A control method for a welded coupling device

Abstract

The application provides a control method of a welding coupling device, comprising: comparing the extracted contour and center line data according to the pre-set welding spot size and fusion width threshold to determine whether the welding spot has size abnormality or fusion width abnormality; if there is abnormality, an image segmentation method is used to identify the abnormal area in the image, the area and geometric features of the abnormal area are calculated through a pixel statistical method to obtain the size and deformation degree of the welding spot; welding instruction data after correction is obtained, and the energy, focal length and welding torch angle parameters in the instruction are extracted as the reference value of welding instruction analysis; whether the welding energy and focal length match the reference value of welding instruction analysis is determined by analyzing the energy distribution curve in the welding process and combining the real-time collected focal length data and welding torch angle change; if the welding energy and focal length do not match the reference value of welding instruction analysis, the control parameters of the welding device are adjusted according to the deviation value, and the correction amount of the welding torch path is recalculated.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to a control method for welding coupling equipment. Background Art

[0002] In the field of precision welding, solder joint quality is a critical factor in determining product performance and reliability. With the increasing demand for welding precision in industries such as electronics and medical, traditional welding quality inspection methods are no longer able to meet these demands. For a long time, the industry has relied primarily on manual visual inspection or random destructive testing to assess solder joint quality. This approach is not only inefficient and labor-intensive, but also fails to provide comprehensive, real-time inspection of every solder joint, making it difficult to detect potential quality issues in a timely manner. To improve welding quality control, efforts have been made to introduce automated inspection technologies. However, accurately acquiring geometric features of solder joints without contacting them presents numerous challenges, such as their small size and complex shape, as well as interference from spatter and smoke generated during the welding process. Even when image data of solder joints can be acquired, accurately extracting key parameters such as their outline, size, and weld width poses challenges such as complex algorithms and poor robustness. Furthermore, subtle deformation and positional deviations in solder joints are often difficult to detect visually, making traditional methods difficult to automatically identify and quantitatively analyze these minor defects. More importantly, during the welding process, parameters such as energy input, focal length adjustment, and welding gun posture all have a significant impact on the quality of the weld. Traditional technologies have difficulty monitoring changes in these parameters in real time, and are even more unable to establish an effective correlation model between these parameters and weld quality, resulting in a lack of scientific basis for welding process control. Therefore, how to automatically adjust welding parameters in a timely manner when abnormalities are detected to ensure stable welding quality, and how to quickly and accurately evaluate the quality of the weld after welding is completed to provide data support for subsequent process optimization, remain key issues that need to be urgently addressed in the field of precision welding. Summary of the Invention

[0003] The present invention provides a control method for a welding coupling device, which mainly includes:

[0004] A high-speed camera is used to capture continuous image sequences of the welding instruction execution process, and an edge detection algorithm is used to extract the contour information of the weld point. The center line of the weld point is generated by combining the morphological processing method.

[0005] Based on the pre-set weld spot size and weld width thresholds, the extracted contour and centerline data are compared to determine whether the weld spot has size anomalies or weld width anomalies.

[0006] If there is an abnormality, the image segmentation method is used to identify the abnormal area in the image, and the area and geometric characteristics of the abnormal area are calculated through pixel statistics to obtain the size and deformation degree of the solder joint;

[0007] Based on the preset welding gun path parameters and the extracted weld centerline coordinates, the deviation between the weld centerline and the preset path is calculated to obtain the weld position offset. The welding instructions are then corrected based on the weld size and deformation.

[0008] Obtain the corrected welding instruction data and extract the energy, focal length and welding gun angle parameters in the instruction as the reference values ​​for welding instruction analysis;

[0009] By analyzing the energy distribution curve during welding, combined with real-time collected focal length data and welding gun angle changes, it is determined whether the welding energy and focal length match the baseline values ​​analyzed in the welding instruction;

[0010] If the welding energy and focal length do not match the reference values ​​analyzed by the welding instruction, the control parameters of the welding equipment are adjusted according to the deviation values ​​and the correction amount of the welding gun path is recalculated;

[0011] The adjusted control parameters are input into the welding equipment control system to monitor the energy and focal length changes during the welding process in real time. By continuously collecting welding process images, the stability of the weld point contour and center line is verified to determine whether the welding quality meets the preset standards.

[0012] Furthermore, the method captures a continuous image sequence of the welding instruction execution process using a high-speed camera, extracts the contour information of the weld spot using an edge detection algorithm, and generates the centerline of the weld spot using a morphological processing method. The method includes: capturing a first image sequence of the welding head motion trajectory during the execution of the welding instruction using an industrial-grade high-speed camera at a preset acquisition frequency; extracting first grayscale image data from the first image sequence; obtaining an optimal grayscale threshold based on image histogram statistics; binarizing the first grayscale image data to obtain first binary image data; performing denoising on the first binary image data using a Gaussian filter to obtain second binary image data; performing edge detection on the second binary image data using a Sobel operator to obtain first edge contour data; calculating the contour pixel density distribution based on the first edge contour data; performing breakpoint connection processing on the edge contour using a density clustering method to obtain second edge contour data; and performing a morphological closing operation on the second edge contour data to obtain closed contour line data. Based on the closed contour line data, a morphological thinning algorithm based on a distance transform is used to extract a skeleton to obtain first skeleton data; and performing width normalization on the first skeleton data based on a preset skeleton width threshold to obtain second skeleton data. The second skeleton data is processed using a cubic spline interpolation algorithm to achieve node continuity, generating the first centerline data. This data is then smoothed using a preset curvature threshold to generate the second centerline data. A morphological erosion operator is used to refine the second centerline data to obtain the weld point contour centerline data. The centerline data integrity is verified by testing the centerline pixel connectivity.

[0013] Furthermore, the method of comparing the extracted contour and centerline data based on pre-set weld size and weld width thresholds to determine whether the weld has a size anomaly or a weld width anomaly includes: reading standard weld size parameters from a preset parameter database, performing feature extraction on the first weld contour data to obtain actual weld size parameters, and using a curve matching method to calculate the deviation between the actual size and the standard size to obtain first abnormal feature data. Using a region growing algorithm to extract the weld width feature region from the first weld contour data to obtain first weld width data, using a distance transformation method to perform statistics on the first weld width data to obtain a weld width mean, reading standard weld width range parameters from the preset parameter database, and calculating second abnormal feature data based on the weld width mean and the standard range. Using an equidistant sampling method to sample the first weld contour data to obtain a first sampling point sequence, calculating the local curvature of the first sampling point sequence to obtain first curvature distribution data, and calculating the roundness parameter by least squares fitting to obtain third abnormal feature data. A second sampling point sequence is obtained by extracting local extreme points from the first curvature distribution data. The spatial distribution period of the second sampling point sequence is calculated to obtain first periodic data. The spectral components of the first periodic data are calculated using a fast Fourier transform to obtain fourth abnormal feature data. The first to fourth abnormal feature data are combined to form a comprehensive abnormal feature vector for the weld. An abnormality discrimination threshold is read from a preset parameter database. A multi-dimensional threshold comparison is performed based on the comprehensive abnormal feature vector and the abnormality discrimination threshold to obtain a weld abnormality type identifier. The weld abnormality type identifier and the corresponding abnormal feature data are written into an abnormality detection result database. A weld image index is established in the abnormality detection result database, and the corresponding first weld contour data is stored in association therewith.

