Welding path priority method for automobile electronic ignition coil and welding equipment thereof

Abstract

The invention discloses a welding path priority method and welding equipment for an automobile electronic ignition coil, and the welding method comprises the steps: obtaining point cloud data through preset track global scanning, generating a digital three-dimensional model consistent with a welding driving motion coordinate system through Gaussian filtering denoising, iterative nearest point registration, B-spline surface fitting and other strategies, and obtaining a welding driving motion coordinate system; the problem of coordinate deviation of traditional path planning is solved. A contour recognition strategy is combined with region growth, feature extraction and template comparison, and a least square method is used for fitting a center line to serve as a reference track; a node setting strategy is adopted, forced and uniform nodes are scientifically arranged, and accurate positioning of a welding head is guaranteed. A double correction mechanism of molten pool monitoring and node feedback is innovatively adopted, a thermal imager collects images and extracts molten pool parameters after preprocessing, a real-time correction instruction is generated in combination with three-level judgment, a lower node adjustment instruction is generated through image processing after node welding, closed-loop control is achieved, welding defects are reduced, and the percent of pass is increased. Corollary equipment adopts an integrated disc design and is matched with a rotary bearing mechanism to realize batch positioning and conveying, a multi-degree-of-freedom mechanical arm performs automatic welding according to a planned path, parameters are adjusted according to welding seam characteristics, and both precision and production efficiency are considered.

Description

Technical Field

[0001] This invention relates to the field of welding technology, and in particular to a welding path prioritization method and welding equipment for automotive electronic ignition coils. Background Technology

[0002] As a core component of the engine ignition system, the automotive electronic ignition coil requires welding to fix and seal the joints of its housing, winding ends, and terminals. The welding precision directly determines the insulation performance, structural stability, and service life of the ignition coil, thus affecting the engine's ignition efficiency and operational reliability. With the development of automotive electronics technology towards higher precision and miniaturization, stringent requirements have been placed on the uniformity of ignition coil weld formation, positional accuracy, and welding stability. Therefore, a precise and efficient welding path planning and welding control method, coupled with suitable welding equipment, is urgently needed to meet the high-precision welding demands of mass production.

[0003] In the existing technology, the contour recognition method does not combine the specific structural features of the ignition coil for accurate region segmentation, the feature extraction of suspected weld areas is not comprehensive, and there is a lack of a standardized comparison mechanism with the preset weld feature template. This makes it easy for welds to be missed or misidentified. In addition, the center contour line extraction accuracy is insufficient, and it cannot provide a reliable benchmark for the welding path.

[0004] The system failed to implement mandatory node placement at abrupt structural changes such as weld inflection points and width transitions. The node spacing in straight sections lacked scientific basis. Welding execution parameters were not adjusted in a gradient manner to account for weld structural differences. Furthermore, the node parameters and the welding drive system exhibited poor coordination, easily leading to weld trajectory deviations and forming defects.

[0005] Existing welding controls are mostly passive execution of preset parameters, lacking real-time monitoring of the molten pool state between nodes, and molten pool offset and temperature fluctuation cannot be corrected in time. At the same time, the feedback of forming and temperature data after node welding is not timely, and the deviation cannot be quickly transmitted to the next node, which can easily lead to the accumulation of deviation and cause defects such as undercut and burn-through. Summary of the Invention

[0006] To achieve the above objectives, one technical solution adopted by the present invention is: a welding path prioritization method for automotive electronic ignition coils, the method comprising: According to the preset scanning time and preset scanning trajectory, the surface of the ignition coil and the area to be welded are scanned in the whole domain to obtain the three-dimensional coordinates of each scanning point and form the original point cloud data; the original point cloud data is processed by the stitching fitting strategy to obtain the digital three-dimensional model of the ignition coil. The contour recognition strategy is used to identify the weld area in the digital 3D model, extract the center contour line of each weld segment, and use the contour line as the reference trajectory of the welding path; according to the shape of the weld, the movement direction, start point and end point of each welding path segment are defined; according to the node setting strategy, the welding execution data of each path node is set on each welding path segment. Welding is performed sequentially along the path nodes according to the welding path and welding execution parameters; Thermal imaging images of the molten pool are acquired between every two path nodes according to a preset acquisition frequency. The coordinates and temperature of the molten pool center are obtained through path image processing strategies. The coordinates and temperature of the molten pool center are compared with the preset molten pool data through a molten pool offset monitoring and correction strategy, and real-time correction instructions are generated. After each welding path is completed, the node thermoforming image of that path node is acquired, and the node forming data and node temperature data are obtained through the node image processing strategy; through the feedback data comparison and adjustment instruction generation strategy, the node forming data and node temperature data of each node are compared with the preset node data, and the adjustment instruction for the next node is generated. By interleaving and coordinating real-time correction and adjustment commands, the stability of the welding process in all welding paths is ensured, and welding defects caused by feedback lag are avoided. After the entire weld is completed, a full-area image is captured and the overall image recognition strategy is used to detect the forming state of the overall weld image. If the detection result meets the preset standard, it is judged as a qualified product.

[0007] Furthermore, the splicing and fitting strategy includes: A Gaussian filtering algorithm is used to smooth the original point cloud data, a Gaussian distribution function is constructed, and a weighted average calculation is performed on the three-dimensional coordinates of each scanning point and a preset number of neighboring points to obtain the coordinate deviation value of each scanning point; sampling points whose coordinate deviation value exceeds the noise point judgment threshold are judged as noise points and removed. A segment of the point cloud covering the reference features of the ignition coil is selected as the target point cloud, and the point clouds obtained from the scanning of the remaining segments are used as source point clouds. The rotation and translation matrices between the source and target point clouds are calculated iteratively. In each iteration, the sampling point in the source point cloud is matched with the nearest point in the target point cloud, and the distance error between the two points is calculated. The iteration stops and the optimal coordinate transformation relationship is determined when the distance error is less than the preset registration accuracy threshold. Based on the optimal coordinate transformation relationship, all source point clouds are uniformly transformed into the coordinate system of the target point cloud so that the coordinate system is consistent with the motion coordinate system of the welding drive. Based on the structural characteristics of the ignition coil, the global point cloud is divided into multiple fitting regions, and surface fitting is performed on each region. Then, through a surface stitching algorithm, the fitted surfaces of each region are integrated into a complete three-dimensional surface model. When the fitting error between the fitted surface and the original point cloud does not exceed the preset fitting error threshold, a digital three-dimensional model is generated. The coordinate system of this digital three-dimensional model is the same as the motion coordinate system of the welding drive.

[0008] Furthermore, the contour recognition strategy includes: The digital 3D model is smoothed and denoised. Based on the preset structural features of the automotive electronic ignition coil, a region growing algorithm is used to segment the digital 3D model into regions, dividing the model into the shell region, winding region, terminal region, and suspected weld seam region. Among them, the suspected weld seam region is the connection and transition region between different structural regions. Three-dimensional features are extracted from suspected weld areas, including the region's normal vector direction, surface curvature, contour edge gradient, and thickness parameter. The normal vector direction distinguishes the orientation difference between the weld area and adjacent structural areas, with a significant angle between the weld area's normal vector direction and that of adjacent structural areas. Surface curvature identifies concave or convex features in the weld area. Contour edge gradient captures the boundary contour of the weld area, with the gradient difference between the weld boundary and the edge gradient of adjacent structures exceeding a preset gradient threshold. The thickness parameter filters out invalid suspected areas, retaining areas with a thickness equal to the preset weld width as target areas for identification. The extracted features of suspected weld areas are compared with the preset weld feature template. When the similarity between the features of the suspected weld area and the preset weld feature template is greater than or equal to the preset comparison threshold, the area is determined to be a valid weld area to be welded. A contour extraction algorithm is used to extract the boundary contour lines of the effective weld seam to be welded area; the center line of the boundary contour lines on both sides is obtained by coordinate fitting calculation, and this center line is used as the center contour line of each weld seam segment.

[0009] Furthermore, node configuration strategies include: Path nodes are set at preset node spacing for weld inflection points, weld width change points, the start and end points of irregular structures, and straight and uniform sections of welds; the welding execution parameters are three-dimensional coordinates, preset welding parameters, standard parameters for welding feedback, and feedback deviation thresholds. Each path node extracts its three-dimensional coordinates from the welding execution parameters on the weld center contour line. These three-dimensional coordinates are consistent with the digital three-dimensional model and the motion coordinate system of the welding drive, serving as the target positioning coordinates of the welding head during node welding. This ensures that the welding head can move accurately to the preset position of each node, providing a positioning reference for the accurate execution of the welding path. The three-dimensional coordinates of each node are unique and continuous with the coordinates of adjacent nodes, avoiding welding path deviation caused by node coordinate discontinuity. For each path node, define the preset welding parameters, welding feedback standard parameters, and feedback deviation threshold in the welding execution parameters; All path nodes on each welding path are numbered sequentially according to the direction of welding movement, with each node corresponding to a unique number, and the welding execution data of each node is associated and stored with it. The preset welding parameters include welding speed, welding power, and welding torch angle; for adjacent nodes on the same weld, the preset welding parameters are adjusted in a gradient according to the differences in weld structure to ensure that the welding parameters are the same as the weld characteristics. The feedback standard parameters are consistent with the feedback standard parameters in the welding execution parameters, including the node weld formation standard parameters and the node temperature standard parameters. The weld formation standard parameters include the preset weld width, preset weld height, and preset penetration depth, which are the same as the preset weld size. The node temperature standard parameters are preset optimal welding temperature ranges, which are used to provide a clear benchmark for the comparison of feedback data after node welding is completed. The feedback deviation threshold is consistent with the feedback deviation threshold in the welding execution parameters, corresponding to the node feedback standard parameter setting; the feedback deviation threshold includes coordinate deviation threshold, forming deviation threshold and temperature deviation threshold; the coordinate deviation threshold is used to determine the degree of offset between the center of the molten pool and the three-dimensional coordinates of the node, the forming deviation threshold is used to determine the degree of deviation between the forming size of the node weld and the feedback standard parameter, and the temperature deviation threshold is used to determine the degree of deviation between the temperature of the node weld and the optimal temperature range.

[0010] Furthermore, path image processing strategies include: The acquired molten pool thermal imaging images are preprocessed with noise reduction, enhancement and geometric correction to obtain the preprocessed molten pool thermal imaging images and geometric correction parameters; An adaptive threshold segmentation algorithm, combined with the grayscale distribution characteristics of the molten pool thermal imaging image, is used to divide the preprocessed molten pool thermal imaging image into molten pool region and non-molten pool region. A fitting algorithm is used to fit the contour of the segmented molten pool region to obtain the contour boundary coordinates of the molten pool region, and the geometric center of the contour boundary is calculated. This geometric center is defined as the image pixel coordinates of the molten pool center. Based on the geometric correction parameters, the pixel coordinates of the molten pool center are converted into the actual coordinates of the molten pool center. The coordinates of the molten pool center are consistent with the 3D coordinates of the nodes in the digital 3D model, the motion coordinate system of the welding drive, and the node setting strategy. The coordinates of the molten pool center are used to compare with the 3D coordinates of the nodes to determine the degree of molten pool offset. Based on the preset calibration relationship between grayscale values ​​and temperature of thermal imaging images, grayscale values ​​are statistically analyzed in the segmented molten pool region; the average grayscale value within the molten pool region is selected as the real-time molten pool temperature; the real-time molten pool temperature is used to compare with the preset optimal welding temperature range and to assist in generating real-time correction commands.

[0011] Furthermore, the melt pool offset monitoring and correction strategy includes: During the welding process, based on the node numbers of the two paths where the current welding is located, the preset molten pool data of the corresponding node is retrieved from the associated stored node welding execution data; the preset molten pool data includes the three-dimensional coordinates of the node corresponding to the current welding segment, the preset optimal welding temperature range, the coordinate deviation threshold, and the temperature deviation threshold. The center coordinates of the molten pool and the real-time temperature of the molten pool output by the synchronous acquisition path image processing strategy are compared with the preset molten pool data and judged into three levels according to the degree of deviation: normal state, slightly abnormal state and severely abnormal state. Based on the three-level judgment, real-time correction instructions are generated for minor anomaly correction instructions and severe anomaly correction instructions; After the welding driver executes the real-time correction command, the path image processing strategy acquires thermal imaging images of the molten pool at a preset acquisition frequency, extracts the molten pool data, and monitors and compares it again until the molten pool state returns to normal before welding is performed. Each generated real-time correction command, the molten pool data before and after correction, the welding execution parameters, and the correction effect are associated with and stored with the path node number of the current welding segment.

[0012] Furthermore, node image processing strategies include: The acquired nodal thermoforming images are preprocessed with noise reduction, enhancement, and geometric correction. An adaptive threshold segmentation algorithm is adopted, which combines the gray-scale distribution characteristics of the node thermoforming image with the preset solder joint width and preset solder joint height parameters in the node setting strategy to automatically determine the segmentation threshold and divide the preprocessed image into node solder joint forming area and non-forming area. Contour fitting and size measurement are performed on the segmented node solder joint forming area, and the extracted node forming data corresponds one-to-one with the node solder joint forming standard parameters in the node setting strategy. Based on the preset calibration relationship between grayscale values ​​and temperature of thermal imaging images, grayscale values ​​are statistically analyzed for the segmented node solder joint forming area; the average grayscale value within the forming area is selected as the real-time temperature data of the node, which reflects the residual heat state of the node solder joint after welding is completed. The extracted node forming data and node temperature data are associated and bound to the unique number of the current node. This is used to provide feedback data comparison and adjustment instruction generation strategy. The data is compared with preset node data to generate adjustment instructions for the next path node. This allows for interleaving with real-time correction instructions to ensure the continuity and stability of the welding process.

