An adaptive generation method and system for multi-layer, multi-pass welding paths in robots

Abstract

The application discloses a kind of robot multilayer multi-pass welding path adaptive generation method and system, method includes using time flight sensor to the three-dimensional point cloud of weld is collected, and the weld point cloud collected is preprocessed;The preprocessed weld point cloud is segmented to extract the feature of weld by random sample consistent weld segmentation algorithm based on density clustering, and using cubic spline interpolation method is interpolated fitting to weld feature point;According to the geometric information of welding workpiece and the weld feature point after interpolation fitting, multilayer multi-pass welding path is generated;Based on multilayer multi-pass welding path, the normal estimation method of weighted principal component analysis is used to calculate the posture of welding torch end.The method of the application can segment the three-dimensional point cloud of welding workpiece containing complex surface to extract weld feature point, and then generate multilayer multi-pass welding path through weld feature point and the geometric information of welding workpiece to solve the problem of low welding precision, slow speed and narrow application range of medium plate.

Description

Technical Field

[0001] This invention belongs to the field of welding technology, specifically relating to an adaptive generation method and system for multi-layer, multi-pass welding paths in robots. Background Technology

[0002] With the rapid development of industrial automation technology, the size of workpieces welded by robots is gradually transitioning from miniaturization to large-scale. Currently, medium-thick plates with a thickness exceeding 4 mm have become common materials in many fields such as automobile manufacturing, marine vessels, and bridge construction. When welding workpieces composed of medium-thick plates, due to the deep weld seam and large welding area, multi-layer multi-pass (MLMP) welding methods are usually used.

[0003] However, existing multi-layer, multi-pass welding technologies, besides manual welding, primarily employ a "teach-and-playback" mode using welding robots. In this mode, the operator must first teach each workpiece, manually setting the robot's welding path, and then the robot welds layer by layer according to the taught path. While this method can adapt to workpieces of different shapes and sizes, the teaching process relies on manual intervention, which is time-consuming and prone to errors, especially when dealing with large and complex-shaped workpieces, where the teaching workload is even more substantial. Furthermore, the playback welding mode after teaching lacks adaptability and is difficult to adjust to minor changes in workpiece shape or uncertainties in the welding environment. Therefore, the applicability and effectiveness of this method are limited when the workpiece shape is complex or the welding environment changes.

[0004] With the increasing demands for production efficiency and welding quality in the manufacturing industry, the traditional "teach-and-playback" model can no longer meet the needs of highly efficient and automated welding. In dealing with complex weld shapes and dynamic welding environments, welding systems urgently require more adaptive welding path generation technology. To this end, a welding path generation method based on vision sensors has been proposed. However, most current mainstream weld feature extraction technologies are limited to two-dimensional planar information processing, neglecting the three-dimensional spatial geometric characteristics of the weld. Patents CN118809035A and CN117260086B plan multi-layer, multi-pass welding paths using two-dimensional planar information of the weld; however, this results in insufficient accuracy in describing the actual shape of the weld during the welding path generation process, leading to welding results deviating from the ideal state. More importantly, existing multi-layer, multi-pass welding path generation methods mostly focus on V-groove welds (such as patent CN113352317A), exhibiting significant adaptability when facing more complex weld shapes or diverse welded workpieces, failing to meet the diverse needs of industrial production. These problems severely restrict the promotion and application of welding technology in complex manufacturing scenarios. Summary of the Invention

[0005] To address the aforementioned problems, this invention provides an adaptive generation method and system for multi-layer, multi-pass welding paths in robots, aiming to offer a method for generating multi-layer, multi-pass welding paths with high accuracy, high speed, and wide applicability.

[0006] An adaptive generation method for multi-layer, multi-pass welding paths in robots, the method comprising the following steps:

[0007] A time-of-flight sensor was used to acquire three-dimensional point clouds of the weld seam, and the acquired weld seam point clouds were preprocessed.

[0008] The preprocessed weld point cloud is segmented using a density clustering-based random sample consistent weld segmentation algorithm to extract weld features, and the weld feature points are interpolated and fitted using a cubic spline interpolation method.