[0014] Further, if there is an anomaly, an image segmentation method is used to identify the abnormal area in the image, and the area and geometric features of the abnormal area are calculated by pixel statistics method to obtain the size and deformation degree of the welding spot, including: extracting the welding spot image to be processed from the image database, regionally marking the welding spot image according to the preset abnormal judgment result to obtain a first abnormal mark image, and extracting the boundary of the first abnormal mark image by using a watershed segmentation algorithm to obtain first region boundary data. Curve fitting is performed on the first region boundary data to obtain a first abnormal contour curve, and a chain code encoding method is used to extract a sequence of contour feature points to obtain first feature point data, and the perimeter and area of the abnormal area are calculated according to the first feature point data to obtain first geometric parameter data. The first shape feature data is obtained according to the first geometric parameter data and the principal axis direction of the center of gravity of the abnormal area. The gray scale histogram data of the abnormal area in the first abnormal mark image is obtained by gray scale statistics, and the second abnormal area image is obtained by multi-level threshold segmentation according to the first gray scale histogram data. The first density distribution map is obtained by labeling the segmented region by using a region growing method. The second shape feature data is obtained by extracting the region centroid position and boundary point coordinates from the first density distribution map. The first size data is obtained by calculating the long axis and short axis parameters of the abnormal area by using the minimum circumscribed rectangle method. The symmetry parameter of the second shape feature data is calculated by using the region projection method to obtain the first deformation parameter. The first geometric parameter data, the first size data and the first deformation parameter are combined to form an abnormal area feature vector. The abnormal area feature vector and the corresponding first abnormal mark image are indexed to write into the abnormal feature database to complete the abnormal area feature storage.

[0015] Furthermore, the method of calculating the deviation between the weld centerline and the preset path based on preset welding gun path parameters and the extracted weld centerline coordinates to obtain the weld position offset, and then correcting the welding instructions based on the weld size and deformation, includes: reading a preset welding path coordinate sequence from a path parameter database, generating a first path curve using a cubic spline interpolation method, curve fitting the weld centerline coordinate data to obtain a second path curve, calculating the distance between corresponding points of the two curves using a least squares method to obtain first deviation data; performing interval sampling on the first deviation data to calculate the local deviation mean to obtain first offset data; smoothing the first offset data using a sliding window method to obtain second offset data; and segmenting the second offset data according to a preset deviation threshold to obtain path offset interval data; reading weld size anomaly data and deformation degree data from a quality inspection database, calculating a coordinate correction coefficient according to a preset compensation rule to obtain a first correction parameter, and normalizing the first correction parameter to obtain a second correction parameter; calculating a path compensation value based on the path offset interval data and the second correction parameter to obtain first path compensation data, and generating a path compensation curve using a segmented interpolation method to obtain second path compensation data. The first instruction parameter is obtained by reading standard welding parameters from the welding process database. The second instruction parameter is generated by generating a position correction value based on the second path compensation data. The first and second instruction parameters are combined to obtain a corrected instruction sequence. The feasibility of the corrected instruction sequence is verified using a kinematic constraint verification method to obtain a qualified instruction sequence. The qualified instruction sequence is written into the welding control database, and an instruction index record is created to associate and store the corresponding path compensation data and correction parameters.

[0016] Furthermore, the method of obtaining and executing the corrected welding instruction data and extracting the energy, focal length, and welding gun angle parameters from the instructions as reference values ​​for welding instruction analysis includes: reading the corrected welding instruction sequence from an instruction database, obtaining the instruction timestamp and instruction type identifier using an instruction parsing method to obtain first instruction data, classifying and filtering the first instruction data based on the instruction type identifier to obtain a welding parameter instruction set; sorting the welding parameter instruction set by timestamp to obtain first parameter time series data; separating the energy parameter, focal length parameter, and angle parameter using a parameter extraction algorithm to obtain a first parameter array; resampling the first parameter array using a fixed time window to obtain a second parameter array; normalizing the second parameter array based on parameter units and ranges to obtain a first standard parameter sequence; calculating the sliding mean and standard deviation of the first standard parameter sequence to obtain first statistical parameter data; extracting the parameter fluctuation amplitude and frequency characteristics using a wavelet decomposition method to obtain first fluctuation characteristic data; calculating the parameter stability interval based on the first statistical parameter data and the first fluctuation characteristic data to obtain first interval data; and segmenting the first interval data using an adaptive threshold method to obtain second interval data. The standard parameter range is read from the process parameter database to obtain parameter limit data. The first eigenvector is calculated based on the second interval data to obtain the characteristics of each parameter interval. A weighted fusion of the first eigenvector and the parameter limit data is performed to obtain a baseline parameter vector. This baseline parameter vector is reconstructed according to the instruction format to obtain the welding instruction baseline value. The welding instruction baseline value is written into the instruction baseline database, and a baseline value index record is created to associate and store the corresponding statistical parameter data with the fluctuation characteristic data.

[0017] Furthermore, the method analyzes the energy distribution curve during the welding process, combines real-time acquired focal length data and welding gun angle changes, and determines whether the welding energy and focal length match the baseline values ​​analyzed in the welding instruction. The method includes: reading real-time energy sampling data during the welding process from a sensor database to obtain first raw data; validating the first raw data based on the sampling timestamp to obtain first valid data; and smoothing the first valid data using a sliding average method to obtain first curve data. Using welding equipment sensors to collect focal length and angle data to obtain second raw data; eliminating outliers from the second raw data using data screening rules to obtain second valid data; and performing a piecewise least squares fit on the second valid data to obtain second curve data. Reading welding benchmark parameters from a benchmark database to obtain a benchmark parameter vector; normalizing the three sets of parameters (energy, focal length, and angle) in the benchmark parameter vector to obtain standard benchmark data; and calculating the parameter fluctuation range based on the standard benchmark data to obtain parameter threshold data. Time-series alignment of the first and second curve data using a fixed time window to obtain a first parameter sequence; and unifying the sampling frequency using a data resampling method to obtain a second parameter sequence. The curve characteristic values ​​of the second parameter sequence are calculated to obtain characteristic parameter data. The matching degree between the characteristic parameter data and the standard reference data is calculated using a curve correlation algorithm to obtain parameter similarity data. The parameter similarity data is graded and determined based on parameter threshold data to obtain a parameter matching result. The determination rules are read from the process database to obtain rule data. A multi-dimensional threshold determination method is used to obtain a parameter matching identifier. The parameter matching identifier and the corresponding parameter similarity data are written into a result database, and a result index record is created to associate and store the corresponding characteristic parameter data with the determination rule data.

[0018] Furthermore, if the welding energy and focal length do not match the reference values ​​analyzed in the welding instruction, the control parameters of the welding equipment are adjusted based on the deviation values, and the correction value of the welding gun path is recalculated. This includes: reading welding parameter deviation data from a parameter analysis database, calculating the energy deviation value and the focal length deviation value using an adaptive mapping algorithm to obtain first deviation data, generating a compensation coefficient based on welding process rules to obtain a first compensation parameter, and numerically optimizing the first compensation parameter to obtain a corrected control parameter. A binocular vision sensor is used to capture a welding position image to obtain first position data, reading standard position parameters of the molten pool from a process database to obtain reference position data, and calculating the position difference using an image registration method to obtain first offset data. The first offset data is converted to a welding gun coordinate system using a coordinate transformation matrix to obtain second offset data, calculating the welding gun position compensation coefficient based on the corrected control parameters to obtain a first compensation vector, and normalizing the first compensation vector to obtain a welding gun position compensation value. Path adjustment parameters are calculated based on the welding gun position compensation value to obtain third offset data, smoothing the third offset data using a cubic spline interpolation method to obtain a first path correction value, and coordinate calibration of the first path correction value to obtain a second path correction value. The Jacobian matrix method is used to map the second path correction value to the joint space to obtain the first path compensation data. Trajectory optimization is performed based on the manipulator's kinematic constraints to obtain the second path compensation data. The second path compensation data is verified for reachability to obtain a path verification result. The compensation trajectory is verified using motion simulation to obtain trajectory verification data. Based on the verification result, compensation instructions are generated to obtain qualified path data. The qualified path data is written into the control parameter database, and a path index record is created to associate and store the corresponding deviation data with the compensation parameter data.