[0013] Furthermore, the feedback data comparison and adjustment instruction generation strategy includes: Upon completion of welding at each path node, the node forming data, node temperature data, and unique node number extracted and bound using the node image processing strategy are retrieved. Simultaneously, the corresponding preset node data for that node, as well as the correction data related to the current welding segment stored in the molten pool offset monitoring and correction strategy, are retrieved from the node welding execution data. This enables the correlation and matching of feedback data, preset data, and historical correction data. The retrieved node forming data is compared one by one with the preset weld point forming standard parameters of the corresponding node, and the forming deviation value of each parameter is calculated; the node temperature data is compared with the preset optimal welding temperature range of the corresponding node, and the temperature deviation value is calculated; during the comparison process, the historical correction data of the molten pool offset monitoring and correction strategy is referenced simultaneously to correct the deviation calculation error and ensure the accuracy of the comparison results. Based on the forming deviation value and temperature deviation value, combined with the forming deviation threshold and temperature deviation threshold in the node setting strategy, the welding quality of the current node is judged in three levels: no deviation state, slight deviation state, and large deviation state.

[0014] Based on the three-level node determination, node adjustment instructions for the next path node are generated, including no adjustment, minor adjustment, and targeted adjustment. The node adjustment instructions are consistent with the welding execution parameters and the motion coordinate system of the welding drive in the node setting strategy. The generated adjustment instructions for the next node are synchronously sent to the welding driver to guide the welding execution of the next node. The feedback data, comparison results, grading judgment results of the current node, and the adjustment instructions for the next node are associated and stored with the current node number and the next node number to realize the interleaving of real-time correction instructions and adjustment instructions, and avoid welding defects caused by feedback lag.

[0015] Another technical solution adopted by the present invention is: a welding device, which is applicable to the above-mentioned welding path prioritization method for automotive electronic ignition coils, and the welding device includes: Base 1, which is used to support all the components of the device; At least one ignition coil carrying mechanism 2 is fixedly mounted on the base 1; each of the ignition coil carrying mechanisms 2 is used to carry the ignition coil integration disk 3; the ignition coil integration disk 3 is evenly arranged with a plurality of ignition coils to be welded; The robotic arm welding mechanism 4 is mounted on the base 1; The robotic arm welding mechanism 4 is equipped with a thermal imager 5 and a three-dimensional laser scanner 6. The 3D laser scanner 6 is used to emit a laser beam to perform a full-area scan of the surface of the ignition coil to be welded and the area to be welded, capturing the 3D spatial information of the area to be welded, converting the scan data into a digital point cloud, and generating a digital 3D model of the ignition coil. The thermal imager 5 is equipped with dual-lens cameras, which simultaneously acquire the forming image and temperature region and data of the weld point or weld seam of the ignition coil after welding, and generate a fused image.

[0016] Furthermore, the ignition coil carrying mechanism 2 includes: Rotary motor 201 is mounted on base 1; The rotating disk 202 is fixedly mounted on the rotating end of the rotating motor 201; At least one magnetic positioning block 203 is uniformly installed on the rotating disk 202; the magnetic positioning block 203 is used to integrate the ignition coil on the rotating disk 202.

[0017] Compared with existing technologies, this invention has several advantages: The method acquires point cloud data through full-domain scanning of a preset trajectory, and employs splicing and fitting strategies such as Gaussian filtering for noise reduction, iterative nearest-point registration, and B-spline surface fitting to generate a digital 3D model that is completely consistent with the coordinate system of the welding-driven motion, thus completely solving the weld seam offset problem caused by coordinate deviations during traditional manual path planning. The contour recognition strategy combines region growing, feature extraction, and template comparison to accurately locate the effective weld seam area to be welded. The centerline is fitted using the least squares method as the reference trajectory, ensuring a high degree of fit between the trajectory and the actual weld seam. The node setting strategy selectively deploys forced and uniform nodes, with strict control over coordinate continuity, providing reliable support for precise positioning of the weld joint, effectively improving weld seam position accuracy and contour regularity, and reducing defects such as incomplete penetration and undercut.

[0018] This innovative method employs a dual correction mechanism of molten pool monitoring and node feedback. The thermal imaging images of the molten pool acquired by the thermal imager undergo noise reduction, enhancement, geometric correction, and threshold segmentation to accurately extract the center coordinates and real-time temperature of the molten pool. Combined with a three-level deviation judgment, differentiated real-time correction commands are generated. For minor and severe anomalies, the robotic arm trajectory, welding power, and speed are adjusted to ensure the molten pool quickly returns to a normal state. After each node welding is completed, the forming and temperature data are extracted through node image processing. Combined with historical correction data, errors are corrected, and adjustment commands for the next node are generated to compensate for minor deviations at the current node, avoiding deviation accumulation. This achieves closed-loop control of "real-time monitoring - deviation judgment - precise correction - re-inspection and optimization," ensuring that the weld joint forming size and temperature meet preset standards, significantly improving the welding qualification rate.

[0019] The supporting welding equipment adopts an integrated ignition coil disk design, which can simultaneously carry multiple ignition coils to be welded. Combined with a rotating motor-driven carrying mechanism, it achieves batch positioning and continuous conveying of ignition coils, eliminating the need for frequent manual loading and unloading. The multi-degree-of-freedom robotic arm welding mechanism can automatically complete the welding of all ignition coils according to a planned path. The node spacing and welding parameters are adjusted according to the weld seam characteristics, enabling efficient welding of straight sections and precise adaptation of irregular sections, significantly improving production cycle time. Attached Figure Description

[0020] Figure 1 This is a schematic flowchart of the pre-reference test method for winding of automotive electronic ignition coils according to the present invention.

[0021] Figure 2 This is a schematic diagram of the winding device for automotive electronic ignition coils according to the present invention.

[0022] Figure 3 for Figure 2 A schematic diagram of the structure of the ignition coil bearing mechanism and the robotic arm welding mechanism.

[0023] Figure 4 for Figure 3 Enlarged view of point A in the middle; Figure 5 for Figure 2 Schematic diagram of the structure of the ignition coil support mechanism; Figure 6 for Figure 5 Top view. Detailed Implementation

[0024] The technical solutions of the welding path prioritization method and welding equipment for automotive electronic ignition coils provided by the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0025] Example 1 like Figure 1 As shown, a welding path priority method for automotive electronic ignition coils includes: performing a full-domain scan of the ignition coil surface and the area to be welded according to a preset scan time and preset scan trajectory, obtaining the three-dimensional coordinates of each scan point and forming point cloud raw data; and processing the point cloud raw data through a stitching fitting strategy to obtain a digital three-dimensional model of the ignition coil.

[0026] Furthermore, the stitching and fitting strategy includes: using a Gaussian filtering algorithm to smooth the original point cloud data, constructing a Gaussian distribution function, performing a weighted average calculation on the three-dimensional coordinates of each scan point and its surrounding preset number of neighboring points to obtain the coordinate deviation value of each scan point; and identifying and removing sampling points whose coordinate deviation values ​​exceed the noise point determination threshold as noise points.

[0027] Specifically, firstly, a Gaussian distribution function is constructed in three-dimensional space. Taking each scanning point as the center, a preset number of neighboring points are selected, and the weight coefficient of each neighboring point is calculated through the Gaussian distribution function.

[0028] The Gaussian distribution function is: , in, The Gaussian weight coefficients for the neighborhood points; The three-dimensional coordinates of the neighboring points; The three-dimensional coordinates of the center scan point to be smoothed; The standard deviation is Gaussian. The larger the value, the stronger the smoothing effect; conversely, the smaller the value, the richer the point cloud details are preserved.

[0029] Then, a weighted average calculation is performed on the three-dimensional coordinates of the center scanning point and its neighboring points to obtain the corrected coordinates of the center scanning point after smoothing.

[0030] The formula for calculating the corrected coordinates of the center scan point after smoothing is: , in, Corrected 3D coordinates of the center scan point after smoothing; The number of neighboring points selected is preset. ; For the first Gaussian weight coefficients corresponding to each neighboring point; For the first The three-dimensional coordinates of the neighboring points; denominator This is the weighted summation and normalization term, used to avoid the cumulative bias of weights affecting the accuracy of the corrected coordinates.

[0031] Simultaneously, the difference between the original coordinates and the corrected coordinates of each scan point is calculated, which is the coordinate deviation value of that scan point.

[0032] The formula for calculating the coordinate deviation of the scan point is: , in, This represents the coordinate deviation value of the scanned point; The original three-dimensional coordinates of the scan point; The corrected 3D coordinates of the scan points after smoothing.

[0033] The splicing and fitting strategy further includes: selecting a segment of the point cloud covering the reference features of the ignition coil as the target point cloud, and using the point clouds obtained from the scanning of the remaining segments as source point clouds; iteratively calculating the rotation and translation matrices between the source point cloud and the target point cloud, and in each iteration, selecting the sampling point in the source point cloud and the nearest point in the target point cloud for matching, calculating the distance error between the two points, until the distance error is less than the preset registration accuracy threshold, stopping the iteration and determining the optimal coordinate transformation relationship; based on the optimal coordinate transformation relationship, uniformly transforming all source point clouds to the coordinate system of the target point cloud, so that the coordinate system is consistent with the motion coordinate system of the welding drive.

[0034] Specifically, firstly, a segment of the point cloud covering the reference features of the ignition coil is selected as the target point cloud. The reference features are selected from structurally stable and easily identifiable areas such as the root of the ignition coil's terminals and the housing's positioning groove. These areas have high point cloud density and low noise, ensuring reliable registration reference. The point clouds obtained from the remaining scan segments are used as source point clouds. ,in, , The number of source point cloud segments is set according to the size of the ignition coil and the scanning range.

[0035] For each segment of source cloud The iterative nearest-point algorithm is used to iteratively calculate the rotation matrix between the source point cloud and the target point cloud. Translation matrix This enables precise matching between the source point cloud and the target point cloud.

[0036] The coordinate transformation formula for converting the source point cloud to the target point cloud coordinate system is: , in, Source Point Cloud The Middle The three-dimensional coordinates of each sampling point; These are the three-dimensional coordinates of the sampling point after transformation to the target point cloud coordinate system; for The rotation matrix is ​​used to correct the pose deviation between the source point cloud and the target point cloud. for The translation matrix is ​​used to correct the positional deviation between the source point cloud and the target point cloud.

[0037] The formula for calculating distance error is: , in, Source point cloud sampling points The distance error between the transformed point and the nearest point in the target point cloud; For target point cloud Zhongyu The three-dimensional coordinates of the nearest sampling point; It is the Euclidean norm, used to calculate the straight-line distance between two points.

[0038] Calculate the average distance error of all matching point pairs in the current iteration round. in, Set a preset registration accuracy threshold for the number of point pairs matched in the current iteration. According to the welding precision requirements of automotive electronic ignition coils, Usually set to ,like If the iteration stops, determine the current rotation matrix. Translation matrix The optimal coordinate transformation relationship; if Then reselect the matching point pair and adjust the rotation matrix. Translation matrix Repeat the above iterative process until the iteration termination condition is met.

[0039] After determining the optimal coordinate transformation relationship, all source point clouds are transformed according to the above coordinate transformation formula. Unified conversion to target point cloud In the coordinate system, the unified coordinate system is kept consistent with the motion coordinate system of the welding drive, that is, the origin and coordinate axis directions of the coordinate system are completely corresponding. This ensures that the welding path planned based on the digital 3D model can be directly adapted to the motion control of the welding robot and avoids welding path deviation caused by coordinate system deviation.

[0040] The splicing and fitting strategy further includes: dividing the global point cloud into multiple fitting regions based on the structural characteristics of the ignition coil, performing surface fitting on each region, and then integrating the fitted surfaces of each region into a complete three-dimensional surface model through a surface splicing algorithm; when the fitting error between the fitted surface and the original point cloud does not exceed a preset fitting error threshold, a digital three-dimensional model is generated; the coordinate system of this digital three-dimensional model is the same as the motion coordinate system of the welding drive.

[0041] Specifically, based on the structural characteristics of the automotive electronic ignition coil, the registered global point cloud is divided into multiple fitting regions. The division is based on the actual structure of the ignition coil and is specifically divided into three regions: the curved surface region of the ignition coil shell, the curved surface region of the winding end, and the curved surface region of the terminal connection. Each region is fitted separately to avoid mutual interference between different structural surfaces and improve the fitting accuracy.

[0042] For each fitting region, a B-spline surface fitting algorithm is used for surface fitting. The formula for B-spline surface fitting is: , in, For the three-dimensional coordinate function of the B-spline fitted surface; , They are respectively direction, The number of vertices controlling the direction; for direction B-spline basis functions; for direction B-spline basis functions; The three-dimensional coordinates of the control vertices of the B-spline surface; The parameter variable has a range of values. It covers the entire fitted region.

[0043] After fitting the surface of a single region, the fitting error between the fitted surface of that region and the original point cloud is calculated. The formula for calculating the fitting error is as follows: , in, This represents the average fitting error of the fitted region; This represents the number of original point cloud sampling points in the fitted region; For the region The three-dimensional coordinates of the original point cloud sampling points; To fit the surface with The corresponding three-dimensional coordinates of the point; Given the Euclidean norm, calculate the distance deviation between the original point cloud and the fitted surface.

