform fusion calculations on the temperature field offset of the welding head, welding position offset and solder joint quality, build a multi-parameter coupling model of the welding process, extract key parameters that reflect the temperature field offset law, welding position offset characteristics and solder joint defect patterns, and obtain the influence of the welding head thermal field coupling effect on the welding quality through data association analysis and mechanism deduction, and establish a quantitative relationship between temperature field offset, welding position offset and solder joint quality.

[0039] A feature vector sequence is constructed according to the temperature field offset, welding head displacement and solder joint quality score, and the feature vector sequence obtains a coupling response function through a bidirectional long short-term memory network; principal component dimensionality reduction is performed on the output data of the coupling response function, and the temperature gradient, displacement increment and acceleration component are obtained through the characteristic contribution rate threshold, and the temperature gradient, displacement increment and acceleration component constitute a displacement response curve under the action of thermal-mechanical coupling; the temperature field offset and the welding head displacement are time-series aligned using the displacement response curve, and a three-dimensional quality response surface is constructed through the time-series alignment result and the solder joint quality score, and the three-dimensional quality response surface obtains a mapping function of welding parameters and quality characteristics through cubic spline interpolation; a state transfer matrix is ​​constructed according to the mapping function, and the state transfer matrix is ​​online identified using the recursive least squares method, and the coupled evolution law of the temperature field offset, welding head displacement and solder joint quality is obtained through the online identification result.

[0040] Exemplarily, based on the synchronous data collected by the infrared thermal imager, industrial camera and solder joint detection device, a 12-dimensional feature vector including temperature field offset, welding head displacement and solder joint quality score is constructed, and the feature vector sequence is recorded at a sampling interval of 0.1 seconds. A bidirectional long short-term memory network is used to extract the temporal correlation of the feature sequence. The network contains 2 hidden layers, 128 neurons in each layer, and the training sample length is set to 50 time steps. The coupled response function of temperature field offset and displacement change is obtained from the training results. The output data of the coupled response function is subjected to principal component dimensionality reduction, and the feature contribution rate threshold is set to 0.95. The key variables of temperature field offset and displacement change are extracted, including temperature gradient, displacement increment and acceleration component. The displacement response curve under the action of thermomechanical coupling is constructed, and the quantitative correspondence between temperature field offset and welding head displacement is established through the displacement response curve. The thermomechanical coupling characteristic parameters are obtained from the quantitative correspondence. The thermomechanical coupling characteristic parameters and the solder joint quality score are time-series registered to construct a three-dimensional quality response surface. The coordinate axes of the quality response surface correspond to the temperature field offset, the welding head displacement and the solder joint quality score. The quality response surface is fitted by cubic spline interpolation. The quality change trend caused by the temperature field offset is extracted from the fitting surface, and the mapping function between welding parameters and quality characteristics is established. The fourth-order state transfer matrix is ​​constructed through the mapping function. The state variables include the temperature field offset, the displacement change, the quality score, and the parameter change rate. The state transfer matrix is ​​identified online by the recursive least squares method with a forgetting factor of 0.95. The identification cycle is set to 0.1 second. The coupled evolution law of the temperature field offset, the welding head displacement and the solder joint quality is obtained through the identification results. The multi-parameter coupling characteristics of the welding process are reflected in the composition of the 12-dimensional feature vector. The temperature field offset includes four components: center temperature value, temperature gradient value, temperature change rate, and temperature distribution uniformity. The welding head displacement includes four components: three-dimensional coordinate displacement value, displacement speed value, and posture angle value. The weld quality score includes four components: penetration value, diameter value, strength value, and appearance value. The feature vector is recorded at intervals of 0.1 seconds to form a time series data stream. The bidirectional long short-term memory network extracts time series features through two hidden layers, forward and backward. Each layer of 128 neurons provides sufficient feature expression capabilities. The coupled response of temperature field offset and displacement change shows significant nonlinear characteristics. The principal component dimensionality reduction analysis shows that the cumulative contribution rate of temperature gradient, displacement increment, and acceleration component reaches 96.8%. When the temperature field offset shows a step change of 20 degrees Celsius, the welding head displacement produces a response displacement of 45 microns within 0.2 seconds. The displacement curve shows obvious overshoot characteristics, with the maximum overshoot reaching 12 microns and the stabilization time being 0.8 seconds. The displacement response of the welding head in the horizontal plane is direction-dependent, with a displacement sensitivity of 2.5 microns per degree Celsius in the X direction and 1.8 microns per degree Celsius in the Y direction.The quality response surface reflects the combined effect of temperature field offset and displacement change on the quality of solder joints. The three-dimensional surface shows a significant quality decline trend under the conditions of temperature offset of 20 degrees Celsius and displacement of 45 microns. The penetration depth of the solder joint is reduced from the standard value of 0.4 mm to 0.32 mm, the diameter is reduced from 1.0 mm to 0.85 mm, and the quality score is reduced from 95 points to 78 points. The quality response surface fitted by cubic spline interpolation shows that the weight of the influence of temperature field offset on quality is 0.65, and the weight of the influence of displacement change is 0.35. The fourth-order state transfer matrix describes the dynamic evolution process of temperature field offset, displacement change and quality score, and the matrix elements are updated in real time by recursive least squares method. Observational data show that there is a phase delay of about 0.15 seconds between temperature field offset and displacement change, and the response delay of quality score reaches 0.25 seconds. The eigenvalue distribution of the state transfer matrix reflects the stability of the system, and the modulus of the dominant eigenvalue is 0.92, indicating that the system has good convergence characteristics. The forgetting factor of 0.95 ensures the rapid tracking capability of parameter identification for changes in working conditions, and the root mean square error of the identification result is controlled within 3%. The coupled evolution law reveals the transmission mechanism by which the temperature field offset causes the welding head displacement through the thermomechanical coupling effect, which in turn leads to the degradation of the solder joint quality. In the steady-state stage of the welding process, the temperature field fluctuates within the range of plus or minus 5 degrees Celsius, the corresponding displacement fluctuation is less than 10 microns, and the quality score is maintained above 90 points. When the temperature field shows a continuous offset, the system enters an unstable state, the displacement increases exponentially, and the quality score decreases approximately linearly.