[0019] Furthermore, the method includes inputting the adjusted control parameters into the welding equipment control system, monitoring the energy and focal length changes during the welding process in real time, and continuously capturing images of the welding process to verify the stability of the weld contour and centerline, thereby determining whether the welding quality meets the preset standard. The method includes: writing the corrected welding control parameters into the welding equipment controller to obtain a first control instruction; collecting energy and focal length data during the welding process from the equipment sensor to obtain first monitoring data; generating second monitoring data through data validity verification; and calculating parameter change trends using a sliding average method to obtain first trend data. A high-speed industrial camera is used to continuously capture weld formation images to obtain a first image sequence; valid images are selected based on image quality evaluation indicators to obtain a second image sequence; and the second image sequence is contrast-enhanced using a histogram equalization method to obtain a third image sequence. A watershed segmentation algorithm is used to extract weld contour data from the third image sequence to obtain a first contour sequence; weld centerline data is calculated using a least squares method to obtain a first centerline sequence; and contour shape parameters and centerline offset parameters are calculated using a continuity detection method to obtain first stability data. A time-domain analysis of the first trend data yields a parameter fluctuation index, while a statistical analysis of the first stability data yields a shape stability index. These two sets of indicators are combined to generate a quality feature vector, yielding first feature data. Preset quality parameters are read from a quality standard database to obtain standard feature data. The feature matching degree is calculated using a multidimensional threshold comparison method to obtain first matching data. The welding quality is then determined based on the matching rules to obtain a quality determination result. Parameter adjustment recommendations are generated based on the quality determination result to obtain first adjustment data. A feedback compensation algorithm is used to calculate control parameter corrections to obtain second adjustment data. The quality determination result and second adjustment data are written to a quality monitoring database, where a quality record index is established and the corresponding monitoring data and feature data are stored in an associated manner.

[0020] The technical solution provided by the embodiment of the present invention may have the following beneficial effects:

[0021] The present invention discloses a control method for a welding coupling device. A high-speed camera is used to capture images of the welding process, and edge detection and morphological processing are used to extract the weld contour and center line. Welding spot anomalies are determined based on a preset threshold, and image segmentation is used to identify abnormal areas and calculate their features. Combined with welding gun path parameters, the weld position offset is calculated and the welding instructions are corrected. By analyzing the energy distribution curve, focal length, and welding gun angle changes, it is determined whether they match the reference value. If not, the control parameters are adjusted and the correction amount is recalculated. Finally, the welding process is monitored in real time, the stability of the weld contour and center line is verified, and the welding quality is determined. The present invention realizes automatic monitoring, anomaly detection, and real-time correction of the welding process, thereby improving welding quality and efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1The figure is a flow chart of a control method of a welding coupling device according to the present invention.

[0023] Figure 2 Schematic diagram of a control method for a welding coupling device according to the present invention. DETAILED DESCRIPTION

[0024] The following will describe the technical solutions in the embodiments of the present invention in detail with reference to the accompanying drawings. The described embodiments are only a part of the embodiments of the present invention.

[0025] like Figure 1-2 In this embodiment, a control method for a welding coupling device may specifically include:

[0026] S101. Capture a continuous image sequence during the execution of welding instructions through a high-speed camera device, extract the contour features of the weld using edge detection technology, and generate the centerline data of the weld in combination with a morphological algorithm. Specifically, the image sequence of the welding head motion trajectory is captured at a preset frequency, and initial contour data is obtained using grayscale processing and binarization methods. After noise reduction and edge detection, a complete contour line is generated, and then the centerline data is generated through a refinement algorithm.

[0027] In this embodiment of the present invention, after the welding equipment is started, a high-speed camera captures images of the welding process at a preset frequency to capture key features of weld formation. The captured image sequence undergoes multiple processing steps to generate the weld centerline. This article does not specify a specific hardware model; appropriate camera equipment can be selected based on the actual scenario.

[0028] S1011. An initial image sequence is captured of the welding joint motion trajectory using an industrial-grade high-speed camera device at a preset acquisition frequency, grayscale image data is extracted from the initial image sequence, an optimal binarization threshold is determined using grayscale histogram analysis, and the grayscale image data is binarized to generate initial binarized image data.

[0029] In this embodiment of the present invention, the acquisition frequency can be set to 1000 to 5000 frames per second to ensure that the dynamics of the melt pool and spatter details are recorded. Grayscale images are stored in 8-bit format, with a grayscale value range of 0 to 255. Using histogram statistics, a threshold with the maximum inter-class variance is selected, typically between 120 and 180, to distinguish the melt pool from the background area.

[0030] S1012. Use a Gaussian filter to perform noise reduction processing on the initial binary image data to generate noise-reduced binary image data, perform edge detection on the noise-reduced binary image data using a Sobel operator to generate initial edge contour data, calculate the pixel density distribution based on the initial edge contour data, and perform breakpoint connection processing to generate complete edge contour data.

[0031] In the embodiment of the present application, the Gaussian filter kernel size can be selected as 5x5 or 7x7, and the standard deviation is set to 1.5-2.0 to smooth the noise and retain the edge features. The Sobel operator calculates the gradient field through horizontal and vertical templates, and the gradient threshold is set to 30 to mark the edge points. When connecting the breakpoints by density clustering, the clustering radius is 10 pixels, and the minimum density is 5 pixels to ensure the continuity of the contour.

[0032] S1013, the complete edge contour data is subjected to morphological closing operation to generate closed contour line data, a morphological thinning algorithm based on distance transformation is used to extract the skeleton of the closed contour line data to generate initial skeleton data, the initial skeleton data is subjected to node continuous processing by a cubic spline interpolation algorithm, and finally the welding spot center line data is generated by combining curvature adjustment.

[0033] In the embodiment of the present application, the closing operation uses a circular structural element with a radius of 3 pixels to fill the gap. The distance transformation threshold is set to one-third of the contour width to extract the skeleton. The cubic spline interpolation control point spacing is 20 pixels, and the curvature threshold is 0.15 for smoothing processing. Finally, the center line width is thinned to 1 pixel by the erosion operator, and its connectivity is verified, with a maximum breakpoint gap of not more than 2 pixels.

[0034] In the embodiment of the present application, the welding head movement speed is usually 50-200 mm / s, and the image frame displacement is less than 3 pixels, which ensures the sequence continuity. The extracted center line has a sub-pixel level accuracy, and the average width is standardized to 1 pixel, which meets the requirements of welding spot trajectory analysis. The Sobel operator can also provide gradient direction for judging the fusion width distribution, improving the comprehensiveness of the analysis.

[0035] It can be understood that the specific values of the parameters in this step can be adjusted by the technician according to the actual application scene to optimize the image processing effect. Through multi-level image processing, the accuracy of the welding spot contour and center line data is ensured, laying a foundation for subsequent quality judgment.

[0036] S102, according to the pre-set welding spot size and fusion width standard value, the welding spot contour and center line data extracted by image processing are compared to judge whether the welding spot has size abnormality or fusion width abnormality. Specifically, the reference data is read from the standard parameter library, the actual welding spot features are extracted and the deviation is calculated, the abnormal type identification is generated by combining multi-dimensional feature analysis, and the results are stored to support subsequent optimization.

[0037] In the embodiment of the present application, this step aims to identify potential defects of the welding spot by comparing the standard and actual parameters. The standard parameters are generated based on a large number of sample statistics, and the actual features are extracted by image analysis. After comparison, a multi-dimensional feature vector is formed for abnormality determination. This step does not limit the performance of the specific computing device, and appropriate hardware can be selected according to actual needs.