[0044] A preset fitting error threshold is set based on the accuracy requirements of the ignition coil welding path planning. , for If the average fitting error of a single region If the surface fitting in that region is satisfactory, then the surface fitting in that region is deemed acceptable. If the B-spline basis function degree is adjusted or the number of vertices is controlled, the surface fitting is repeated until the fitting error requirement is met.

[0045] In this embodiment, after all fitted regions have passed the fitting test, a surface stitching algorithm is used to integrate the fitted surfaces of each region into a complete three-dimensional surface model. During the stitching process, it is ensured that the joints between adjacent surface regions are smooth and gapless, and that the normal directions at the joints are continuous, to avoid stitching deviations affecting the accuracy of subsequent weld contour recognition. The overall average fitting error of the complete three-dimensional surface model does not exceed a preset fitting error threshold. At that time, a digital three-dimensional model of the ignition coil is officially generated; the coordinate system of this digital three-dimensional model is the same as the motion coordinate system of the welding drive, and can be directly used for subsequent identification of the weld seam waiting area, welding path planning and node parameter setting.

[0046] The method further includes: identifying the weldable area in the digital 3D model using a contour recognition strategy, extracting the center contour line of each weld segment, and using this contour line as the reference trajectory for the welding path. Based on the shape of the weld, the movement direction, start point, and end point of each welding path segment are defined; and based on a node setting strategy, welding execution data for each path node is set on each welding path segment.

[0047] Furthermore, the contour recognition strategy includes: smoothing and denoising the digital 3D model, combining the preset structural features of the automotive electronic ignition coil, and using a region growing algorithm to segment the digital 3D model into regions such as the outer shell region, winding region, terminal region, and suspected weld region; wherein, the suspected weld region is the connecting transition region between different structural regions.

[0048] Specifically, the core purpose of smoothing and denoising the digital 3D model is to eliminate residual minor errors from the surface fitting process and redundant noise on the model surface, thus avoiding interference with region segmentation and weld identification accuracy. Considering the surface characteristics of the digital 3D model, the Laplacian smoothing algorithm is used for denoising. This algorithm can smooth out localized minor protrusions or depressions while preserving the overall model structure and the characteristics of the weld area. The specific smoothing formula and process are as follows: The Laplace smoothing formula is: , in: For the first in the digital 3D model Smoothed 3D coordinates of each vertex ; For the first Number of adjacent vertices of each vertex indivual; For the first The vertex of the first vertex The original 3D coordinates of the adjacent vertices ; The average weighting coefficient is used to ensure that the vertex coordinates after smoothing fit the original model contour, avoiding excessive smoothing that could lead to the loss of weld features.

[0049] During the smoothing process, the number of smoothing iterations is set. In each iteration, all vertices of the model are smoothed using the formula described above. After each iteration, the coordinate deviation between the smoothed vertices and the original vertices is calculated. The calculation formula is as follows: , like If the result is satisfactory, the smoothing and denoising is deemed acceptable; if the result exceeds the threshold, the number of iterations is increased, and the smoothing is repeated until the requirements are met.

[0050] After smoothing and denoising, the digital 3D model is segmented into regions based on the preset structural features of the automotive electronic ignition coil and a region growth algorithm. This is done according to the growth criterion of the similarity of the model surfaces, dividing the model into different structural regions.

[0051] First, seed points are selected. Typical characteristic points from the ignition coil housing, winding ends, and terminals are chosen as initial seed points. The principle for selecting seed points is to ensure structural stability and distance from transitional regions, thereby ensuring accurate growth direction in each region. Secondly, the similarity determination formula is as follows: , in, For the model The vertex and the first The surface similarity of the vertices, with values ​​ranging from 0 to 1. The closer the similarity is to 1, the higher the similarity. , The first , The unit normal vector of each vertex; The dot product of the normal vectors is used to determine the similarity of the orientation of the surfaces; , Let be the magnitude of the normal vector; The Euclidean distance between the two vertices; This is the distance weighting coefficient, used to adjust the impact of distance on similarity.

[0052] Finally, region growing and segmentation are performed. A similarity threshold is set. Centered on each initial seed point, the similarity is sequentially... Adjacent vertices are included in the corresponding region until no new vertices can be included; the remaining vertices not included in the above three regions are all classified as suspected weld seam regions. This region is the connecting transition area between the shell region, winding region, and terminal region. Its surface features and normal vector direction are between adjacent structural regions, and it is the core candidate range for the weld seam to be welded.

[0053] The contour recognition strategy further includes: extracting three-dimensional features from suspected weld areas. The extracted features include the region's normal vector direction, surface curvature, contour edge gradient, and thickness parameter. The normal vector direction is used to distinguish the orientation difference between the weld area and adjacent structural areas, and there is a significant angle between the normal vector direction of the weld area and the adjacent structural areas. The surface curvature is used to identify the concave or convex features of the weld area. The contour edge gradient is used to capture the boundary contour of the weld area, and the difference between the edge gradient of the weld boundary and the adjacent structure is greater than a preset gradient threshold. The thickness parameter is used to filter invalid suspected areas and retain areas with the same thickness as the preset weld width as target areas to be identified.

[0054] Specifically, the normal vector direction is extracted by calculating the unit normal vector of each vertex within the suspected weld area using the covariance matrix method. The calculation formula is: First, select each vertex. Construct a neighborhood set of adjacent vertices. Calculate the covariance matrix of the neighborhood point set. The formula is: , in, The average coordinates of the neighborhood point set. ; For the first The deviation vector between each neighboring point and the average coordinate; This is the transpose of the deviation vector.

[0055] For covariance matrix Eigenvalue decomposition yields three eigenvalues. This corresponds to three eigenvectors, where the smallest eigenvalue is... The corresponding eigenvector is the unit normal vector of that vertex. .

[0056] The purpose of the normal vector direction is to distinguish the orientation difference between the weld area and adjacent structural areas. A threshold for the included angle of the normal vector is set. If the angle between the normal vector of the suspected region vertex and the normal vector of the adjacent structural region vertex... If the vertex is positive, it is determined to belong to the weld-related area; otherwise, it is determined to be an invalid transition area vertex.

[0057] The surface curvature extraction method uses the principal curvature method to calculate the surface curvature of the suspected weld area. The core of this method is to extract the principal curvature of each vertex. and Gaussian curvature The calculation formula is: , , , in, , , These are the covariance matrices. The three eigenvalues; These are the two principal curvatures of the vertex, which reflect the degree of curvature of the surface in the two perpendicular directions at that point; is the Gaussian curvature, which is used to reflect the overall bending characteristics of the surface.

[0058] The purpose of surface curvature is to identify the depressions or bulges in the weld area. Weld areas are mostly characterized by slight depressions or gentle bulges; therefore, a Gaussian curvature threshold is set. If the Gaussian curvature of the suspected region vertex If the vertex is within the specified range, it is considered a candidate vertex for the weld; otherwise, it is considered an invalid vertex.

[0059] The Sobel gradient algorithm is used to extract the contour edge gradient of the suspected weld area, which is used to capture the boundary contour of the weld area.

[0060] First, the three-dimensional coordinates of the suspected weld area are projected onto a two-dimensional plane, i.e., the projection plane is perpendicular to the weld direction, to obtain the two-dimensional coordinates. Then calculate the gradient magnitude of the two-dimensional projected image. The calculation formula is as follows: , in, for gradient components in the direction, ; for gradient components in the direction, ; The grayscale value is the value of the two-dimensional projected image.

[0061] The purpose of the contour edge gradient is to capture the weld boundary. A preset gradient threshold is set based on the grayscale range of the projected image. If the gradient magnitude at a certain point If the point is selected, it is determined to be a candidate point for the weld boundary; simultaneously, the edge gradient difference between the weld boundary and the adjacent structure is calculated. ,like If it is, then it is determined to be a valid weld boundary point.

[0062] Thickness parameters are extracted. The thickness parameter is the maximum thickness of the suspected weld area, calculated using the two-point distance method. The calculation formula is as follows: , in, The thickness of the suspected weld area; , These are the two endpoints along the thickness direction within the suspected region; , These are the three-dimensional coordinates of the two endpoints.

[0063] The thickness parameter is used to filter out invalid or suspected areas. First, set the preset weld width. If the thickness of the suspected area If the area is within the specified range, it will be retained as the target area to be identified; otherwise, it will be determined as an invalid suspected area.

[0064] The contour recognition strategy further includes: comparing the extracted features of the suspected weld area with a preset weld feature template; when the similarity between the features of the suspected weld area and the preset weld feature template is greater than or equal to a preset comparison threshold, the area is determined to be a valid weld area to be welded.

[0065] Specifically, the preset weld feature template is a set of three-dimensional features of the standard weld seam of an automotive electronic ignition coil. These correspond to the standard normal vector direction, standard principal curvature, standard edge gradient, and standard thickness parameters, respectively. The cosine similarity algorithm is used to calculate the overall similarity between the suspected region features and the template features. The calculation formula is: , in, The overall feature similarity has a value range of 100%. ; These are the weight coefficients for each feature. Based on the priority setting for weld identification, typically ; Extraction features for suspected weld seam areas; Standard features for preset weld feature templates.

[0066] Set preset comparison threshold Preset comparison threshold The threshold is adjusted based on welding precision requirements; the higher the precision requirement, the higher the threshold. This applies to the overall feature similarity of suspected weld areas. When this occurs, the area is determined to be a valid weldable area; if If the region is invalid, it will be removed.

[0067] The contour recognition strategy further includes: using a contour extraction algorithm to extract the boundary contour lines of the effective weld seam to be welded area; and using coordinate fitting calculation to obtain the center line of the boundary contour lines on both sides, and using the center line as the center contour line of each weld seam segment.

[0068] Specifically, starting from one boundary vertex of the effective weld area, adjacent boundary vertices are selected sequentially until the starting point is reached, forming a closed or continuous boundary profile. If the weld is non-closed, the two endpoints are extracted to form a continuous open boundary profile, ultimately obtaining the two side boundary profiles of the weld area to be welded. .

[0069] The centerlines of the two boundary contour lines are obtained through coordinate fitting calculation, and these centerlines are used as the center contour lines of each weld segment. The least squares method is used to fit the coordinates of the two boundary contour lines, yielding the fitting equations for each contour line. Then, the centerlines of the two fitted lines are calculated using the following formula: First, boundary contour fitting is performed. Taking two-dimensional fitting as an example, three-dimensional fitting follows the same principle and needs to be combined with… (Coordinate correction). Let the coordinates of the vertices of one side of the boundary contour line be... A linear fitting suitable for straight welds or a quadratic curve fitting suitable for circular arc welds is adopted. The linear fitting equation is: , where the coefficient The solution formulas are as follows: , , in, The number of vertices of the boundary contour line; For the first Two-dimensional coordinates of each vertex; The slope of the fitted line; The intercept of the fitted line is denoted as .

[0070] Next, the centerline is calculated. Let the fitting equations for the two boundary contour lines be: , .

[0071] Because the parallelism of the contour lines on both sides of the weld is high, that is When, the fitting equation for the midline is

[0072] , .

[0073] For the 3D contour lines, additional information is needed. Calculation of the median of the coordinate system. ,in, Points corresponding to the two side contour lines coordinate.

[0074] After fitting, the vertex coordinates of the centerline are extracted to form a continuous center contour line. This center contour line is the welding path reference trajectory for each weld segment. Its coordinate system is consistent with the digital 3D model and the motion coordinate system of the welding drive, providing a precise trajectory reference for subsequent welding path motion direction definition and node setting.

[0075] In this embodiment, the contour recognition strategy extracts the center contour line of the weld as the reference trajectory for the welding path. Through a standardized contour fitting process, it ensures the accuracy and continuity of the reference trajectory, providing reliable support for subsequent welding path planning. After the effective weld area is identified, the contour lines of the two side boundaries are obtained through a contour extraction algorithm. Then, the least squares method is used for coordinate fitting, combined with two-dimensional and three-dimensional coordinate correction, to accurately calculate the centerline of the two side boundaries as the center contour line. This contour line maintains consistency with the digital three-dimensional model and the welding drive motion coordinate system, ensuring a high degree of fit between the reference trajectory and the actual weld.

[0076] The method further includes: defining the movement direction, start point, and end point of each welding path segment according to the shape of the weld; and setting welding execution data for each path node on each welding path segment according to the node setting strategy. The welding execution parameters are three-dimensional coordinates, preset welding parameters, standard parameters for welding feedback, and feedback deviation thresholds.

[0077] Furthermore, the node setting strategy includes: path nodes are set at preset node spacing for weld inflection points, weld width change points, the start and end points of irregular structures, and straight and uniform sections of welds.

[0078] Specifically, path nodes are set at preset node intervals for weld inflection points, weld width change points, the start and end points of irregular structures, and straight and uniform weld sections. Weld inflection points are points where the tangent direction changes abruptly in the weld centerline contour, and weld width change points are points where the weld width... Exceeding the preset fluctuation range The starting and ending points of the irregular structure are the connection points between the irregular weld and the straight or circular arc weld. These nodes are all forcibly placed nodes and do not need to be set according to the spacing, ensuring the welding accuracy at the abrupt changes in the weld structure. For the straight and uniform sections of the weld, nodes are placed according to the preset node spacing. The spacing is set according to the welding accuracy requirements, typically 1.0-2.0mm. The node spacing can be fine-tuned according to the weld size, and the adjusted spacing must meet the following requirements. .