[0009] Multi-layer, multi-pass welding paths are generated based on the geometric information of the workpiece and the interpolated weld feature points.

[0010] Based on a multi-layer, multi-pass welding path, the welding torch tip orientation is calculated using a weighted principal component analysis method for normal estimation.

[0011] A further technical solution of the present invention is: preprocessing the collected weld point cloud includes using a radius filtering method to remove outliers and using a voxel-based point cloud downsampling method to reduce the amount of point cloud data.

[0012] A further technical solution of the present invention is: a random sample consistent weld seam segmentation algorithm based on density clustering, specifically including the following steps:

[0013] The iterations are performed using the following steps until the maximum number of iterations is reached:

[0014] Initialize three non-collinear points randomly from the point cloud set.

[0015] Calculate the plane equation based on the initialized points;

[0016] Calculate the distance from all points to the plane based on the plane equation, and obtain the total number of all interior points whose distance from a point to the plane is less than a threshold.

[0017] If the number of interior points is greater than the threshold, calculate the plane equation represented by each interior point as the input data for extracting weld feature points, and remove each interior point from the point cloud set.

[0018] After the iteration is complete, if there are any remaining point clouds in the point cloud set, perform the following steps on each of the remaining point clouds:

[0019] Using the point cloud as the center and the radius distance threshold as the radius, calculate the number of points inside the circle;

[0020] If the number of points inside the circle is greater than the threshold for the number of adjacent points, then mark the point cloud with the circle as the center, and randomly select another point cloud inside the circle until all the remaining point clouds have been marked.

[0021] The point clouds are classified according to the labels, and the final segmented point cloud results are obtained.

[0022] A further technical solution of the present invention is as follows: extracting weld features includes identifying and extracting the connecting lines between different surfaces in the weld area from the segmented point cloud, and further analyzing the geometric features. The specific steps include: taking the point cloud closest to plane P1 on plane P2 as feature points, and these feature points constitute the weld path of the corner joint.

[0023] A further technical solution of the present invention is as follows: before extracting weld features, the region of interest is extracted from the point cloud, and a distance threshold from the point to the plane is set when extracting feature points.

[0024] A further technical solution of the present invention is as follows: For fillet welds, a multi-layer, multi-pass welding path is generated, specifically including the following steps:

[0025] Based on the extracted weld features A, B, and C, the expression for the number of weld layers L' is:

[0026]

[0027] Where h is the predetermined weld bead thickness, and th is the set error threshold; to ensure that the weld bead can precisely cover the entire weld seam, the predetermined weld bead thickness h is adaptively adjusted to h′. When the remaining depth is within the allowable error range, the number of weld bead layers is rounded down; otherwise, the number of weld bead layers L′=[L]+1, and the predetermined weld bead thickness is adjusted as follows:

[0028]

[0029] After initial welding to obtain new weld features D and E, the first weld pass is welded starting from the right side of the weld in each layer; the number of weld passes is determined by the bevel width of each layer and the width of the weld pass, specifically the number of weld passes P′ in each layer. m As shown below:

[0030]

[0031] In the formula, w is the predetermined weld width, and D m The width of the m-th layer cross section;

[0032] To ensure that the weld bead is precisely covered by the weld seam, the predetermined weld bead width w is adaptively adjusted so that when P m -[P mWhen ]≤th, th is the set error threshold, representing the remaining width th·w after rounding to the nearest whole number w; if the remaining width is within the allowable error range, then for P m Round down to the nearest integer to ensure the number of weld passes is an integer; if P m -[P m ]>th indicates that the remaining width is too large, requiring an additional weld pass, P′ m =[P m +1, and adjust the predetermined weld width as follows:

[0033]

[0034] Based on the number of welding layers and weld passes in each layer, the coordinate positions of each pass are further determined, and the weld pass positions on each cross section are planned. The height between two adjacent layers is the corrected weld pass thickness h′, and the width between two adjacent passes in the same layer is the corrected weld pass width w′. Therefore, the coordinates W of the j-th pass in the i-th layer are... ij for:

[0035]

[0036] A further technical solution of the present invention is as follows: For the lap weld formed by connecting two workpieces together, a multi-layer, multi-pass welding path is generated, specifically including the following steps:

[0037] Weld features B and C are obtained based on the extracted weld feature point A and the preset welding height H. Then, the number of welding layers is calculated using the preset weld thickness h and error range th. The number of layers L′ is:

[0038]

[0039] The weld thickness h is adjusted to obtain h′. The number of weld passes per layer needs to be determined by the cross-sectional width of that layer and the width of the weld pass. The cross-sectional width can be obtained based on the similarity ratio, and then the number of weld passes can be calculated based on the error range th.