[0038] S1021. In an embodiment of the present invention, the reference data of the solder joint size is read from a preset standard parameter library, the collected solder joint contour data is processed using contour feature extraction technology to generate actual size parameters, and the deviation value between the actual size and the standard size is calculated using a curve matching method and used as preliminary feature data of the size anomaly.

[0039] In an embodiment of the present invention, a standard parameter library stores reference values ​​for weld diameter and depth, with diameters typically ranging from 3.0 mm to 5.0 mm and depths from 1.0 mm to 2.0 mm. Contour data is represented by 512 two-dimensional coordinate points. The actual contour is matched to the standard contour using the least squares method, and the deviation is calculated. A deviation exceeding 0.3 mm is considered a preliminary dimensional anomaly. The curve matching process ensures accuracy through iterative optimization, and deviation calculations consider the overall contour distribution rather than single, local anomalies.

[0040] S1022. In an embodiment of the present invention, a regional growing algorithm is used to extract the weld width characteristic region from the weld point contour data to generate initial weld width data, a distance transformation method is used to perform statistical analysis on the initial weld width data to calculate the average weld width value, a weld width reference range is read from a standard parameter library, and compared with the average weld width value to generate characteristic data of weld width abnormality.

[0041] The region growing algorithm uses the weld center as the seed point and expands based on grayscale similarity, with a threshold of 20 to ensure accurate weld width segmentation. A distance transform samples 32 points along the circumference and calculates the average weld width. The normal range is 2.5 mm to 3.5 mm, with a standard deviation of less than 0.2 mm. If the average value is out of range or the deviation is excessive, abnormal weld width feature data is generated. Uniform sampling improves statistical reliability and avoids local noise interference.

[0042] S1023. In an embodiment of the present invention, an equidistant sampling method is used to extract a point sequence from the weld point contour data to generate sampling point data. The local curvature distribution of the sampling point data is calculated by a three-point method and the roundness parameters are fitted to generate shape anomaly feature data. At the same time, the curvature extreme points are extracted and the periodic distribution characteristics are calculated to generate edge anomaly feature data.

[0043] The sampling interval was set to 0.1 mm, with a 0.3 mm spacing between each three adjacent points. Curvature was calculated to reflect contour smoothness. A least-squares method was used to fit a circle and calculate the root mean square deviation. The normal roundness deviation was less than 0.15 mm. Curvature extremes were extracted using a 1.0 mm sliding window. Fast Fourier transform analysis of the periodic distribution revealed that the primary frequency of a normal solder joint was less than 8 waves per revolution, with an amplitude ratio less than 0.3. High-frequency fluctuations indicate edge irregularities or cracks. These two characteristics together provide the basis for anomaly analysis.

[0044] In the embodiment of the present application, the size abnormal feature data, the fusion width abnormal feature data, the shape abnormal feature data and the edge abnormal feature data are integrated to form a comprehensive abnormal feature vector of the welding spot, a multi-dimensional abnormal discrimination threshold value is read from a standard parameter library, the type of the welding spot abnormality is judged by a multi-dimensional threshold value comparison method and an identification is generated, and the type of the abnormality and the related feature data are stored in a detection result database and are associated with the original contour data.

[0045] The comprehensive feature vector contains six-dimensional data: maximum size deviation, average fusion width deviation, fusion width standard deviation, roundness deviation, cycle number and maximum ripple amplitude. The multi-dimensional threshold value comparison checks each item, and any dimension exceeding the range is determined as the corresponding abnormal type, such as size deviation, insufficient fusion width, irregular shape or edge ripple. The detection result database records the feature values and image indexes, and the response time is controlled within 50 milliseconds, supporting real-time monitoring and tracing. The storage process optimizes the retrieval efficiency through hash indexing, ensuring data correlation.

[0046] In the embodiment of the present application, this step improves the detection accuracy through multi-dimensional analysis. The diversity of abnormal types reflects different causes of welding spot defects, such as process parameter adjustment or equipment jitter. The automation of feature extraction and threshold value comparison reduces human intervention and ensures the objectivity of the results. Subsequent process optimization can directly call database data to analyze abnormal distribution patterns to adjust welding parameters.

[0047] For actual application scenarios, welding spot contour analysis can also be extended to molten pool dynamic feature detection, further improving accuracy by increasing sampling density or adjusting algorithm parameters, such as increasing sampling points to 1024 or optimizing region growing threshold to 15, to adapt to the needs of high-speed welding or complex weld scenarios. This extension design provides flexibility for different welding environments. High sampling density can capture minor deformations, and threshold optimization enhances anti-interference capability. The adjusted method significantly improves the sensitivity to abnormal features while keeping the core logic unchanged, making it suitable for high-precision fields such as electronic devices or medical devices.

[0048] S103, if the welding spot is detected to have an abnormality, an abnormal region is identified and its boundary is extracted through image segmentation technology, the area and geometric characteristics of the abnormal region are calculated using pixel statistics method to determine the size and deformation degree of the welding spot, and specifically, a feature vector of the abnormal region is generated through region labeling, watershed segmentation and multi-dimensional feature analysis and stored in a database.

[0049] In the embodiment of the present application, this step aims to perform fine analysis on the abnormal welding spot to quantify the degree of its defects. The identification of the abnormal region is based on image segmentation technology, and the extraction of geometric features is completed through multi-dimensional calculation. The results provide data basis for subsequent process adjustment.

[0050] The weld image to be analyzed is obtained from the image database, and the region is annotated according to the preset abnormality judgment result to generate an initial abnormality marked image. The boundary of the initial abnormality marked image is extracted using the watershed segmentation algorithm to generate abnormal area boundary data, and the boundary features are calculated by curve fitting and chain code encoding method to obtain the geometric parameters.

[0051] An initial anomaly-marked image is generated using a preset anomaly type identifier. A watershed segmentation algorithm constructs a watershed line based on grayscale gradients. The gradient threshold is set between 50 and 80 to distinguish high-gradient areas such as cracks, while the gradient in normal areas is less than 20. At an image resolution of 40 pixels per millimeter, defects larger than 0.1 mm can be detected. After boundary data is smoothed using curve fitting, an 8-directional chain code is used to track the contour point by point, calculating the perimeter and area. The frequency of chain code direction changes in normal welds ranges from 0.2 to 0.3 per cycle, while in abnormal areas, it can exceed 0.5, reflecting the difference in contour complexity.

[0052] S1031. In an embodiment of the present invention, the coordinates of the center of gravity and the direction of the main axis are calculated based on the boundary data of the abnormal area to generate preliminary shape feature data, grayscale statistics are performed on the initial abnormal mark image to generate grayscale histogram data, and the density distribution characteristics of the abnormal area are extracted through multi-level threshold segmentation and region growing methods for further refined analysis.

[0053] The centroid coordinates are calculated by averaging the boundary points, and the principal axis directions are determined by principal component analysis, forming preliminary shape features. A grayscale histogram summarizes the pixel distribution in abnormal areas. Normal solder joints exhibit a unimodal distribution with a peak value between 180 and 220 pixels, while defective areas, such as slag inclusions, exhibit secondary peaks between 100 and 140 pixels. Multi-level threshold segmentation uses three thresholds of 120, 160, and 200, creating four grayscale levels. Region growing uses the grayscale centroid as the seed, a grayscale difference threshold of 15, and an 8-neighborhood algorithm to determine spatial adjacency. A density distribution map is generated to identify features with crack aspect ratios greater than 5 or pores that are approximately circular.