[0079] The formula for calculating the number of nodes in a straight segment is: , in, This represents the total number of nodes on the straight path segment. The length of the straight section of the weld is calculated from the distance between two points on the center outline of the weld. Preset node spacing; The floor symbol, for example This ensures that no nodes are omitted or left unattended.

[0080] The welding execution parameters are three-dimensional coordinates, preset welding parameters, standard parameters for welding feedback, and feedback deviation thresholds. These four parameters are interconnected and work together to form a complete welding control basis for each path node, ensuring that the welding process is controllable, feedback-enabled, and correctable.

[0081] The node setting strategy also includes: extracting the three-dimensional coordinates of each path node from the welding execution parameters on the weld center contour line. These three-dimensional coordinates are consistent with the digital three-dimensional model and the motion coordinate system of the welding drive, and serve as the target positioning coordinates of the welding head during node welding. This ensures that the welding head can move accurately to the preset position of each node, providing a positioning reference for the accurate execution of the welding path. The three-dimensional coordinates of each node are unique and continuous with the coordinates of adjacent nodes, avoiding welding path offset caused by node coordinate discontinuity.

[0082] Specifically, for each path node, its three-dimensional coordinates in the welding execution parameters on the weld center contour line are extracted. These three-dimensional coordinates are consistent with the digital 3D model and the motion coordinate system of the welding drive, serving as the target positioning coordinates of the welding head during node welding. This ensures that the welding head can accurately move to the preset position of each node, providing a positioning reference for the precise execution of the welding path. Coordinate extraction uses interpolation; if the node is a vertex of the contour line, the vertex's three-dimensional coordinates are directly extracted. If the nodes are non-vertex nodes arranged at intervals on straight segments, then the coordinates are calculated using linear interpolation, with the following formula: , in, For the first The three-dimensional coordinates of a non-vertex path node; For the first The three-dimensional coordinates of the previous neighboring node of each node; , , This represents the coordinate difference between two adjacent vertex nodes; These are the interpolation coefficients. ,in For the first The actual distance between each node and the previous vertex node (equal to the preset distance) ), This represents the distance between two adjacent vertex nodes.

[0083] Each node has unique 3D coordinates and is continuous with the coordinates of its neighboring nodes. The formula for determining the continuity of coordinates between neighboring nodes is as follows: , in, For the first The node and the first The actual distance between each node; For the first The three-dimensional coordinates of each node; The maximum allowable deviation of the spacing is defined by this formula. If the formula is satisfied, the coordinates are considered continuous. Otherwise, the node coordinates are adjusted to avoid node coordinate discontinuity, which could lead to welding path offset and defects in the welding joint.

[0084] The node setting strategy also includes defining preset welding parameters, welding feedback standard parameters, and feedback deviation thresholds for each path node in the welding execution parameters. The preset welding parameters include welding speed, welding power, and welding torch angle. For adjacent nodes on the same weld seam, the preset welding parameters are adjusted in a gradient according to the differences in weld seam structure to ensure that the welding parameters are consistent with the weld seam characteristics.

[0085] Specifically, the wider and more curved the weld, the slower the welding speed, ensuring a fully formed molten pool; conversely, the narrower the weld width and the smaller the curvature, the faster the welding speed, improving welding efficiency. For adjacent nodes on the same weld, the preset welding speed is adjusted gradient according to the weld width variation. The gradient adjustment formula is: , in, For the first The preset welding speed for each node; For the first The preset welding speed for each node; For the first Weld width at each node; For the first The weld width at each node. This formula ensures that the welding speed is inversely proportional to the weld width; as the width increases, the speed gradient decreases, and as the width decreases, the speed gradient increases, avoiding defects such as uneven penetration and undercut caused by sudden speed changes. The preset welding speed range is typically [value missing]. The midpoint of the velocity at the nodes of a straight, uniform segment is taken. The velocity at the inflection point and the point of width change is minimized. .

[0086] Welding power is positively correlated with welding speed and weld width. Higher speed and wider welds require higher power to ensure sufficient weld pool temperature; conversely, slower speed and narrower welds require lower power to avoid defects such as burn-through and overheating. The formula for adjusting the power gradient between adjacent nodes is: , in, For the first Preset welding power for each node; For the first Preset welding power for each node; The first , Welding speed of each node; The first , The weld width at each node. The preset welding power value range is typically [value missing]. Fine-tune according to the ignition coil material; for copper, use the upper limit. Aluminum alloys take the lower limit .

[0087] The welding torch angle is related to the direction of the normal vector of the weld center contour line. To ensure that the welding torch axis coincides with the center of the molten pool, the angle setting formula is: , in, For the first Preset welding torch angles for each node; For the first Unit normal vector of the weld surface at each node; For the first Welding velocity direction vector at each node; This is the inverse cosine function, used to calculate the angle between the normal vector and the velocity direction vector. The welding torch angle typically ranges from [value missing]. Straight segment nodes are taken Inflection points and arc segment nodes are finely adjusted according to the direction of the normal vector. .

[0088] For adjacent nodes on the same weld, the preset welding parameters are adjusted in a gradient according to the differences in weld structure to ensure that the welding parameters are the same as the weld characteristics, avoid weld formation defects caused by parameter abrupt changes, and ensure that the entire weld is uniform in shape and consistent in strength.

[0089] The feedback standard parameters are consistent with the feedback standard parameters in the welding execution parameters, including node weld point forming standard parameters and node temperature standard parameters. The weld point forming standard parameters include preset weld point width, preset weld point height, and preset penetration depth, which are the same as the preset weld size. The node temperature standard parameters are preset optimal welding temperature ranges, used to provide a clear benchmark for comparison of feedback data after node welding is completed. The feedback deviation threshold is consistent with the feedback deviation threshold in the welding execution parameters and corresponds to the node feedback standard parameter settings. The feedback deviation threshold includes coordinate deviation threshold, forming deviation threshold, and temperature deviation threshold. The coordinate deviation threshold is used to determine the degree of offset between the weld pool center and the node's three-dimensional coordinates. The forming deviation threshold is used to determine the degree of deviation between the node weld point forming size and the feedback standard parameters. The temperature deviation threshold is used to determine the degree of deviation between the node weld point temperature and the optimal temperature range.

[0090] Specifically, 1. Standard parameters for node solder joint formation: preset solder joint width Preset solder joint height It is 1 / 3 to 1 / 2 of the weld width, calculated using the following formula: Preset melt depth It is 1 / 2 to 2 / 3 of the weld width, calculated using the following formula: In the formula The width of the weld at the kth node is used to ensure that the forming parameters match the weld size and avoid underforming or overforming.

[0091] Set the node temperature standard parameters. Preset the optimal welding temperature range. The setting is based on the material of the ignition coil, i.e. , (copper) or (Aluminum alloy), where The melting point of the welding material is used to ensure that the welding temperature is above the melting point and below the overheating temperature, taking into account both the formation of the molten pool and the material properties.

[0092] The feedback deviation threshold is consistent with the feedback deviation threshold in the welding execution parameters and corresponds to the node feedback standard parameter setting. It is used to determine whether the feedback data during the welding process exceeds the reasonable range. If it does, the deviation correction mechanism is triggered. The feedback deviation threshold includes the coordinate deviation threshold, the forming deviation threshold, and the temperature deviation threshold.

[0093] Coordinate deviation threshold Used to determine the degree of offset between the center of the molten pool and the three-dimensional coordinates of the nodes, set to the preset node spacing. The calculation formula is: In the formula To preset the node spacing, typically To match the welding accuracy requirements mentioned above, if the actual deviation between the center of the molten pool and the three-dimensional coordinates of the node... This triggers the welding head position correction, adjusting it to the target coordinates.

[0094] The forming deviation threshold is used to determine the degree of deviation between the forming size of the node solder joint and the feedback standard parameters, including the width deviation threshold. Height deviation threshold Melt depth deviation threshold The calculation formula is: , , , in, These represent the preset width, height, and penetration depth in the standard parameters for weld joint formation; 0.1 is the deviation coefficient, which can be adjusted to 0.08-0.12 according to welding accuracy. If the actual formed size deviates from the standard parameters beyond the corresponding threshold, the welding power and speed will be corrected to ensure that the forming quality meets the standards.

[0095] Temperature deviation threshold Used to determine the degree of deviation between the temperature of the joint solder point and the optimal temperature range, and is set as the span of the optimal temperature range. The calculation formula is: In the formula To preset the upper and lower limits of the optimal welding temperature range, if the actual weld temperature deviates from the midpoint of the range by more than [a certain amount]... This triggers welding power adjustment, increasing or decreasing the power to control the temperature within the optimal range.

[0096] All feedback deviation thresholds are associated with and stored in relation to the corresponding node's feedback standard parameters and preset welding parameters. During the welding process, feedback data on coordinates, forming, and temperature are collected in real time and compared with the standard parameters to calculate the deviation. If the deviation exceeds the corresponding threshold, a correction mechanism is immediately triggered to adjust the welding head position and welding parameters to ensure that the welding quality of each node meets the standards, ultimately achieving precise and efficient welding of the automotive electronic ignition coil weld.

[0097] The node setting strategy also includes: numbering all path nodes on each welding path in an orderly manner according to the direction of welding movement, with each node corresponding to a unique number, and storing the welding execution data of each node in association with it.

[0098] Specifically, each weld segment is numbered in the format of "segment number-node number". For example, the fifth node of the first weld segment is numbered "1-5", and the third node of the second weld segment is numbered "2-3". The numbers are consecutive and without repetition, which facilitates accurate calling of nodes, data traceability and fault diagnosis during the welding process. The welding execution data is associated with the node number and stored in the welding control system. It is synchronized to the welding robot in real time to ensure that the corresponding parameters can be quickly called when welding each node.

[0099] In this embodiment, the node setting strategy ensures welding accuracy across the entire weld area through scientific and reasonable node placement. The strategy specifically selects weld inflection points, width change points, and other structurally abrupt locations as mandatory nodes. Straight sections are evenly distributed at preset intervals of 1.0-2.0mm. Combined with the node quantity formula and coordinate continuity judgment, this ensures no nodes are missed or discontinuous, avoiding welding deviations at structurally abrupt locations and guaranteeing a continuous welding trajectory in straight sections, providing a reliable basis for precise weld joint positioning. Precise welding execution parameters reduce weld formation defects. Parameter settings are tailored to the weld structure and material characteristics. Welding speed and power are adjusted according to weld width and curvature gradient, and the welding torch angle matches the contour normal vector, avoiding uneven penetration and burn-through caused by parameter abrupt changes. Standard parameters for weld formation and temperature, along with feedback deviation thresholds, clearly define quality judgment criteria, enabling controllable and feedback-enabled welding processes. Node numbers and data are stored in association, improving process standardization and traceability. Using a unique "segment number-node number" identification system, welding execution data is bound to each node, facilitating rapid parameter retrieval, fault diagnosis, and quality traceability. This adapts to the standardized requirements of mass production and provides accurate data support for subsequent molten pool correction and feedback comparison, helping to form a closed-loop control system.

[0100] The method also includes: performing welding sequentially along the path nodes according to the welding path and welding execution parameters.

[0101] In this embodiment, during the welding process, the welding robot arm strictly follows the direction of the weld center contour and the node setting rules, completing the welding operation of each node sequentially along the path. The overall welding process is consistent with the node numbering sequence and welding movement direction, that is, starting from the starting node of each weld segment and advancing point by point to the ending node. Adjacent weld segments transition smoothly at the connecting nodes to avoid defects such as weld breaks and joint misalignment. For different types of welds, an appropriate point-by-point welding method is adopted: straight welds are welded point by point at a uniform speed to maintain stable welding speed and power; for arc welds, the movement posture of the welding robot arm is adjusted in real time according to the curvature changes at the nodes, and the fine-tuning requirements of the welding gun angle are adapted to ensure that the welding posture of each node fits the weld surface; for irregular welds, they are welded point by point in segmented node sequence. After each segment is completed, the welding quality of the connecting nodes is checked, and the next segment is welded only after confirmation that there are no errors.

[0102] During welding at each node, the welding control system will call upon the welding execution parameters associated with that node in real time, including three-dimensional positioning coordinates, preset welding speed, welding power, welding torch angle, as well as feedback standard parameters and feedback deviation thresholds, to achieve precise control of the node welding. After the welding head moves to the target node, it will remain there for a preset time to ensure that the molten pool is fully formed. At the same time, the temperature detection module and the forming detection module will collect the welding feedback data of that node in real time, including the actual coordinates of the molten pool center, the weld point forming size, and the actual temperature of the weld point.

[0103] The collected feedback data is transmitted to the control system in real time and compared with the feedback standard parameters of that node. The actual deviation is calculated using the deviation calculation formula set earlier, and it is determined whether the deviation exceeds the corresponding feedback deviation threshold. If the deviation does not exceed the threshold, the welding of that node is deemed qualified, and the welding robot arm will smoothly move to the next node according to the preset node spacing and movement direction, repeating the above welding process. If the deviation exceeds the threshold, the deviation correction mechanism is immediately triggered, and precise adjustments are made according to the deviation type: when the coordinate deviation exceeds the threshold, the position of the welding robot arm is adjusted to correct it to the preset three-dimensional coordinates of the node; when the forming deviation exceeds the threshold, the welding power and speed of the next node are finely adjusted to compensate for the forming defects of the current node; when the temperature deviation exceeds the threshold, the welding power is adjusted in real time to control the weld temperature within the optimal range, and welding continues after the correction is completed.