[0041] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. An array detection and positioning method combined with multi-welding head battery welding, characterized in that: The method comprises: The arrangement spacing and material properties of the welding head array are obtained, and a mathematical model of heat conduction of the welding head array is established based on the arrangement spacing and material properties of the welding head array. The temperature field distribution of the welding heads under different arrangement spacings is simulated and calculated by the finite element analysis method to determine the influence of the welding head spacing on heat conduction. Infrared thermal imaging technology is used to monitor the temperature field distribution of the welding head in real time, and the preset temperature threshold is combined to determine whether the temperature field of the welding head is offset. If it is detected that the temperature of a welding head exceeds the threshold range, it is determined that the temperature field of the welding head is offset; After determining that the temperature field has shifted, the temperature field offset is calculated by the temperature difference before and after the welding head temperature field offset, and the thermal expansion coefficient of the welding head base material is obtained. Combined with the temperature field offset, the thermal expansion deformation of the welding head base under the condition of temperature field offset is calculated, and the displacement deformation of the welding head base is detected in real time through machine vision measurement technology; The actual position of the welding head is obtained, and the thermal expansion deformation and temperature field offset are combined and correlated. Through theoretical derivation and experimental verification, a ternary mathematical model of welding position offset, thermal expansion deformation and temperature field offset is constructed to calculate the welding position offset of each welding head caused by thermal effect in real time; According to the offset of welding position, the motion control parameters of digital welding system are adaptively adjusted, and the welding head offset error is compensated in real time through closed-loop feedback control algorithm; Adopt the online monitoring technology of solder joint quality, extract the solder joint penetration depth and solder joint diameter, and combine with the pre-established solder joint quality evaluation model to judge whether the solder joint quality is qualified in real time. If the quality of multiple consecutive solder joints is unqualified, it is determined that the temperature field offset of the welding head affects the quality of the solder joint; A machine learning algorithm is used to perform fusion calculation of the welding head temperature field offset, welding position offset and solder joint quality, and a multi-parameter coupling model of the welding process is constructed. The key parameters reflecting the temperature field offset law, welding position offset characteristics and solder joint defect mode are extracted. Through data association analysis and mechanism deduction, the influence of the welding head thermal field coupling effect on the welding quality is obtained, and the quantitative relationship between temperature field offset, welding position offset and solder joint quality is established.