[0040]

[0041] The weld width w is adaptively adjusted according to the error range, and the adjusted w′ is:

[0042] Based on the number of welding layers L′ and the number of passes P′ in each layer m Further determine the starting coordinate position W of each weld bead ij :

[0043]

[0044] For butt welds, a multi-layer, multi-pass welding path is generated, specifically including the following steps:

[0045] The number of layers required for welding is calculated using the extracted weld feature connection lines A, B, C, and D, as well as the predetermined weld thickness h and allowable error range th. The number of layers L′ is:

[0046]

[0047] Adaptive adjustment of weld bead thickness:

[0048]

[0049] After calculating the number of welding layers, it is necessary to calculate the coordinates of the start and end points of the welding path, and then perform tack welding from the start to the end point; the coordinates of the start and end points... and for:

[0050]

[0051] A further technical solution of the present invention is as follows: For vertical plate welding, a multi-layer, multi-pass welding path is generated, specifically including the following steps:

[0052] Based on the extracted weld feature points A, B, and C, let the feature points be located in the XOY plane, with the CB direction being the positive X-axis direction;

[0053] The included angle θ of the V-groove weld is:

[0054]

[0055] Based on the known weld thickness h and weld angle θ, the number of weld passes P can be calculated as follows:

[0056]

[0057] The adaptive adjustment of weld bead thickness h′ is as follows:

[0058]

[0059] The coordinates of the start, intermediate, and end points of the welding path are calculated based on the number of welding passes. Then, the welding is performed from the start point to the intermediate point and then to the end point. The coordinates of the start point, intermediate point, and end point are calculated. and for:

[0060]

[0061] A further technical solution of the present invention is: calculating the welding torch tip attitude using a weighted principal component analysis normal estimation method, specifically including the following steps:

[0062] For point pi Select its surrounding neighborhood point set S i ;

[0063] Calculate the neighborhood point set S i point p in i Euclidean distance d to neighborhood center point c i d i For: d i =||p i -c||;

[0064] The weights of different points are defined based on their Euclidean distances from the center within the neighborhood. The formula for the weight w is:

[0065] in It is the average distance between neighboring points and the center point;

[0066] The neighborhood point set S i Subtract the coordinates of the center point c from the coordinates of the points in the matrix to achieve mean centering of the coordinates, and then minimize its weighted covariance matrix:

[0067] By using singular value decomposition of the weighted covariance matrix, the eigenvector corresponding to the smallest eigenvalue is obtained, which is then used as the eigenvector of point p. i The normal vector.

[0068] On the other hand, a robotic multi-layer, multi-pass welding path adaptive generation system is provided, the system comprising:

[0069] The weld seam point cloud acquisition module is used to acquire the three-dimensional point cloud of the weld seam using a time-of-flight sensor and to preprocess the acquired weld seam point cloud.

[0070] The weld feature point acquisition module is used to segment the preprocessed weld point cloud to extract weld features by using a density clustering-based random sample consistent weld segmentation algorithm, and to interpolate and fit the weld feature points using a cubic spline interpolation method.

[0071] The multi-layer, multi-pass welding path generation module is used to generate multi-layer, multi-pass welding paths based on the geometric information of the workpiece and the weld feature points after interpolation fitting.

[0072] The welding torch end attitude acquisition module is used to calculate the welding torch end attitude based on a multi-layer, multi-pass welding path using a weighted principal component analysis normal estimation method.