[0104] The method further includes: acquiring thermal imaging images of the molten pool between every two path nodes at a preset acquisition frequency, and obtaining the coordinates and temperature of the molten pool center through a path image processing strategy; comparing the coordinates and temperature of the molten pool center with preset molten pool data through a molten pool offset monitoring and correction strategy, and generating real-time correction instructions.

[0105] Specifically, firstly, preset the thermal imaging image acquisition parameters for the molten pool. The acquisition frequency needs to be adapted to the welding speed and node spacing to ensure that a sufficient number of thermal imaging images can be acquired between every two path nodes, achieving full coverage of the molten pool state. The formula for setting it is: , in, Preset the acquisition frequency for thermal imaging images; The average welding speed of the current welding segment is calculated by taking the average welding speed of all nodes in that segment. ,in, This represents the total number of nodes in this segment. Preset node spacing; The minimum interval for monitoring the molten pool between adjacent nodes is set to ensure that at least 3 frames of thermal imaging images are acquired between each node segment, thereby enabling continuous monitoring of the molten pool status. To ensure the sampling frequency is an integer and the rounding is rounded up, the thermal imaging camera's shooting angle is kept consistent with the welding torch angle during acquisition, covering the entire molten pool area and its surroundings. The weld area is designed to ensure that the outline and location information of the molten pool can be fully captured.

[0106] Furthermore, the path image processing strategy includes: performing noise reduction, enhancement, and geometric correction preprocessing on the acquired molten pool thermal imaging image to obtain the preprocessed molten pool thermal imaging image and geometric correction parameters.

[0107] Specifically, the noise reduction process uses a median filtering algorithm to eliminate isolated noise points in the molten pool thermal imaging image while preserving the temperature characteristics and contour information of the molten pool region. The filtering formula is as follows: , in, To the denoised image in coordinates The grayscale value at that location (grayscale value is positively correlated with temperature; the higher the grayscale value, the higher the temperature). For the original thermal imaging image in coordinates The grayscale value at that location; This is the filter kernel size parameter (usually 1 or 2, i.e., 3×3 or 5×5 filter kernels). The median operation takes the median of all pixel grayscale values ​​within the area covered by the filter kernel as the corrected grayscale value for the current pixel. This algorithm can effectively suppress impulse noise while avoiding excessive smoothing that leads to blurred melt pool contours.

[0108] Image enhancement uses a grayscale stretching algorithm to increase the difference between the solder joint forming area and the background area, facilitating subsequent region segmentation. The enhancement formula is as follows: , in, To enhance the image in The grayscale value at the specified location (range 0-255, standard grayscale range); This represents the grayscale value at the corresponding location in the denoised image. This represents the minimum grayscale value of the image after noise reduction. This represents the maximum grayscale value of the image after noise reduction. This formula maps the original grayscale range to 0-255 through linear stretching, significantly improving the grayscale difference between the molten pool area and the background.

[0109] Geometric correction eliminates image geometric distortion based on camera calibration parameters, ensuring that the image coordinates are consistent with the actual molten pool coordinates. The correction formula and geometric correction parameters are explained as follows: , in, These are the pixel coordinates of the corrected image; The pixel coordinates of the image before correction of distortion; These are geometric correction parameters, obtained through pre-calibration of the camera. , For the camera intrinsic parameter matrix, This represents the camera distortion parameter matrix; 1 represents the homogeneous coordinate coefficients, used to unify the coordinate transformation dimensions. Geometric correction parameters must match the current welding scene, be calibrated, stored in the control system, and synchronously retrieved during image acquisition.

[0110] The path image processing strategy also includes: using an adaptive threshold segmentation algorithm and combining it with the grayscale distribution characteristics of the molten pool thermal imaging image to divide the preprocessed molten pool thermal imaging image into molten pool region and non-molten pool region.

[0111] Specifically, first, the mean grayscale value of the preprocessed image is calculated. and grayscale standard deviation , which serves as the basic parameter for threshold calculation: , , in, The mean gray level of the preprocessed image; The grayscale standard deviation of the preprocessed image; Image width; Image height; For the preprocessed image in The grayscale value at that location.

[0112] Secondly, an adaptive segmentation threshold is set based on the characteristics of the molten pool temperature. The formula is: , in, This is the threshold for separating the molten pool region from the non-molten pool region; The threshold adjustment coefficient ensures that the molten pool area is completely preserved after segmentation, while removing interference areas such as electric arc light and spatter in the background.

[0113] Segmentation rule: grayscale value The area is identified as the molten pool region, with a grayscale value. The region was determined to be a non-molten pool region, and a binarized image of the molten pool region was finally obtained. .

[0114] The path image processing strategy also includes: using a fitting algorithm to perform contour fitting on the segmented molten pool region, obtaining the contour boundary coordinates of the molten pool region, and calculating the geometric center of the contour boundary, defining the geometric center as the image pixel coordinates of the molten pool center; based on geometric correction parameters, converting the pixel coordinates of the molten pool center into the actual coordinates of the molten pool center, which are consistent with the 3D coordinates of the nodes in the digital 3D model, the motion coordinate system driven by welding, and the node setting strategy; the coordinates of the molten pool center are used to compare with the 3D coordinates of the nodes to determine the degree of molten pool offset.

[0115] Specifically, the least squares method is used to fit the molten pool profile to the binary molten pool region. To perform contour fitting, first extract the outer boundary contour points of the molten pool region. ,in, , This represents the total number of boundary contour points. Then, the elliptical equation for the molten pool contour is fitted.

[0116] The equation for fitting the ellipse profile of the molten pool is: .

[0117] The coefficients are solved using the least squares method, and the objective function is constructed. ,make Find the minimum value and obtain the coefficients by solving the system of equations. Finally, the precise coordinates of the molten pool outline boundary are obtained. .

[0118] The geometric center of the molten pool contour boundary is calculated using the centroid method, and the image pixel coordinates of the molten pool center are used as the centroid method. The calculation formula is: , in, The coordinates of the center of the molten pool are two-dimensional image coordinates; This represents the total number of pixels within the molten pool outline region; Let be the two-dimensional image coordinates of the i-th pixel within the molten pool contour region.

[0119] Based on geometric correction parameters (Camera distortion correction matrix) and camera intrinsic parameter matrix Convert the image pixel coordinates of the molten pool center into actual three-dimensional coordinates. The conversion formula is: , in, The actual three-dimensional coordinates of the molten pool center (maintaining consistency with the digital three-dimensional model, welding drive motion coordinate system, and node three-dimensional coordinates); The intrinsic parameter matrix of the thermal imaging camera (obtained in advance through calibration, 3×3 matrix); It is the inverse of the intrinsic parameter matrix; These are the geometric correction parameters (the 3×3 distortion correction matrix mentioned above); The z-axis coordinate of the center of the molten pool (compared to the z-axis coordinate of the current weld node) Consistency, that is This ensures that the three-dimensional coordinates of the molten pool center and the coordinates of the welding path nodes are at the same height, closely matching the actual position of the weld.

[0120] The path image processing strategy also includes: performing gray value statistics on the segmented molten pool region based on the preset calibration relationship between gray values ​​and temperature of the thermal imaging image; selecting the average gray value in the molten pool region as the real-time molten pool temperature; and using the real-time molten pool temperature to compare with the preset optimal welding temperature range and assist in generating real-time correction instructions.

[0121] Specifically, the grayscale values ​​of the molten pool region are first statistically analyzed: the average grayscale value of all pixels within the segmented molten pool region is calculated. As the basis for temperature conversion, its calculation formula is: , in, This represents the average grayscale value of the molten pool region. This represents the total number of pixels within the molten pool area; The set of pixels in the molten pool region; This represents the grayscale value of the pixel within the melt pool area after preprocessing.

[0122] Secondly, a pre-defined calibration relationship was established, and the linear relationship between the grayscale value of the molten pool thermal imaging image and the actual temperature was calibrated experimentally beforehand. After calibration, a linear fitting equation was obtained: ,in To calibrate the temperature, The image grayscale value, This is the temperature grayscale coefficient. This represents the temperature offset under different welding materials and welding power. They need to be calibrated and stored separately, and then called up in real time according to the current working conditions during the welding process.

[0123] Finally, the real-time molten pool temperature is calculated. Based on a preset calibration relationship, and combined with the average grayscale value of the molten pool region... Calculate the real-time molten pool temperature The formula is: , in, Real-time molten pool temperature; The linear fitting coefficients in the preset calibration relationship are called in real time to adapt to the current working conditions. This represents the average grayscale value of the molten pool region. The calculation results are synchronously transmitted to the molten pool offset monitoring and correction system for comparison with the preset optimal welding temperature range.

[0124] Furthermore, the molten pool offset monitoring and correction strategy includes: during the welding process, based on the node numbers of the two paths where the current welding is located, retrieving the preset molten pool data of the corresponding node from the associated stored node welding execution data; the preset molten pool data includes the three-dimensional coordinates of the node corresponding to the current welding segment, the preset optimal welding temperature range, the coordinate deviation threshold, and the temperature deviation threshold.

[0125] Specifically, the data retrieval logic uses a node number association mechanism. If the current welding position is at the node number association mechanism, the data retrieval logic will retrieve the data from the node number associated with the node. The node and the first The nodes are connected by node number " "and" "From the associated database of the welding control system, the preset molten pool data corresponding to the welding segment can be retrieved with one click. The retrieval delay is ≤0.1s, ensuring that the monitoring and welding process are synchronized and avoiding feedback lag."

[0126] Wherein, the three-dimensional coordinates of the node corresponding to the current welding segment are i.e., the first... Node coordinates and the Node coordinates This is used to calculate the reference trajectory coordinates corresponding to the current welding position, and the coordinates are consistent with the unified coordinate system. Preset optimal welding temperature range , , or , The melting point of the welding material; Coordinate deviation threshold Used to determine the degree of deviation between the molten pool center and the welding reference trajectory, it is set to 1 / 10 of the preset node spacing, and the calculation formula is as follows: , The preset node spacing is 1.0-2.0mm, therefore It is compatible with the overall welding precision requirements; Temperature deviation threshold The deviation between the real-time molten pool temperature and the optimal temperature range is used to determine the degree of deviation, set to 1 / 10 of the optimal temperature range span, and the calculation formula is as follows: For example, the temperature deviation threshold for copper is .

[0127] The melt pool offset monitoring and correction strategy also includes: synchronously acquiring the melt pool center coordinates and real-time melt pool temperature output by the path image processing strategy, comparing them with the retrieved preset melt pool data, and classifying them into three levels of judgment according to the degree of deviation: normal state, slightly abnormal state, and severely abnormal state.

[0128] Specifically, coordinate offset deviation The calculation process is as follows: calculate the three-dimensional coordinates of the molten pool center. Coordinates of the reference trajectory of the current welding position The offset, the reference trajectory coordinates are obtained by linear interpolation of the first... , Calculate the node coordinates.

[0129] The formula for calculating the interpolation of the reference trajectory coordinates is: , in, , , For nodes and The coordinate difference; These are the interpolation coefficients. , For the current welding position and node The actual distance For nodes and The distance between them .

[0130] The formula for calculating coordinate offset deviation is: , in, ,none Directional offset, only calculated , Planar offset.

[0131] By calculating the real-time molten pool temperature Midpoint of the preset optimal welding temperature range The deviation is used to obtain the temperature deviation. Temperature deviation The formula is: , , In the formula: The midpoint temperature of the preset optimal welding temperature range; This represents the temperature deviation value, which is non-negative.

[0132] Coordinate offset deviation and temperature deviation Define three levels of judgment criteria. (1) Normal state: when and When the molten pool is deemed to be in normal condition and the welding process is normal, no correction command needs to be generated.

[0133] (2) Minor abnormal condition: when or If the error is detected, it is considered a minor anomaly, indicating a slight deviation or temperature fluctuation, and a minor anomaly correction command needs to be generated.

[0134] (3) Severe abnormal state: when or If the condition is deemed serious, indicating a significant deviation or temperature runaway, a serious anomaly correction command must be generated.

[0135] The melt pool offset monitoring and correction strategy also includes: generating real-time correction instructions for minor anomalies and severe anomalies based on the three-level judgment.

[0136] Specifically, the real-time correction instructions are consistent with the motion coordinate system of the welding drive and the node welding execution parameters.

[0137] The minor anomaly correction command corresponds to a minor anomaly state, and its correction amount does not exceed 10% of the preset parameters. It adjusts the welding robot arm trajectory or welding power in a targeted manner. The minor anomaly correction command includes the following commands.

[0138] (1) Corrected by slight coordinate offset ( The trajectory of the robotic arm is adjusted, and the correction formula is as follows: , , in, These are slight corrections made by the robotic arm in the x and y directions. This is the unit vector component in the offset direction, ensuring that the correction direction is opposite to the offset direction; For offsets exceeding a threshold, ensure that the corrected offset falls back within the threshold range. For example, if the instruction is to adjust the robotic arm trajectory... Direction correction , Direction correction The welding parameters remain unchanged.

[0139] (2) Corrected by slight temperature fluctuations ( The welding power is adjusted using the following formula: , in, This is a slightly modified welding power; The current welding power is 0.1; 0.1 is the maximum correction factor to ensure the correction does not exceed 10%. For example, the instruction would be to adjust the welding power to... The robotic arm's trajectory remains unchanged.