2. The method according to claim 1, characterized in that: The method obtains the arrangement spacing and material properties of the welding head array, establishes a mathematical model of heat conduction of the welding head array based on the arrangement spacing and material properties of the welding head array, simulates and calculates the temperature field distribution of the welding heads under different arrangement spacings through the finite element analysis method, and determines the influence of the welding head spacing on heat conduction, including: According to the arrangement of the welding head array, the horizontal and vertical spacing values ​​of the welding head array are obtained, and the initial value of the surface temperature field of the welding head array is collected by an infrared thermal imager to obtain the temperature distribution function; Extracting the thermal conductivity of the welding head material according to the initial value of the temperature field, and establishing a heat diffusion function using the thermal conductivity and thermal impedance, wherein the heat diffusion function includes a temperature diffusion coefficient; The heat conduction calculation area of ​​the welding head array is divided into square grid units, and the heat conduction differential equation in the grid unit is solved by using the fourth-order Runge-Kutta algorithm to obtain a temperature-time variation function; The temperature distribution function is fitted with the temperature-time variation function, and the corresponding relationship between the welding head spacing and the heat diffusion amount is calculated by using the linear least square method to obtain the spatial distribution law of the temperature field.

3. The method according to claim 1, characterized in that The infrared thermal imaging technology is used to monitor the temperature field distribution of the welding head in real time, and a preset temperature threshold is combined to determine whether the temperature field of the welding head is offset. If a welding head temperature is detected to exceed the threshold range, it is determined that the temperature field of the welding head is offset, including: The temperature field image of the welding head array surface is collected by an infrared thermal imager, and the temperature field image is processed by Gaussian filtering to obtain a temperature field filtered image; Acquire the temperature color scale value of the pixel point in the welding head area from the temperature field filter image, and calculate the temperature gradient value according to the temperature color scale value to obtain the temperature time series data; Calculate the deviation between the temperature value of the welding head area and the reference temperature according to the temperature time series data, and record the welding head position coordinates and the temperature deviation value when the temperature deviation exceeds a preset range; The temperature deviation value at the welding head position is accumulated. If the accumulated temperature deviation value exceeds the preset temperature fluctuation threshold, it is determined that the welding head temperature field has an abnormal deviation, and the welding head center point coordinates and the accumulated temperature deviation value are obtained as temperature anomaly records.

4. The method according to claim 1, characterized in that After determining that the temperature field is offset, the temperature field offset is calculated by the temperature difference before and after the welding head temperature field is offset, the thermal expansion coefficient of the welding head base material is obtained, and the thermal expansion deformation of the welding head base under the condition of temperature field offset is calculated in combination with the temperature field offset, and the displacement deformation of the welding head base is detected in real time by machine vision measurement technology, including: The temperature field value of the welding head is collected by the infrared thermal imager, and the temperature difference is calculated by the temperature value of the central area of ​​the welding head in the time window before and after the temperature field abnormality judgment moment, so as to obtain the temperature field offset of the welding head base; The theoretical thermal expansion displacement is calculated based on the three-dimensional size parameters of the base using the linear thermal expansion coefficient value in the physical property database of the copper material of the welding head base and the temperature field offset; Extracting the coordinates of corner feature marks from the base surface image captured by the industrial camera, tracking the feature mark coordinates through a sub-pixel edge detection algorithm, and obtaining the actual deformation of the base surface; The Kalman filter algorithm is used to filter the theoretical displacement of the thermal expansion of the base and the actual deformation to obtain the three-dimensional spatial deformation vector of the base. If the deformation vector exceeds the positive and negative threshold range, it is determined that the thermal strain of the base is abnormal.

5. The method according to claim 1, characterized in that The actual position of the welding head is obtained, and the thermal expansion deformation and the temperature field offset are combined, and a correlation analysis is performed. Through theoretical derivation and experimental verification, a ternary mathematical model of welding position offset, thermal expansion deformation and temperature field offset is constructed, and the welding position offset of each welding head caused by the thermal effect is calculated in real time, including: Collecting a welding head surface image using an industrial camera, extracting a calibration pattern center coordinate from the welding head surface image, and using the center coordinate to calibrate camera parameters to obtain a welding head position deviation vector; Obtain temperature field data of the welding head area with an infrared thermal imager, calculate a temperature deviation value according to the temperature field data, and obtain a thermal expansion displacement vector by combining the temperature deviation value with a thermal expansion coefficient; Building a Bayesian network model according to the welding head position deviation vector and the thermal expansion displacement vector, and extracting a position compensation vector from the Bayesian network model; A recursive filtering operation is performed on the position compensation vector, a temperature compensation value is obtained through the recursive filtering operation, and a three-dimensional space compensation function is constructed according to the temperature compensation value.