[0054] S1032. In an embodiment of the present invention, the minimum circumscribed rectangle method is used to calculate the major and minor axis parameters of the abnormal area in the density distribution map to generate size feature data, the symmetry of the preliminary shape features is analyzed by the regional projection method to generate deformation parameters, and the geometric parameters, size features and deformation parameters are integrated to form a feature vector of the abnormal area and stored in an associated manner.

[0055] The minimum enclosing rectangle is calculated using the convex hull of the boundary points. The long axis reflects the crack propagation trend. A normal solder joint has an aspect ratio of 0.8 to 1.2, while a crack has an aspect ratio of 3 to 5. Regional projection generates curves along the horizontal and vertical directions. Normal solder joints exhibit a bell-shaped distribution, while abnormal areas exhibit skewed or bimodal distributions. Deformation parameters are used to quantify symmetry. The feature vector includes six components: area ratio, perimeter ratio, shape factor, grayscale mean, standard deviation, and projective symmetry. This feature vector is linked to the original image via a hash index and stored in a database, with retrieval time under 5 milliseconds.

[0056] In this embodiment of the present invention, high-precision boundary extraction using watershed segmentation is suitable for complex defect identification, while chain coding and grayscale statistics enhance sensitivity to subtle deformations. The multidimensional design of feature vectors not only quantifies defect types, but also allows for adjustments to welding torch angle or energy input based on crack aspect ratio, for example.

[0057] In an embodiment of the present invention, the analysis accuracy can be further improved by increasing the grayscale threshold level or optimizing the projection direction to meet the needs of the production site. For example, the threshold can be increased to five levels or oblique projection can be introduced to adapt to the detection requirements of various welding defects, thereby improving the robustness and applicability of the method. The flexible design allows the method to be extended to different welding scenarios. The five-level threshold can refine the grayscale level, and the oblique projection can capture non-orthogonal deformation features. By adjusting the parameters, the method can more accurately identify multiple types of defects such as pores, slag inclusions or cracks while maintaining the core logic, thereby improving the reliability of real-time monitoring.

[0058] S104. Calculate the deviation between the preset welding gun path parameters and the extracted weld centerline coordinates to determine the position offset, and generate a corrected welding instruction based on the weld size and deformation data. Path optimization is achieved through path fitting, deviation smoothing, and kinematic verification, and is stored in the control database.

[0059] In this embodiment of the present invention, this step dynamically adjusts the welding gun's trajectory by analyzing the difference between the actual weld point location and the preset path to ensure welding accuracy. Path parameters are pre-set based on process requirements, while deviation calculation and correction rely on real-time data analysis to ultimately generate executable instructions.

[0060] S1041. In an embodiment of the present invention, a welding path coordinate sequence is extracted from a preset path database and a smooth first path curve is generated using a cubic spline interpolation method. The weld centerline coordinates are curve fitted to generate a second path curve. The distance between corresponding points of the two curves is calculated using the least squares method to generate initial deviation data.

[0061] The path coordinate sequence is stored as discrete points, with point spacing of approximately 2 mm for straight welds and 1 mm for curved welds. Cubic spline interpolation uses natural boundary conditions, with the first-order derivative at the endpoints being zero, to ensure smooth torch motion. During fitting, the number of weld centerline points is sampled at five times the path length; for example, a 100 mm weld generates 500 interpolation points. The least squares method is used to calculate deviations along the vertical direction, with the mean normal deviation kept within 0.2 mm, reflecting the fit between the weld point and the path.

[0062] S1042. In an embodiment of the present invention, the initial deviation data is interval sampled to calculate the local deviation mean to generate preliminary offset data, the preliminary offset data is smoothed by a sliding window method to generate smoothed offset data, and the smoothed offset data is segmented and marked according to a preset deviation threshold to determine the path offset interval.

[0063] Interval sampling divides the path into segments, taking the average deviation value for each segment. A sliding window is set to 15 points wide, with weights decreasing with distance from the center point. For example, Gaussian weighting is used to smooth out noise. If a segment's deviation exceeds 0.5 mm, it is marked as an offset interval. This segmented processing preserves local features while avoiding overcorrection caused by excessive global deviation, ensuring targeted adjustments.

[0064] In an embodiment of the present invention, the weld size anomaly and deformation data are read from the quality inspection database, the coordinate correction coefficient is calculated according to the deviation direction and degree and normalized to generate the final correction parameter, the path compensation value is calculated in combination with the offset interval data and a smooth path compensation curve is generated by segmented interpolation.

[0065] Dimensional anomalies include diameter deviations exceeding 15% or uneven widths. The direction of deformation determines the compensation direction, for example, for elliptical welds, compensation is applied along the minor axis. The correction factor is directly proportional to the deviation and should not exceed 30% of the standard diameter. Segmented interpolation extends the offset interval by 50 mm at both ends, using a cubic polynomial to ensure first-order continuity, generating a smooth compensation curve to optimize the welding gun trajectory.

[0066] In an embodiment of the present invention, a position correction value is generated according to a path compensation curve and combined with standard welding parameters to generate a correction instruction sequence. The feasibility of the instruction sequence is checked by a kinematic constraint verification method to generate a final qualified instruction sequence. The qualified instruction sequence and related parameters are stored in a welding control database to support subsequent tracing.

[0067] Standard parameters include speeds of 5 to 15 mm / s, voltages of 18 to 24 V, and currents of 80 to 120 amps. Position corrections are inversely proportional to speed. Kinematic verification ensures that the welding gun acceleration is less than 1000 mm / s², the angular velocity is less than 90 degrees / s, and the attitude angle variation is less than 3 degrees. Qualified instructions are stored with a 4-millisecond sampling period and contain position, attitude, and process data. Compensation parameters are linked to the original path using a weld number index.

[0068] S1043. In an embodiment of the present invention, for complex weld scenarios, correction accuracy can be improved by increasing the density of interpolation points or adjusting the window width. For example, the sampling points can be increased to 8 times the path length or the window can be expanded to 20 points to accommodate high-precision welding requirements with a curvature radius of less than 10 mm, thereby improving the flexibility and robustness of path adjustment. The optimized design enhances the method's adaptability to irregular welds. High-density sampling captures subtle deviations, while window expansion smooths complex fluctuations. Through these adjustments, correction instructions can more accurately match the actual weld state, improving welding quality and equipment operating efficiency.

[0069] S105. Obtain the corrected welding instruction data and extract the energy, focal length and welding gun angle parameters therein as analysis reference values. Specifically, generate a standardized reference parameter vector through instruction parsing, resampling and feature analysis and store it in the database for subsequent use.

[0070] In this embodiment of the present invention, this step aims to extract key parameters from a complex instruction sequence and establish a stable baseline value system. This extraction process combines time series analysis and fluctuation characteristic analysis to ensure that the parameters reflect the actual welding state. This method has no special requirements for the instruction generation device and can be flexibly adapted to different control systems.

[0071] S1051. In an embodiment of the present invention, the corrected welding instruction sequence is read from the instruction database and the timestamp and type identification are obtained through the parsing method to generate initial instruction data. The welding parameter-related instructions are filtered out according to the type identification and sorted by the timestamp to generate timing parameter data. The parameter extraction algorithm is then used to separate the energy, focal length and angle parameters to form a preliminary parameter array.

[0072] The command sequence is recorded with millisecond-level timestamps and includes three categories of instructions: motion, welding, and process. Welding parameter instructions account for approximately one-third of these instructions. The parsing process identifies the command type identifier for each frame and selects instructions containing voltage, current, focal length, and angle. After sorting, a 200mm weld generates approximately 2000 frames of data, with a sampling period of 10 milliseconds. A parameter extraction algorithm decomposes the command fields frame by frame, generating an array containing the three parameter categories for subsequent processing.