[0140] The severe anomaly correction command corresponds to a severe anomaly state. Its correction amount does not exceed 20% of the preset parameters. At the same time, it adjusts the robotic arm trajectory and welding parameters, and reduces the welding speed if necessary. The severe anomaly correction command includes the following commands.

[0141] (1) Correction by severe coordinate offset ( The formula for trajectory correction and speed fine-tuning is as follows: , , , in, For robotic arms , Significant correction amount for direction; 1.2 is the correction factor to ensure rapid correction of offset; The corrected welding speed will be reduced to 90% of the current speed to allow time for trajectory adjustment. For example, the instruction could be "Adjust robotic arm trajectory." Direction correction , Direction correction The welding speed was adjusted to .

[0142] (2) Correction through severe temperature fluctuations ( The welding power can be significantly adjusted using the following formula: , in, This represents the severely corrected welding power; 0.2 is the maximum correction factor, ensuring the correction does not exceed 20%. For example, the instruction would be: "Emergency adjustment of welding power to..." The temperature of the molten pool is closely monitored, and the trajectory of the robotic arm remains unchanged.

[0143] The molten pool offset monitoring and correction strategy also includes: after the welding drive executes the real-time correction command, the path image processing strategy acquires molten pool thermal imaging images at a preset acquisition frequency, extracts molten pool data, and monitors and compares it again until the molten pool state returns to normal before welding is performed.

[0144] In this embodiment, after receiving a real-time correction command, the welding robotic arm immediately performs trajectory or parameter adjustments with an adjustment delay of ≤0.05s. After the adjustment is completed, the path image processing strategy still operates at the preset acquisition frequency. Acquire thermal imaging images of the molten pool, repeat the aforementioned preprocessing, region segmentation, center localization, and temperature calculation steps, and quickly extract new molten pool data.

[0145] The newly extracted molten pool data is compared with the preset molten pool data again, and the aforementioned deviation calculation and three-level judgment steps are repeated to determine whether the molten pool state has returned to normal.

[0146] If the second determination is a normal state, the correction stops and the welding robot continues to advance the welding along the preset path; if it is still a slight or serious abnormality, the correction amount is recalculated and a real-time correction command is generated based on the new molten pool data. The process of correction, data acquisition and comparison is repeated until the molten pool state returns to normal, ensuring that the welding process is not interrupted and defects are corrected in a timely manner.

[0147] The molten pool offset monitoring and correction strategy also includes associating and storing each generated real-time correction command, molten pool data before and after correction, welding execution parameters and correction effect with the path node number of the current welding segment.

[0148] Specifically, the storage format adopts an associative format of "welding segment node number (k-k+1) - correction time - real-time correction command - molten pool data before correction - molten pool data after correction - current welding parameters - correction effect," ensuring that each correction record can be accurately bound to a specific welding position. The stored data includes molten pool center coordinates, real-time molten pool temperature, and corresponding deviation values ​​before / after correction; current welding parameters include welding speed, welding power, and welding torch angle; and the correction effect includes the molten pool state judgment result after correction and the change in offset / temperature deviation. The stored data is used for subsequent welding quality traceability, correction strategy optimization, and fault diagnosis. It also provides historical correction data for feedback data comparison and adjustment command generation strategies, assisting in the calculation of correction deviation errors.

[0149] In this embodiment, the molten pool monitoring and correction strategy scientifically sets the acquisition frequency to ensure full coverage of the molten pool status between nodes. Combined with preprocessing techniques such as median filtering and grayscale stretching, along with adaptive threshold segmentation and ellipse fitting, it accurately extracts the molten pool center coordinates and real-time temperature. Geometric correction ensures data consistency with actual working conditions, effectively avoiding interference from noise and distortion, and ensuring monitoring accuracy meets welding requirements. Based on coordinate and temperature deviation thresholds, a three-level state is defined, with differentiated correction amounts for minor and severe anomalies. The robotic arm trajectory, welding power, and speed are adjusted accordingly, and the correction formula closely matches the working conditions. This avoids both insufficient correction leading to defect accumulation and overcorrection causing new problems, ensuring the molten pool quickly returns to normal. After the correction command is executed, real-time re-acquisition and re-inspection are performed, iterating until the molten pool is normal, achieving a closed loop of monitoring, comparison, correction, and re-inspection. This timely correction of offsets and temperature fluctuations effectively reduces defects such as undercut, burn-through, and uneven formation, improving weld formation accuracy and consistency. The correction of relevant data and node numbers provides a historical basis for subsequent feedback adjustments and error correction, facilitates welding quality traceability and fault diagnosis, and can optimize correction parameters in the long term to further improve welding stability.

[0150] The method further includes: after each welding path is completed, acquiring the node thermoforming image of the path node, and obtaining node forming data and node temperature data through a node image processing strategy; through a feedback data comparison and adjustment instruction generation strategy, comparing the node forming data and node temperature data of each node with preset node data, and generating an adjustment instruction for the next node.

[0151] Furthermore, the node image processing strategy includes: preprocessing the acquired node thermoforming images by denoising, enhancing, and geometrically correcting them.

[0152] Specifically, the noise reduction process uses a Gaussian filtering algorithm, and the filtering formula is as follows: , , in, It is a two-dimensional Gaussian filter kernel function; The standard deviation of the filtered Gaussian standard; The radius of the filter kernel; The grayscale value of the image after noise reduction; This represents the grayscale value of the original node thermoforming image.

[0153] Image enhancement uses the same grayscale stretching algorithm as described above, which enhances the grayscale difference between the solder joint forming area and the background.

[0154] Geometric correction follows the formula described above, using the same geometric correction parameters as those used in path image processing. Ensure that the image coordinates match the actual solder joint coordinates.

[0155] The node image processing strategy also includes: using an adaptive threshold segmentation algorithm, combining the grayscale distribution characteristics of the node thermoforming image with the preset solder joint width and preset solder joint height parameters in the node setting strategy, to automatically determine the segmentation threshold and divide the preprocessed image into node solder joint forming area and non-forming area.

[0156] Specifically, calculate the mean gray level of the preprocessed image. and grayscale standard deviation The formula is the same as above.

[0157] Adaptive segmentation threshold calculation: Combining preset solder joint size parameters, a forming size correction coefficient is introduced. The threshold formula is: , in, This is the segmentation threshold; This is the threshold adjustment coefficient; This is the forming size correction factor. ,in, To preset the solder joint width, To preset the solder joint height, The preset weld width is 1.5-3.0mm, which is used to dynamically adjust the threshold according to the weld size to avoid missing or over-segmenting.

[0158] Segmentation rules are based on grayscale values. For the formation area of ​​node solder joints, For the non-shaped region, a binarized image of the shaped region is obtained. .

[0159] The node image processing strategy also includes: contour fitting and size measurement of the segmented node solder joint forming region, with the extracted node forming data corresponding one-to-one with the node solder joint forming standard parameters in the node setting strategy. The node forming data includes the actual width of the solder joint. Actual height Actual melting depth Each corresponds to a preset parameter. , , .

[0160] Specifically, the contour fitting uses the least squares method to fit the contour of the binary forming region. Taking linear fitting in the width direction as an example, the fitting formula is: , , , in, The slope of the fitted line; The intercept of the fitted line; This represents the total number of contour points in the width direction. These are the pixel coordinates of the contour points. The height and depth directions are aligned accordingly.

[0161] Size measurement converts pixel coordinates into actual physical coordinates, and then calculates the actual forming parameters.

[0162] Pixel coordinates to physical coordinates are: , in, These are the actual physical coordinates; The pixel coordinates of the outline points; The focal length of the camera; The pixel coordinates of the center point of the camera image; , This represents the z-axis coordinate of the current node.

[0163] The actual forming parameters are calculated using the following formula: , , , in, This refers to the actual width of the weld joint; The maximum and minimum values ​​of the x-axis of the forming area; This refers to the actual height of the solder joint; This represents the actual penetration depth of the weld joint. This is the melting depth correction factor.

[0164] The node image processing strategy also includes: based on the preset calibration relationship between gray values ​​and temperature of thermal imaging images, performing gray value statistics on the segmented node solder joint forming area; selecting the average gray value within the forming area as the real-time temperature data of the node, which reflects the residual heat state of the node solder joint after welding.

[0165] Specifically, the average grayscale value of the formed area The calculation formula is: , in, This represents the total number of pixels in the formed area. The set of pixels in the shaped region; This is the grayscale value after preprocessing.

[0166] Waste heat temperature correction factor The conversion formula is: , in, Provides real-time temperature data for the nodes; These are the linear fitting coefficients in the preset calibration relationship; This is the waste heat temperature correction factor, used to correct temperature deviations caused by waste heat decay.

[0167] The node image processing strategy also includes: associating the extracted node forming data and node temperature data with the unique number of the current node, which is used for feedback data comparison and adjustment instruction generation strategy. It compares with preset node data and then generates adjustment instructions for the next path node, realizing the interleaving and coordination with real-time correction instructions to ensure the continuity and stability of the welding process.

[0168] Specifically, each node has a unique identifier that corresponds to a set of finished data. , , and temperature data Once bound, the data is stored in the control system and associated with the node's preset parameters and real-time correction historical data, facilitating accurate retrieval and comparison later.

[0169] The unique node number is in the format of “segment number-node number”, for example, 1-5 represents the 5th node of the first weld segment.

[0170] Furthermore, the feedback data comparison and adjustment instruction generation strategy includes: after each path node is welded, retrieving the node forming data, node temperature data, and node unique number extracted and bound by the node image processing strategy; at the same time, retrieving the preset node data corresponding to the node from the node welding execution data, as well as the correction data related to the current welding segment stored by the molten pool offset monitoring and correction strategy, to achieve the correlation and matching of feedback data, preset data, and historical correction data.

[0171] Specifically, after the node welding is completed, the system automatically identifies the unique number of the current node and retrieves three types of data synchronously through the number association: (1) node feedback data of forming data and temperature data; (2) node preset data of preset forming standard parameters, preset optimal temperature range and deviation threshold; (3) real-time correction instructions of the current welding segment, molten pool data before and after correction, and historical correction data of correction effect. The retrieval delay is ≤0.1s to ensure rapid completion of comparative analysis.

[0172] Using the unique node number as the core matching keyword, we ensure that the three types of data accurately correspond to the same node and the same welding segment. Among them, the historical correction data is associated with the node number of the current welding segment to correct the error in deviation calculation and ensure that the comparison results are consistent with the actual welding conditions.

[0173] The feedback data comparison and adjustment instruction generation strategy also includes: comparing the retrieved node forming data with the preset weld point forming standard parameters of the corresponding node one by one, and calculating the forming deviation value of each parameter; comparing the node temperature data with the preset optimal welding temperature range of the corresponding node, and calculating the temperature deviation value; during the comparison process, simultaneously referring to the historical correction data of the molten pool offset monitoring and correction strategy, correcting the deviation calculation error, and ensuring the accuracy of the comparison results.

[0174] Specifically, the formula for calculating the forming deviation value is as follows: , , , in, For the deviation of the weld joint width, For height deviation, These represent the penetration depth deviation, and all are non-negative values. This refers to the actual formed data of the node; Preset standard parameters for solder joint formation ( , , , (Where the width of the node weld is given).

[0175] 2. The formula for calculating the temperature deviation is: , , in, The midpoint temperature of the preset optimal temperature range; This is the temperature deviation value; This is the real-time temperature data of the node. This deviation value reflects the difference between the residual heat state of the weld joint and the preset optimal state, and indirectly reflects the accuracy of temperature control during the welding process.

[0176] The feedback data comparison and adjustment instruction generation strategy also includes: based on the forming deviation value and temperature deviation value, combined with the forming deviation threshold and temperature deviation threshold in the node setting strategy, the welding quality of the current node is judged in three levels: no deviation state, slight deviation state, and large deviation state.

[0177] 3. Deviation Calculation Error Correction: Combining historical correction data stored with the molten pool offset monitoring and correction strategy, an error correction coefficient is introduced. The forming deviation and temperature deviation values ​​are corrected to ensure that the comparison results are closer to the actual welding conditions. The correction formula is as follows: , , , , in, This is the corrected forming deviation value; This is the corrected temperature deviation value; This is the error correction factor (range 0.02-0.05), whose value is calculated from historical correction data. ( This represents the number of historical corrections for the current welding segment. This is the preset deviation value before the i-th correction. (where is the actual deviation value after the i-th correction). The larger the deviation, the more corrections are needed. The closer the value is to 0.05, the more targeted the error correction will be.

[0178] Specifically, the welding quality grade is determined based on the corrected deviation value. (1) Excellent grade: When , , and When the current node is deemed high-quality, the welding formation and temperature control of the current node meet the optimal standards, and no parameter adjustment is required for the next node.

[0179] (2) Pass / Fail Level: When , , or If the current node is deemed to be qualified, the welding quality of the current node meets the standard, but there is a slight deviation. A small parameter adjustment (adjustment amount ≤ 5%) is required for the next node.

[0180] (3) Level to be adjusted: when , , or If the current node welding is significantly deviated, it is determined that the current node needs to be adjusted. The parameters of the next node need to be adjusted accordingly (adjustment amount 5%-10%) to avoid the accumulation of deviations.

[0181] in, This represents the width of the weld at the current node. Preset the weld height and penetration depth parameters; The temperature deviation threshold is used to determine the criteria that are compatible with the overall welding accuracy requirements and the ignition coil weld quality standards.

[0182] The feedback data comparison and adjustment instruction generation strategy also includes: based on the three-level judgment of the node, generating node adjustment instructions for the next path node with no adjustment, minor adjustment, and targeted adjustment. The node adjustment instructions are consistent with the welding execution parameters and the motion coordinate system of the welding drive in the node setting strategy.