6. The method according to claim 1, characterized in that The method of adaptively adjusting the motion control parameters of the digital welding system according to the welding position offset and compensating the welding head offset error in real time through a closed-loop feedback control algorithm includes: According to the numerical comparison between the welding head position offset and the preset compensation threshold, the compensation gain coefficient is calculated by using the particle swarm optimization algorithm; A three-dimensional space compensation vector is constructed by multiplying the compensation gain coefficient by the position offset, and a compensation response curve containing a position component, a velocity component and an acceleration component is generated from the compensation vector; Extract the compensation amount from the compensation response curve to construct a third-order state space equation, and use a Kalman state observer to iteratively calculate the state space equation to obtain a compensation control amount; The position loop gain parameters, speed loop gain parameters and current loop gain parameters are updated according to the compensation control amount, the actual compensation displacement is obtained by adjusting the gain parameters, the actual compensation displacement is numerically compared with the theoretical compensation amount to obtain the compensation tracking error, and a proportional-integral-differential regulator is used to perform closed-loop regulation on the compensation tracking error.

7. The method according to claim 1, characterized in that The online monitoring technology for solder joint quality is used to extract solder joint penetration depth and solder joint diameter, and combined with the pre-established solder joint quality evaluation model, to judge in real time whether the solder joint quality is qualified. If the quality of multiple consecutive solder joints is unqualified, it is determined that the temperature field offset of the welding head affects the solder joint quality, including: Receive a solder joint surface image captured by an industrial camera, obtain a grayscale image through grayscale quantization, extract the solder joint edge contour using a Sobel operator based on the grayscale image, obtain the sub-pixel coordinates of the solder joint contour using Gaussian curve fitting, and obtain a feature vector including a solder joint diameter value, a penetration value, a grayscale mean, and a contour roundness; According to the feature vector, a deep convolutional neural network is used to identify and calculate the quality score of the solder joint, wherein the deep convolutional neural network is composed of three convolutional layers and two fully connected layers to obtain a standardized quality score; For the standardized quality score, a sliding time window is used to scan the score sequence. If the quality scores of the solder joints in the time window are all lower than a preset threshold, it is determined that the temperature field of the soldering head corresponding to the time window is abnormally offset; According to the characteristic vector and quality score data in the temperature field offset time window, a recursive least squares algorithm is used to establish a mapping function from the characteristic vector to the quality score, and the temperature field offset law is determined from the convergence trend of the mapping function.

8. The method according to claim 1, characterized in that The machine learning algorithm is used to perform fusion calculation on the welding head temperature field offset, welding position offset and solder joint quality, a multi-parameter coupling model of the welding process is constructed, and key parameters reflecting the temperature field offset law, welding position offset characteristics and solder joint defect mode are extracted. Through data association analysis and mechanism deduction, the influence law of the welding head thermal field coupling effect on the welding quality is obtained, and the quantitative relationship between the temperature field offset, welding position offset and solder joint quality is established, including: Constructing a feature vector sequence according to the temperature field offset, the welding head displacement and the welding point quality score, and obtaining a coupling response function through a bidirectional long short-term memory network. Performing principal component dimensionality reduction on the output data of the coupling response function, obtaining temperature gradient, displacement increment and acceleration component through a characteristic contribution rate threshold, wherein the temperature gradient, displacement increment and acceleration component constitute a displacement response curve under the action of thermal-mechanical coupling; The displacement response curve is used to perform time series registration on the temperature field offset and the welding head displacement, and a three-dimensional quality response surface is constructed through the time series registration result and the weld quality score. The three-dimensional quality response surface obtains the mapping function of welding parameters and quality characteristics through cubic spline interpolation; A state transfer matrix is ​​constructed according to the mapping function. The state transfer matrix is ​​identified online using a recursive least squares method. The coupled evolution law of the temperature field offset, the welding head displacement and the welding spot quality is obtained through the online identification result.

Images