[0073] In an embodiment of the present invention, a preliminary parameter array is resampled using a fixed time window to generate resampled parameter data, normalized according to the unit and range of each parameter to generate a standardized parameter sequence, and the parameter stability is determined by calculating statistical characteristics and analyzing the fluctuation characteristics through wavelet decomposition to generate a reference parameter vector.

[0074] The resampling window width was set to 50 milliseconds with a 50% overlap to ensure data continuity. The voltage range of 18 to 24 volts, the focal length of 12 to 15 mm, and the angle of ±15 degrees were normalized to 0 to 1. The steady-state level was calculated using a sliding mean and standard deviation, and a three-layer wavelet decomposition was used to extract high-, medium-, and low-frequency fluctuations. High-frequency amplitudes were limited to 5% and medium-frequency amplitudes to 10%. Voltage deviations exceeding 0.5 volts or focal length fluctuations exceeding 0.5 mm were marked as unstable. The final vector was generated based on the minimum fluctuation principle and reconstructed into instruction format.

[0075] S1052. In an embodiment of the present invention, the generated benchmark parameter vector is weightedly fused with the process parameter limit to generate the final welding instruction benchmark value, which is written into the instruction benchmark database. At the same time, an index is established to associate and store the corresponding statistical parameters and fluctuation characteristic data to support rapid retrieval and process optimization.

[0076] During fusion, the mean deviation weight is 0.4, the standard deviation is 0.3, and the fluctuation feature is 0.3, with preference given to intervals with small fluctuations. The baseline value record is 32 bytes, and the feature data is 128 bytes. A hash index ensures efficient retrieval. The database supports retrieval by weld type or time, adapting to the parameter configuration requirements of batch tasks and improving welding consistency.

[0077] In an embodiment of the present invention, for dynamic welding scenarios, the accuracy of parameter extraction and anti-interference ability can be further improved by adjusting the window width to 30 milliseconds or increasing the wavelet decomposition level to 5 layers, thereby adapting to parameter analysis requirements under complex conditions such as arc fluctuations or workpiece deformation.

[0078] This optimized design enhances the method's adaptability to non-steady-state processes. A short window captures rapid changes, while multi-layer decomposition refines frequency characteristics. Through these adjustments, the baseline value more accurately reflects process status, providing more reliable data support for real-time control and quality improvement.

[0079] S106. By analyzing the energy distribution curve during the welding process and combining it with the real-time collected focal length data and welding gun angle changes, determine whether these parameters match the welding instruction reference values. Specifically, generate parameter matching identification through data verification, curve fitting and correlation analysis, and store the results to support process monitoring.

[0080] In the embodiment of the present application, this step aims to evaluate the execution effect of the welding parameters in real time, to ensure consistency with the expected benchmark. The energy distribution reflects the welding stability, and the focal length and angle affect the weld formation, thereby improving the quality control accuracy through multi-dimensional comparative analysis. The method is suitable for welding systems with various sensor configurations.

[0081] In the embodiment of the present application, energy sampling data during the welding process are extracted from the sensor database and validity verification is performed according to the time stamp to generate valid energy data. The valid energy data are smoothed by using the moving average method to generate an energy distribution curve. Meanwhile, focal length and angle data are collected by the sensor, and after the abnormal values are removed, the focal length and angle curves are generated by using piecewise least squares fitting.

[0082] The energy data are sampled at 1000 Hz, with a range of 800 to 3000 W. When verifying, the value range and continuity are checked. If there are more than 5 consecutive abnormal points, the segment is removed. The moving average is used to smooth high-frequency noise with a 21-point window. The focal length data are 10 to 20 mm, sampled at 200 Hz, and the angle data are plus or minus 30 degrees, sampled at 500 Hz. The abnormal values are removed according to 3 times the standard deviation. The fitting is segmented at 50 ms, and a quadratic polynomial is used to ensure the continuity of the curve, reflecting the parameter trend.

[0083] In the embodiment of the present application, welding benchmark parameters are read from the benchmark value database, and the energy, focal length and angle values thereof are normalized to generate standard benchmark data. The energy distribution curve and the focal length and angle curve are time-aligned and resampled by using a fixed time window to generate a parameter sequence with a uniform frequency. Then, the characteristic values of the curves are calculated, and correlation analysis is performed with the standard benchmark data to evaluate the matching degree.

[0084] The benchmark values are normalized by using the maximum and minimum value method to map to 0 to 1. The energy benchmark is 85% to 95% of the rated power, and the focal length benchmark is matched with 15 mm and 5 mm steel plates. The time alignment window is 100 ms, the energy is 100 points, the focal length is 20 points, and the angle is 50 points, which are resampled to 100 Hz. The characteristic values include the mean, standard deviation, peak factor and waveform factor. The cosine similarity is used to calculate the matching degree. A high matching degree is higher than 0.95, a medium matching degree is from 0.9 to 0.95, and a low matching degree is lower than 0.9.

[0085] In the embodiment of the present application, threshold data are generated according to the process parameter fluctuation range, and multi-dimensional threshold judgment is performed in combination with the matching degree results to generate a parameter matching identifier. The matching identifier and the similarity data are written into the result database and are associated with the characteristic parameters and the determination rules for subsequent tracing and optimization.

[0086] Thresholds are set for energy fluctuations of ±5%, focal lengths of ±0.5 mm, and angles of ±2 degrees. Weighted voting is used to determine matches, with energy weights of 0.5, focal lengths of 0.3, and angles of 0.2. The results database records a 16-byte timestamp, 32-byte parameter value, and a 16-byte identifier, packaged in 5-minute increments for fast retrieval. Matching analysis improves weld height standard deviation by 40% and width consistency by 35%.

[0087] In this embodiment, for highly dynamic welding scenarios, real-time analysis can be enhanced by shortening the sampling window to 50 milliseconds or increasing the sampling frequency to 2000 Hz. This allows for more accurate capture of the effects of sudden energy changes or angle adjustments, further optimizing parameter matching reliability and process adaptability, and improving responsiveness to transient changes. High-frequency sampling refines data granularity, while short sampling windows focus on local features, ensuring parameter matching results are more accurate and accurate for actual working conditions, providing technical support for high-quality welds.

[0088] S107. If the welding energy and focal length are detected to be mismatched with the welding instruction baseline values, the equipment control parameters are adjusted based on the deviations, and the welding gun path correction is recalculated. Specifically, adaptive mapping is used to calculate the deviations, binocular vision positioning offsets, and kinematic mapping to generate optimized path data, which is stored in a database to improve welding accuracy. Real-time feedback is used to adjust parameters and paths, mitigating the impact of deviations on weld quality. Adaptive algorithms ensure that adjustments match the degree of deviation, while visual positioning and kinematic optimization ensure the accuracy and feasibility of path corrections. This method is applicable to a variety of welding processes and robotic systems.

[0089] S1071. In an embodiment of the present invention, welding parameter deviation data is extracted from a parameter analysis database and an adaptive mapping algorithm is used to calculate the deviation values ​​of energy and focal length to generate initial deviation data. Compensation coefficients are generated according to process rules and optimized to obtain corrected control parameters. At the same time, a binocular vision sensor is used to capture welding position images and the position difference is calculated through image registration to generate initial offset data.

[0090] Deviation data includes the difference between the real-time energy and focal length values ​​and the baseline values. Adaptive mapping linearly compensates for deviations within ±5%, with nonlinear adjustments beyond this. For example, at 2000 watts, a 50-watt deviation corresponds to a coefficient of 0.8, while a 100-watt deviation corresponds to a coefficient of 1.2. The binocular vision system has a resolution of 1280×1024 and a field of view of 40×32 mm. Phase correlation registration accuracy is 0.1 mm. The deviation vector between the melt pool center and the reference point is calculated, providing a basis for subsequent corrections.