[0183] Generate the next node adjustment instruction. (1) Forming parameter adjustment (corresponding to width, height, and penetration deviation): Adjust the welding power and welding speed of the next node to indirectly correct the weld point forming size. The adjustment formula is as follows: Width deviation adjusted to , ; Height deviation adjusted to , ; Melt depth deviation adjusted to , ; In the formula: For the next node ( The welding power is adjusted according to the corresponding forming parameters; The welding speed is adjusted according to the forming parameters corresponding to the next node; For the current node ( Welding power and welding speed; The forming adjustment coefficient is used to determine the shape. The larger the deviation, the closer the adjustment coefficient should be to 0.5, ensuring that the adjustment effect is significant and does not exceed the safe range of parameters.

[0184] (2) Temperature parameter adjustment corresponding to temperature deviation: Adjust the welding power of the next node to correct the molten pool temperature, thereby optimizing the residual heat state of the weld. The adjustment formula is: , in, The welding power adjusted for the next node temperature; This is the temperature adjustment factor; These are the corrected temperature deviation value and the temperature deviation threshold, respectively. If the temperature deviation is positive, the power is reduced; if the deviation is negative, the power is increased to ensure that the molten pool temperature at the next node is in the optimal range.

[0185] The feedback data comparison and adjustment instruction generation strategy also includes: synchronously sending the generated next node adjustment instruction to the welding driver to guide the welding execution of the next node; and storing the feedback data, comparison results, grading judgment results of the current node and the adjustment instruction of the next node in association with the current node number and the next node number to realize the interleaving of real-time correction instructions and adjustment instructions, and avoid welding defects caused by feedback lag.

[0186] Specifically, when both forming deviation and temperature deviation exist simultaneously, the optimal adjustment values ​​of welding power and welding speed in each adjustment direction are taken and integrated to generate the final adjustment instruction for the next node. The instruction format is "node number-adjustment type-adjusted welding power-adjusted welding speed-adjustment basis", ensuring that the instruction is clear and executable.

[0187] In this embodiment, the node image processing and feedback data comparison and adjustment strategy employs preprocessing techniques such as Gaussian filtering and grayscale stretching. Combined with adaptive threshold segmentation and contour fitting tailored to the weld point size, it accurately extracts the actual width, height, penetration depth, and real-time residual heat temperature of the weld point. Through pixel coordinate to physical coordinate conversion and error correction, it ensures that the data accurately matches the preset standard, avoiding interference from noise and segmentation deviations. Simultaneously, it introduces an error correction coefficient based on historical correction data to optimize the calculation of forming and temperature deviations. A three-level quality judgment standard is adopted to accurately distinguish between excellent, qualified, and adjustment-needed levels, which not only meets welding precision requirements but also effectively identifies minor deviations, avoiding misjudgments or omissions. Differentiated adjustment instructions are generated based on the quality level to accurately adjust the welding power and speed of the next node. The adjustment amount is set in a formulaic manner, balancing adjustment effect and parameter safety, promptly correcting deviations at the current node, preventing deviations from being passed on to subsequent nodes, and ensuring uniform weld quality throughout the entire weld segment.

[0188] The method also includes: ensuring the stability of the welding process throughout the entire welding path by interleaving and coordinating real-time correction instructions and adjustment instructions, thereby avoiding welding defects caused by feedback lag.

[0189] Specifically, during the welding process, real-time correction commands are continuously executed, simultaneously capturing thermal imaging images of the molten pool between every two path nodes. The coordinates, temperature, and offset of the molten pool center are calculated in real time. Once the molten pool offset exceeds the threshold or the temperature deviates from the optimal range, real-time correction operations such as robotic arm trajectory adjustment and welding power fine-tuning are immediately triggered. This eliminates the need to wait for the welding of a single node to be completed, enabling rapid response to immediate interference during the welding process and preventing immediate deviations from escalating into welding defects due to feedback lag. Adjustment commands are generated after each path node is completed. Based on the forming data and temperature data extracted from the node image processing, and combined with preset parameters, deviation comparison and three-level judgment are performed. Adjustment commands for the next node are generated in a targeted manner, optimizing the welding speed, power, coordinates, and other parameters of the next node in advance, compensating for any slight deviations that may exist at the current node, and avoiding subsequent welding defects caused by the accumulation of deviations. The core of this collaborative approach lies in data exchange and timing: the execution history of real-time correction commands is synchronously fed back to the feedback data comparison system to correct deviation calculation errors in the next node, making adjustment commands more targeted; the parameters for the next node set by the adjustment command serve as the baseline threshold for the real-time correction command, ensuring that real-time correction always revolves around the optimized parameters and preventing a disconnect between the two. Simultaneously, through the associated storage of node numbers, both real-time correction data and adjustment commands can be precisely bound to their corresponding nodes, with command sending delays controlled within 0.1 seconds. This ensures that when the next node welding begins, the adjustment command is already in effect, and the real-time correction command follows synchronously, forming a continuous process of node welding, real-time correction, node feedback, generation of adjustment commands, execution by the next node, and real-time correction.

[0190] In this embodiment, this interleaved cooperation mode not only solves the shortcomings of real-time correction commands in correcting immediate deviations but also makes up for the defects of adjustment commands lagging behind the welding process and being unable to cope with immediate interference. It effectively ensures the stability of welding parameters, the accuracy of welding trajectory, and the uniformity of weld formation in all welding paths, and completely avoids welding defects such as weld offset, incomplete penetration, undercut, and overheating burn-through caused by feedback lag. It further adapts to the stringent requirements of automotive electronic ignition coils for welding precision and stability, ensuring that each weld segment meets the preset design standards and improving the overall welding qualification rate.

[0191] Example 2 like Figure 2-6 As shown, the welding equipment is applicable to the welding path prioritization method for automotive electronic ignition coils described in Embodiment 1 above. The welding equipment includes a base 1, which supports all components of the equipment. Specifically, the base 1 is integrally formed from high-strength cast iron and undergoes aging treatment to eliminate internal stress, ensuring sufficient rigidity and stability to effectively support the weight of all components and prevent welding deviations caused by equipment vibration during the welding process. The upper surface of the base 1 is precision milled to provide a reference for the accurate installation of each component. Adjustable anchor bolts are also provided at the bottom of the base for easy equipment leveling and on-site installation adaptation. The welding equipment also includes: at least one ignition coil carrying mechanism 2, which is fixedly mounted on the base 1; each of the ignition coil carrying mechanisms 2 is used to carry the ignition coil integration plate 3; the ignition coil integration plate 3 has multiple ignition coils to be welded evenly arranged on it. Specifically, the installation position of the ignition coil carrying mechanism 2 can be finely adjusted according to the size of the ignition coil integration plate 3 and the welding requirements to ensure precise correspondence with the working range of the robotic arm welding mechanism 4; each of the ignition coil carrying mechanisms 2 is specifically used to carry the ignition coil integration plate 3, adapting to batch welding scenarios and improving welding efficiency; the ignition coil integration plate 3 is a circular structure, integrally injection molded from high-temperature resistant engineering plastic, and its surface has multiple ignition coil installation positions to be welded evenly arranged in a ring. Each position is provided with a positioning groove that matches the shell of the ignition coil, which can realize the synchronous positioning and carrying of multiple ignition coils, making it convenient for the robotic arm welding mechanism 4 to complete the welding of all ignition coils in sequence, greatly improving the batch production cycle.

[0192] Furthermore, the ignition coil carrying mechanism 2 includes a rotary motor 201, which is mounted on the base 1. Specifically, the rotary motor 201 is fixedly mounted on the base 1 via a motor mounting bracket. A servo rotary motor is selected, which features adjustable speed and high positioning accuracy. It can precisely adjust the rotation speed according to the welding rhythm and has a rapid response. It can cooperate with the working progress of the robotic arm welding mechanism 4 to achieve precise rotation of the ignition coil integration disk 3, ensuring that the weld seam area of ​​each ignition coil to be welded can be accurately moved to the welding operation position.

[0193] The rotating disk 202 is fixedly mounted on the rotating end of the rotating motor 201. Specifically, the rotating disk 202 is fixedly mounted on the rotating end of the rotating motor 201 by a key connection and is coaxially arranged with the rotating motor 201. The rotating disk 202 is made of aluminum alloy, which is lightweight and has sufficient rigidity, effectively reducing rotational inertia and ensuring the stability of the rotation process. The upper surface of the rotating disk 202 is provided with an annular positioning groove for preliminary positioning of the ignition coil integrated disk 3, preventing radial displacement of the integrated disk during rotation and ensuring the accuracy of the welding position.

[0194] At least one magnetic positioning block 203 is evenly installed on the rotating disk 202. The magnetic positioning block 203 is used to hold the ignition coil integrated disk 3 on the rotating disk 202. Specifically, the magnetic positioning blocks 203 are evenly installed in a ring shape inside the annular positioning groove of the rotating disk 202. The magnetic positioning blocks 203 are made of high-strength permanent magnet material, and the surface is treated with anti-rust. They have the characteristics of strong adsorption force, rapid positioning, and not easy to demagnetize. They are used to firmly adsorb and fix the ignition coil integrated disk 3 on the rotating disk 202. With the help of the annular positioning groove, dual positioning is achieved. This can prevent the ignition coil integrated disk 3 from shifting due to vibration and rotation during the welding process, and can also quickly complete the clamping and disassembly of the integrated disk, improving the loading and unloading efficiency. The number of magnetic positioning blocks 203 is matched with the number of installation positions of the ignition coil integrated disk 3 to ensure that the integrated disk is subjected to uniform force and the positioning is more accurate.

[0195] The welding equipment also includes a robotic arm welding mechanism 4, which is mounted on the base 1. Specifically, the robotic arm welding mechanism 4 is fixedly mounted on one side of the upper surface of the base 1 via a mounting bracket. The installation position corresponds to the position of the ignition coil carrying mechanism 2, ensuring that the working range of the robotic arm can completely cover all welding positions on the ignition coil integration plate 3. The robotic arm welding mechanism 4 uses a multi-degree-of-freedom industrial robotic arm, which has the characteristics of flexible movement and high positioning accuracy. It can accurately execute welding actions according to the welding path generated by the welding path optimization method mentioned above and the adjustment instructions, flexibly adapting to the welding needs of welds of different shapes and positions, and can realize continuous welding operations, greatly improving welding efficiency and welding accuracy.

[0196] The robotic arm welding mechanism 4 is equipped with a thermal imager 5 and a 3D laser scanner 6. The 3D laser scanner 6 emits a laser beam to perform a full-area scan of the surface of the ignition coil to be welded and the weld seam area, capturing the 3D spatial information of the area to be welded, converting the scan data into a digital point cloud, and generating a digital 3D model of the ignition coil. The thermal imager 5 is equipped with dual-lens cameras, which simultaneously acquire the forming image and temperature region and data of the weld point or weld seam of the ignition coil after welding, and generate a fused image.

[0197] The two components work together to provide precise data support for the contour recognition, molten pool monitoring, and node image processing steps in the welding path optimization method described above, ensuring full controllability of the welding process. The 3D laser scanner 6 is a high-precision industrial-grade laser scanner, which emits a high-frequency laser beam to perform a full-area, dead-angle-free scan of the surface of the ignition coil to be welded and the area to be welded. After the laser beam is reflected by the surface of the ignition coil, the receiving module of the scanner captures the reflected signal, thereby accurately capturing the 3D spatial information of the area to be welded. The scanner has a built-in data processing module that can quickly convert the captured 3D spatial information into digital point cloud data. After removing redundant point cloud data, a complete digital 3D model of the ignition coil is generated through a point cloud fitting algorithm and transmitted synchronously to the welding control system. This provides precise model data support for the smoothing and denoising, region segmentation, and weld recognition steps of the digital 3D model in the contour recognition strategy described above, ensuring the accurate generation of the welding path reference trajectory.

[0198] The thermal imager 5 is equipped with dual lenses: a visible light lens and an infrared thermal imaging lens. It can simultaneously acquire two types of image data to meet different welding monitoring needs. The visible light lens acquires the forming image of the weld point or weld seam after welding, clearly capturing the weld point's outline and forming state, providing clear image data for steps such as weld point forming size measurement and outline fitting in the node image processing strategy. The infrared thermal imaging lens acquires the temperature area distribution and real-time temperature data of the weld point or weld seam, accurately capturing the spatial distribution characteristics and temperature change trends of the weld point temperature, generating a temperature thermogram. The thermal imager 5 has a built-in image fusion module that can fuse the acquired forming image with the temperature thermogram, generating a fused image with superimposed forming and temperature information. This image is synchronously transmitted to the welding control system, providing accurate temperature and forming data support for the aforementioned molten pool offset monitoring and correction, feedback data comparison and adjustment, etc. This allows staff to intuitively view the weld point forming quality and temperature state, while also providing a reliable basis for generating adjustment commands, ensuring that welding quality meets standards.

[0199] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A method for prioritizing soldering paths for automotive electronic ignition coils, characterized in that, The method includes: According to the preset scanning time and preset scanning trajectory, the surface of the ignition coil and the area to be welded are scanned in the whole domain to obtain the three-dimensional coordinates of each scanning point and form the original point cloud data; the original point cloud data is processed by the stitching fitting strategy to obtain the digital three-dimensional model of the ignition coil. The contour recognition strategy is used to identify the weld area in the digital 3D model, extract the center contour line of each weld segment, and use the contour line as the reference trajectory of the welding path; according to the shape of the weld, the movement direction, start point and end point of each welding path segment are defined; according to the node setting strategy, the welding execution data of each path node is set on each welding path segment. Welding is performed sequentially along the path nodes according to the welding path and welding execution parameters; Thermal imaging images of the molten pool are acquired between every two path nodes according to a preset acquisition frequency. The coordinates and temperature of the molten pool center are obtained through path image processing strategies. The coordinates and temperature of the molten pool center are compared with the preset molten pool data through a molten pool offset monitoring and correction strategy, and real-time correction instructions are generated. After each welding path is completed, the node thermoforming image of that path node is acquired, and the node forming data and node temperature data are obtained through the node image processing strategy; through the feedback data comparison and adjustment instruction generation strategy, the node forming data and node temperature data of each node are compared with the preset node data, and the adjustment instruction for the next node is generated. By interleaving and coordinating real-time correction and adjustment commands, the stability of the welding process throughout the entire welding path is ensured, welding defects caused by feedback lag are avoided, and the welding of the area to be welded is completed.

2. The welding path prioritization method for automotive electronic ignition coils according to claim 1, characterized in that, The splicing and fitting strategy includes: A Gaussian filtering algorithm is used to smooth the original point cloud data, a Gaussian distribution function is constructed, and a weighted average calculation is performed on the three-dimensional coordinates of each scanning point and a preset number of neighboring points to obtain the coordinate deviation value of each scanning point; sampling points whose coordinate deviation value exceeds the noise point judgment threshold are judged as noise points and removed. A segment of the point cloud covering the reference features of the ignition coil is selected as the target point cloud, and the point clouds obtained from the scanning of the remaining segments are used as source point clouds. The rotation and translation matrices between the source and target point clouds are calculated iteratively. In each iteration, the sampling point in the source point cloud is matched with the nearest point in the target point cloud, and the distance error between the two points is calculated. The iteration stops and the optimal coordinate transformation relationship is determined when the distance error is less than the preset registration accuracy threshold. Based on the optimal coordinate transformation relationship, all source point clouds are uniformly transformed into the coordinate system of the target point cloud so that the coordinate system is consistent with the motion coordinate system of the welding drive. Based on the structural characteristics of the ignition coil, the global point cloud is divided into multiple fitting regions, and surface fitting is performed on each region. Then, through a surface stitching algorithm, the fitted surfaces of each region are integrated into a complete three-dimensional surface model. When the fitting error between the fitted surface and the original point cloud does not exceed the preset fitting error threshold, a digital three-dimensional model is generated. The coordinate system of this digital three-dimensional model is the same as the motion coordinate system of the welding drive.

3. The welding path prioritization method for automotive electronic ignition coils according to claim 2, characterized in that, The contour recognition strategy includes: The digital 3D model is smoothed and denoised. Based on the preset structural features of the automotive electronic ignition coil, a region growing algorithm is used to segment the digital 3D model into regions, dividing the model into the shell region, winding region, terminal region, and suspected weld seam region. Among them, the suspected weld seam region is the connection and transition region between different structural regions. Three-dimensional features are extracted from suspected weld areas, including the region's normal vector direction, surface curvature, contour edge gradient, and thickness parameter. The normal vector direction distinguishes the orientation difference between the weld area and adjacent structural areas, with a significant angle between the weld area's normal vector direction and that of adjacent structural areas. Surface curvature identifies concave or convex features in the weld area. Contour edge gradient captures the boundary contour of the weld area, with the gradient difference between the weld boundary and the edge gradient of adjacent structures exceeding a preset gradient threshold. The thickness parameter filters out invalid suspected areas, retaining areas with a thickness equal to the preset weld width as target areas for identification. The extracted features of suspected weld areas are compared with the preset weld feature template. When the similarity between the features of the suspected weld area and the preset weld feature template is greater than or equal to the preset comparison threshold, the area is determined to be a valid weld area to be welded. The boundary contour lines of the effective weld seam to be welded area are extracted using a contour extraction algorithm; the center lines of the two boundary contour lines are obtained by coordinate fitting calculation, and these center lines are used as the center contour lines of each weld seam segment.

4. The welding path prioritization method for automotive electronic ignition coils according to claim 3, characterized in that, Node configuration strategies include: Path nodes are set at preset node spacing for weld inflection points, weld width change points, the start and end points of irregular structures, and straight and uniform sections of welds; the welding execution parameters are three-dimensional coordinates, preset welding parameters, standard parameters for welding feedback, and feedback deviation thresholds. Each path node extracts its three-dimensional coordinates from the welding execution parameters on the weld center contour line. These three-dimensional coordinates are consistent with the digital three-dimensional model and the motion coordinate system of the welding drive, serving as the target positioning coordinates of the welding head during node welding. This ensures that the welding head can move accurately to the preset position of each node, providing a positioning reference for the accurate execution of the welding path. The three-dimensional coordinates of each node are unique and continuous with the coordinates of adjacent nodes, avoiding welding path deviation caused by node coordinate discontinuity. For each path node, define the preset welding parameters, welding feedback standard parameters, and feedback deviation threshold in the welding execution parameters; All path nodes on each welding path are numbered sequentially according to the direction of welding movement, with each node corresponding to a unique number, and the welding execution data of each node is associated and stored with it. The preset welding parameters include welding speed, welding power, and welding torch angle; for adjacent nodes on the same weld, the preset welding parameters are adjusted in a gradient according to the differences in weld structure to ensure that the welding parameters are the same as the weld characteristics. The feedback standard parameters are consistent with the feedback standard parameters in the welding execution parameters, including the node weld formation standard parameters and the node temperature standard parameters. The weld formation standard parameters include the preset weld width, preset weld height, and preset penetration depth, which are the same as the preset weld size. The node temperature standard parameters are preset optimal welding temperature ranges, which are used to provide a clear benchmark for the comparison of feedback data after node welding is completed. The feedback deviation threshold is consistent with the feedback deviation threshold in the welding execution parameters, corresponding to the node feedback standard parameter setting; the feedback deviation threshold includes coordinate deviation threshold, forming deviation threshold and temperature deviation threshold; the coordinate deviation threshold is used to determine the degree of offset between the center of the molten pool and the three-dimensional coordinates of the node, the forming deviation threshold is used to determine the degree of deviation between the forming size of the node weld and the feedback standard parameter, and the temperature deviation threshold is used to determine the degree of deviation between the temperature of the node weld and the optimal temperature range.

5. The welding path prioritization method for automotive electronic ignition coils according to claim 4, characterized in that, Path image processing strategies include: The acquired molten pool thermal imaging images are preprocessed with noise reduction, enhancement and geometric correction to obtain the preprocessed molten pool thermal imaging images and geometric correction parameters; An adaptive threshold segmentation algorithm, combined with the grayscale distribution characteristics of the molten pool thermal imaging image, is used to divide the preprocessed molten pool thermal imaging image into molten pool region and non-molten pool region. A fitting algorithm is used to fit the contour of the segmented molten pool region to obtain the contour boundary coordinates of the molten pool region, and the geometric center of the contour boundary is calculated. This geometric center is defined as the image pixel coordinates of the molten pool center. Based on the geometric correction parameters, the pixel coordinates of the molten pool center are converted into the actual coordinates of the molten pool center. The coordinates of the molten pool center are consistent with the 3D coordinates of the nodes in the digital 3D model, the motion coordinate system of the welding drive, and the node setting strategy. The coordinates of the molten pool center are used to compare with the 3D coordinates of the nodes to determine the degree of molten pool offset. Based on the preset calibration relationship between grayscale values ​​and temperature of thermal imaging images, grayscale values ​​are statistically analyzed in the segmented molten pool region; the average grayscale value within the molten pool region is selected as the real-time molten pool temperature; the real-time molten pool temperature is used to compare with the preset optimal welding temperature range and to assist in generating real-time correction commands.

6. The welding path prioritization method for automotive electronic ignition coils according to claim 5, characterized in that, Molten pool offset monitoring and correction strategies include: During the welding process, based on the node numbers of the two paths where the current welding is located, the preset molten pool data of the corresponding node is retrieved from the associated stored node welding execution data; the preset molten pool data includes the three-dimensional coordinates of the node corresponding to the current welding segment, the preset optimal welding temperature range, the coordinate deviation threshold, and the temperature deviation threshold. The center coordinates of the molten pool and the real-time temperature of the molten pool output by the synchronous acquisition path image processing strategy are compared with the preset molten pool data and judged into three levels according to the degree of deviation: normal state, slightly abnormal state and severely abnormal state. Based on the three-level judgment, real-time correction instructions are generated for minor anomaly correction instructions and severe anomaly correction instructions; After the welding driver executes the real-time correction command, the path image processing strategy acquires thermal imaging images of the molten pool at a preset acquisition frequency, extracts the molten pool data, and monitors and compares it again until the molten pool state returns to normal before welding is performed. Each generated real-time correction command, the molten pool data before and after correction, the welding execution parameters, and the correction effect are associated with and stored with the path node number of the current welding segment.

7. The welding path prioritization method for automotive electronic ignition coils according to claim 6, characterized in that, Node image processing strategies include: The acquired nodal thermoforming images are preprocessed with noise reduction, enhancement, and geometric correction. An adaptive threshold segmentation algorithm is adopted, which combines the gray-scale distribution characteristics of the node thermoforming image with the preset solder joint width and preset solder joint height parameters in the node setting strategy to automatically determine the segmentation threshold and divide the preprocessed image into node solder joint forming area and non-forming area. Contour fitting and size measurement are performed on the segmented node solder joint forming area, and the extracted node forming data corresponds one-to-one with the node solder joint forming standard parameters in the node setting strategy. Based on the preset calibration relationship between grayscale values ​​and temperature of thermal imaging images, grayscale values ​​are statistically analyzed for the segmented node solder joint forming area; the average grayscale value within the forming area is selected as the real-time temperature data of the node, which reflects the residual heat state of the node solder joint after welding is completed. The extracted node forming data and node temperature data are associated and bound to the unique number of the current node. This is used to provide feedback data comparison and adjustment instruction generation strategy. The data is compared with preset node data to generate adjustment instructions for the next path node. This allows for interleaving with real-time correction instructions to ensure the continuity and stability of the welding process.

8. The welding path prioritization method for automotive electronic ignition coils according to claim 7, characterized in that, The feedback data comparison and adjustment instruction generation strategy includes: Upon completion of welding at each path node, the node forming data, node temperature data, and unique node number extracted and bound using the node image processing strategy are retrieved. From the node welding execution data, the corresponding preset node data, as well as the correction data related to the current welding segment stored in the molten pool offset monitoring and correction strategy, are retrieved to achieve correlation and matching between feedback data, preset data, and historical correction data. The retrieved node forming data is compared one by one with the preset weld point forming standard parameters of the corresponding node, and the forming deviation value of each parameter is calculated; the node temperature data is compared with the preset optimal welding temperature range of the corresponding node, and the temperature deviation value is calculated; during the comparison process, the historical correction data of the molten pool offset monitoring and correction strategy is referenced simultaneously to correct the deviation calculation error and ensure the accuracy of the comparison results. Based on the forming deviation value and temperature deviation value, combined with the forming deviation threshold and temperature deviation threshold in the node setting strategy, the welding quality of the current node is judged in three levels: no deviation state, slight deviation state, and large deviation state. Based on the three-level node determination, node adjustment instructions for the next path node are generated, including no adjustment, minor adjustment, and targeted adjustment. The node adjustment instructions are consistent with the welding execution parameters and the motion coordinate system of the welding drive in the node setting strategy. The generated adjustment instructions for the next node are synchronously sent to the welding driver to guide the welding execution of the next node. The feedback data, comparison results, grading judgment results of the current node, and the adjustment instructions for the next node are associated and stored with the current node number and the next node number to realize the interleaving of real-time correction instructions and adjustment instructions, and avoid welding defects caused by feedback lag.

9. Welding equipment, characterized in that, The welding equipment is applicable to the welding path prioritization method for automotive electronic ignition coils as described in any one of claims 1-8, and the welding equipment comprises: Base (1), which is used to support all the components of the device; At least one ignition coil support mechanism (2) is fixedly mounted on the base (1); each of the ignition coil support mechanisms (2) is used to support the ignition coil integration disk (3); the ignition coil integration disk (3) is uniformly arranged with a plurality of ignition coils to be welded; The robotic arm welding mechanism (4) is mounted on the base (1); The robotic arm welding mechanism (4) is equipped with a thermal imager (5) and a three-dimensional laser scanner (6). The three-dimensional laser scanner (6) is used to emit a laser beam to perform a full-area scan of the surface of the ignition coil to be welded and the area to be welded, capture the three-dimensional spatial information of the area to be welded, convert the scan data into a digital point cloud, and generate a digital three-dimensional model of the ignition coil. The thermal imager (5) is equipped with dual-lens cameras, which simultaneously acquire the forming image and temperature area and data of the weld point or weld seam of the ignition coil after welding, and generate a fused image.

10. The welding equipment according to claim 9, characterized in that, The ignition coil carrying mechanism (2) includes: A rotary motor (201) is mounted on a base (1); A rotating disk (202) is fixedly mounted on the rotating end of a rotary motor (201); At least one magnetic positioning block (203) is uniformly mounted on the rotating disk (202); the magnetic positioning block (203) is used to place the ignition coil integrated disk (3) on the rotating disk (202).

Images