[0091] In the embodiment of the present application, the initial offset data is converted into the welding gun coordinate system by a coordinate transformation matrix to generate welding gun offset data, the position compensation coefficient is calculated according to the correction control parameter and normalized to generate a compensation vector, and the offset data is smoothed by cubic spline interpolation to generate path correction and mapped to the joint space by the Jacobian matrix to generate path compensation data.

[0092] The coordinate transformation is realized by a 4*4 matrix, the origin of the welding gun coordinate system is at the nozzle end, and the z-axis is along the axis. The compensation coefficient is adjusted with the speed, which is 1.2 times the deviation at 10 mm / s, and increases to 1.5 at 20 mm / s. The cubic spline interpolation control point spacing is 10 mm to ensure path smoothing. The inverse solution of the Jacobian matrix converts the end pose into joint angle to generate compensation data and optimize path continuity.

[0093] In the embodiment of the present application, the path compensation data is verified by kinematic constraints and trajectory optimization to generate executable path data, the trajectory reachability is verified by motion simulation to generate verification results, and qualified path data is generated accordingly. Finally, the qualified path data and related deviation and compensation parameters are written into the control parameter database and indexed for traceability.

[0094] The constraints include joint angle positive and negative 170 degrees, angular velocity 120 degrees per second, acceleration 200 degrees per square second, gradient descent optimization smoothness and time. The simulation step is 0.001 seconds, and the singular position and workspace boundary are checked. The database records 64 bytes of path, 32 bytes of deviation and compensation data, supports multi-condition retrieval, and the accuracy is improved to 0.1 mm.

[0095] In the embodiment of the present application, for complex welds or high-speed welding, the correction accuracy can be further improved by encrypting the registration points to 16 points or shortening the interpolation interval to 5 mm, so as to adapt to the dynamic change of the molten pool or the posture adjustment demand, and ensure the consistency of the weld formation and the process stability.

[0096] This optimization enhances the adaptability of the method. The encrypted registration improves the positioning accuracy, and the short-interval interpolation refines the path adjustment, so that the system can cope with high dynamic scenarios and significantly improve the quality consistency.

[0097] S108, input the adjusted control parameters into the welding equipment control system and monitor the energy and focal length changes in real time, and at the same time verify the stability of the contour and center line by continuously collecting the weld point image to judge whether the welding quality meets the preset standard. Specifically, the quality judgment result is generated by sensor data analysis and image processing, and adjustment suggestions are put forward.

[0098] In the embodiment of the present application, this step ensures the stability of the welding process and the controllability of the weld quality by real-time monitoring and image analysis. Parameter monitoring provides dynamic feedback, image verification quantifies forming features, and the combination of the two improves the reliability of quality evaluation. The method can be adapted to different welding equipment and workpiece types.

[0099] S1081. In an embodiment of the present invention, the corrected control parameters are written into the controller to generate an execution instruction and the energy and focal length data of the welding process are collected by the equipment sensor to generate initial monitoring data. The outliers are eliminated through validity verification to generate valid monitoring data. The sliding average method is then used to analyze the parameter change trend to generate trend data. At the same time, a high-speed industrial camera is used to continuously collect weld formation images and extract contour and centerline data after screening and enhancement.

[0100] Energy data was sampled at 1000 Hz, with a focal length of 200 Hz. Verification used a 3x standard deviation rule to eliminate outliers and a 25-point sliding window for smoothing. Under a 2000-watt process, energy fluctuations of ±100 watts and focal lengths of ±0.5 mm were considered normal. The camera frame rate was 200 frames per second, with an exposure of 0.5 milliseconds. After screening for clarity and contrast, grayscale was enhanced to a range of 0 to 255 using histogram equalization. A watershed algorithm was used to extract contours based on gradients, and the least-squares method was used to fit the centerline.

[0101] In an embodiment of the present invention, shape parameters and centerline offset parameters of contour data are calculated to generate stability data and statistical analysis is performed on trend data to generate quality feature vectors. Standard feature data are read from a quality standard database and the matching degree is calculated through multi-dimensional threshold comparison to generate a quality judgment result. At the same time, parameter adjustment suggestions are generated based on the results and correction amounts are calculated to optimize control parameters.

[0102] Shape parameters include roundness, symmetry, and smoothness, while offset parameters include average and maximum offset. The fitting error for a normal solder joint is less than 0.2 mm. Trend data is extracted to include mean, standard deviation, crest factor, and form factor. The feature vector is 8-dimensional, and the weighted Euclidean distance is used to measure the degree of matching. Adjustments utilize proportional-integral control with a 20-watt energy step size, a 0.1-mm focal length, and a 100-ms compensation period to ensure real-time response.

[0103] S1082. In an embodiment of the present invention, the quality judgment results and adjustment data are written into a quality monitoring database and an index is established to associate and store the monitoring data and feature data for traceability. At the same time, a closed-loop feedback mechanism is used to implement adaptive parameter adjustment to improve welding stability and consistency.

[0104] The database records 64 bytes of quality data and 128 bytes of process parameters, supporting sub-millisecond retrieval. Closed-loop feedback significantly reduces dimensional standard deviation by 40% and centerline deviation by 50%, increasing pass rate by 15 percentage points and reducing rework.

[0105] In an embodiment of the present invention, for high-precision or complex workpieces, the quality analysis can be further refined by increasing the camera frame rate to 500 frames per second or increasing the feature vector dimension to 12 dimensions, thereby more accurately capturing tiny defects and optimizing parameter adjustment effects to meet strict process requirements.

[0106] This optimization improves the ability to detect subtle changes. High frame rates reveal dynamic details, and multi-dimensional features enhance identification accuracy, ensuring high-quality weld formation.

[0107] It will be apparent to those skilled in the art that the present application is not limited to the details of the exemplary embodiments described above and that the present application can be implemented in other specific forms without departing from the spirit or essential characteristics of the present application. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the present application is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be included therein. Any reference sign in a claim should not be construed as limiting the claim to which it relates.

Claims

1. A control method for a welding coupling device, characterized in that: The method comprises: A high-speed camera is used to capture continuous image sequences of the welding instruction execution process, and an edge detection algorithm is used to extract the contour information of the weld point. The center line of the weld point is generated by combining the morphological processing method. Based on the pre-set weld spot size and weld width thresholds, the extracted contour and centerline data are compared to determine whether the weld spot has size anomalies or weld width anomalies. If there is an abnormality, the image segmentation method is used to identify the abnormal area in the image, and the area and geometric characteristics of the abnormal area are calculated through pixel statistics to obtain the size and deformation degree of the solder joint; Based on the preset welding gun path parameters and the extracted weld centerline coordinates, the deviation between the weld centerline and the preset path is calculated to obtain the weld position offset. The welding instructions are then corrected based on the weld size and deformation. Obtain the corrected welding instruction data and extract the energy, focal length and welding gun angle parameters in the instruction as the reference values ​​for welding instruction analysis; By analyzing the energy distribution curve during welding, combined with real-time collected focal length data and welding gun angle changes, it is determined whether the welding energy and focal length match the baseline values ​​analyzed in the welding instruction; If the welding energy and focal length do not match the reference values ​​analyzed by the welding instruction, the control parameters of the welding equipment are adjusted according to the deviation values ​​and the correction amount of the welding gun path is recalculated; The adjusted control parameters are input into the welding equipment control system to monitor the energy and focal length changes during the welding process in real time. By continuously collecting welding process images, the stability of the weld point contour and center line is verified to determine whether the welding quality meets the preset standards.

2. The method according to claim 1, characterized in that The method collects a continuous image sequence of the welding instruction execution process by a high-speed camera, extracts the contour information of the welding spot by using an edge detection algorithm, and generates the center line of the welding spot by combining a morphological processing method, including: Acquire a first image sequence of the welding head motion trajectory according to a preset acquisition frequency, extract first grayscale image data from the first image sequence, and obtain first binary image data by using image histogram statistics to obtain an optimal grayscale threshold; Performing noise reduction processing on the first binarized image data using a Gaussian filter to obtain second binarized image data, and performing edge detection on the second binarized image data using a Sobel operator to obtain first edge contour data; Calculating the density distribution of contour pixels for the first edge contour data, performing breakpoint connection processing on the edge contour using a density clustering method to obtain second edge contour data, and processing the second edge contour data through a morphological closing operation to obtain closed contour line data; A morphological thinning algorithm based on distance transformation is used to perform skeleton extraction on the closed contour line data to obtain first skeleton data, and a cubic spline interpolation algorithm is used to perform node continuity processing on the first skeleton data to obtain welding point contour centerline data.

3. The method according to claim 1, characterized in that The method of comparing the extracted contour and centerline data based on the preset weld spot size and weld width thresholds to determine whether the weld spot has abnormal size or weld width includes: Reading standard solder joint size parameters from a preset parameter database, and processing the first solder joint contour data using a contour feature extraction method to obtain actual solder joint size parameters; Extracting weld width feature regions using a region growing algorithm based on the first weld spot contour data to obtain weld width data, and performing statistical analysis on the weld width data using a distance transformation method to obtain a weld width mean; Processing the first solder joint contour data using an equidistant sampling method to obtain a first sampling point sequence, and obtaining first curvature distribution data by local curvature calculation; If the first to fourth abnormal feature data form a comprehensive abnormal feature vector of a weld spot, a multi-dimensional threshold comparison is performed based on the comprehensive abnormal feature vector and the abnormality discrimination threshold to obtain a weld spot abnormality type identifier.

4. The method according to claim 1, wherein If an abnormality exists, an image segmentation method is used to identify the abnormal area in the image, and the area and geometric features of the abnormal area are calculated by a pixel statistics method to obtain the size and deformation degree of the solder joint, including: Marking the weld spot image according to a preset abnormality judgment result to obtain a first abnormality marked image; Using a watershed segmentation algorithm to extract the boundary of the first abnormality marked image to obtain first region boundary data, wherein the first region boundary data is used to calculate geometric parameters of the abnormal region; Calculating the coordinates of the center of gravity and the direction of the principal axis of the abnormal region according to the first region boundary data to obtain first shape feature data, wherein the first shape feature data is used for density distribution analysis of the abnormal region; A symmetry parameter is calculated for the first shape feature data using a region projection method to obtain a first deformation parameter, and the first deformation parameter is combined with the first geometric parameter data to form an abnormal region feature vector.

5. The method according to claim 1, wherein The method of calculating the deviation between the center line of the weld point and the preset path based on the preset welding gun path parameters and the extracted weld center line coordinates to obtain the weld position offset and correcting the welding instructions based on the size and deformation of the weld point includes: Obtaining a welding path coordinate sequence from a preset path database, generating a first path curve using a cubic spline interpolation method, and calculating the distance between corresponding points based on the first path curve and a fitting curve of the weld centerline coordinates to obtain first deviation data; Performing interval sampling on the first deviation data and calculating a local deviation average to obtain first offset data, and smoothing the first offset data using a sliding window method to obtain path offset interval data; Calculating a coordinate correction coefficient based on the path offset interval data and the abnormal solder joint size data in the quality inspection database to obtain a second correction parameter, and generating a path compensation curve using a piecewise interpolation method to obtain second path compensation data; A position correction amount is generated according to the second path compensation data to obtain a second instruction parameter, the second instruction parameter is integrated with the standard welding parameter to obtain a corrected instruction sequence, and a kinematic constraint verification method is used to perform a feasibility test on the corrected instruction sequence to obtain a qualified instruction sequence.

6. The method according to claim 1, characterized in that The step of obtaining and executing the corrected welding instruction data and extracting the energy, focal length, and welding gun angle parameters in the instruction as reference values ​​for welding instruction analysis includes: Reading a welding instruction sequence from an instruction database using an instruction parsing method, and obtaining an instruction timestamp and an instruction type identifier using the instruction parsing method to obtain a welding parameter instruction set; Sorting the welding parameter instruction set according to the timestamp, separating the energy parameter, the focal length parameter and the angle parameter by a parameter extraction algorithm to obtain a first parameter array; Resampling the first parameter array using a fixed time window, and normalizing the resampled data according to parameter units and ranges to obtain a first standard parameter sequence; Statistical parameters and fluctuation characteristics are calculated for the first standard parameter sequence. If the statistical parameters and fluctuation characteristics meet the stability interval requirements, a reference parameter vector is generated. The reference parameter vector is reconstructed according to the instruction format to obtain a welding instruction reference value.

7. The method according to claim 1, characterized in that The analysis of the energy distribution curve during the welding process, combined with the real-time collected focal length data and welding gun angle changes, determines whether the welding energy and focal length match the reference values ​​analyzed in the welding instruction, including: Reading energy sampling data during the welding process from a sensor database, verifying the validity of the energy sampling data according to a sampling timestamp, and obtaining first curve data using a sliding average method; The focal length data and angle data are subjected to outlier elimination, and the second curve data is obtained by piecewise least square fitting; Reading welding reference parameters from a reference value database, and normalizing energy parameters, focal length parameters, and angle parameters in the welding reference parameters to obtain standard reference data; The first curve data and the second curve data are time-series aligned to obtain characteristic parameter data, a curve correlation algorithm is used to calculate the matching degree between the characteristic parameter data and the standard reference data, and a parameter matching identifier is obtained according to parameter threshold data.

8. The method according to claim 1, characterized in that If the welding energy and focal length do not match the reference values ​​analyzed by the welding instruction, the control parameters of the welding equipment are adjusted according to the deviation value, and the correction amount of the welding gun path is recalculated, including: Reading welding parameter deviation data from a parameter analysis database, and calculating an energy deviation value and a focal length deviation value using an adaptive mapping algorithm according to the welding parameter deviation data to obtain first deviation data; A binocular vision sensor is used to collect a welding position image, and a position difference is calculated based on the welding position image by an image registration method to obtain first offset data; The first offset data is converted into a welding gun coordinate system using a coordinate transformation matrix to obtain second offset data, and a welding gun position compensation coefficient is calculated using the second offset data to obtain a first compensation vector; A first path correction value is obtained by using a cubic spline interpolation method according to the first compensation vector, and path compensation data is obtained by mapping the first path correction value to a joint space using a Jacobian matrix method.

9. The method according to claim 1, characterized in that The adjusted control parameters are input into the welding equipment control system to monitor the energy and focal length changes during the welding process in real time. By continuously collecting welding process images, the stability of the weld point contour and center line is verified to determine whether the welding quality meets the preset standards, including: Using a device sensor to collect energy data and focal length data during the welding process to obtain first monitoring data, and obtaining second monitoring data through data validity verification based on the first monitoring data; Continuously capturing weld formation images using a high-speed industrial camera according to the second monitoring data to obtain a first image sequence, and extracting weld contour data from the first image sequence using a watershed segmentation algorithm to obtain a first contour sequence; Calculating contour shape parameters and centerline offset parameters of the first contour sequence using a continuity detection method to obtain first stability data; A quality feature vector is generated by performing statistical analysis on the first stability data to obtain first feature data, and a quality determination result is obtained by calculating a matching degree between the first feature data and standard feature data using a multi-dimensional threshold comparison method.

Images