[0073] This disclosure provides a method and system for adaptive generation of multi-layer, multi-pass welding paths for robots based on 3D point clouds. The method acquires 3D point cloud data of the welding surface using a TOF sensor and combines this data with the geometric features of the workpiece to automatically plan the number of layers and passes required for multi-layer, multi-pass welding, and then adaptively generates the welding path. This method not only improves welding accuracy and efficiency but also enhances the system's adaptability to different welding tasks. By adapting to various workpiece types, the system can achieve efficient and high-precision automated operation in medium-thick plate welding, significantly reducing the need for manual intervention, minimizing errors and defects, and thus effectively improving welding efficiency and quality. Specific advantages are as follows:

[0074] 1. This invention proposes a random sample consistent weld seam segmentation algorithm based on density clustering. This algorithm can not only handle simple planar structures, but also cope with complex curved surface structures, and can be effectively applied to the curved surface segmentation of medium and thick plate welded workpieces.

[0075] 2. This invention fully considers the diversity and complexity of welded workpieces in industrial production, and designs multi-layer, multi-pass welding paths for various common medium and thick plate welds. It is widely applicable to welded workpieces of various shapes and sizes in industrial production and can be efficiently applied to multi-layer, multi-pass welding technology.

[0076] 3. Experiments were conducted on various welded workpieces using this invention, and the maximum error (ME) and root mean square error (RMSE) of the generated multi-layer, multi-pass weld paths were controlled below 1.103 mm and 0.677 mm, respectively. Furthermore, the time required is significantly reduced compared to the traditional "teach and playback" mode. This achieves an efficient and stable weld path generation process.

[0077] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0078] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0079] Figure 1 This is a schematic diagram of the adaptive generation method for multi-layer, multi-pass welding paths of robots in an embodiment of the present invention;

[0080] Figure 2 This is a schematic diagram illustrating the principle of the adaptive generation method for multi-layer, multi-pass welding paths of robots in this embodiment of the invention.

[0081] Figure 3 This is a point cloud noise map of the V-groove weld in an embodiment of the present invention;

[0082] Figure 4 This is a flowchart of the point cloud preprocessing process in an embodiment of the present invention. Figure 4 (a) is the view before point cloud processing; Figure 4 (b) Display the graph after removing outliers; Figure 4 (c) is the display image after point cloud downsampling;

[0083] Figure 5 This is a flowchart of weld seam segmentation in an embodiment of the present invention. Figure 5 (a) is a rendering of the original point cloud data of the weld; Figure 5 (b) is a diagram showing the effect of weld seam segmentation;

[0084] Figure 6 This is a diagram showing the effect of weld feature extraction in an embodiment of the present invention. Figure 6 (a) is a point cloud ROI region map; Figure 6 (b) is a diagram showing the feature extraction results of the ROI region;

[0085] Figure 7 This is a comparison diagram of interpolation of weld feature points in an embodiment of the present invention. Figure 7 (a) Image of feature points without interpolation; (b) Image of interpolated feature points.

[0086] Figure 8 This is a schematic diagram of the overall and cross-sectional view of the V-groove weld in an embodiment of the present invention;

[0087] Figure 9 This is a schematic diagram of the overall and cross-sectional view of the lap weld in an embodiment of the present invention;

[0088] Figure 10 This is a schematic diagram of the overall and cross-sectional view of the S-shaped butt weld in an embodiment of the present invention;

[0089] Figure 11 This is a schematic diagram illustrating the stress principle of molten droplets during welding of the V-groove weld seam vertical plate in an embodiment of the present invention. Figure 11 (a) Schematic diagram of the force principle of the intermediate molten droplet; Figure 11 (b) Schematic diagram of the force principle of the molten droplets at both ends;

[0090] Figure 12 This is a schematic diagram of the welding path planning for the vertical plate of the V-groove weld in an embodiment of the present invention;

[0091] Figure 13 This is a comparison image of point cloud normal vector estimation in an embodiment of the present invention. Figure 13 (a) shows the PCA point cloud normal vector estimation results; Figure 13 (b) shows the weighted PCA point cloud normal vector estimation results;

[0092] Figure 14This is a schematic diagram of the robot multi-layer multi-pass welding path adaptive generation system in an embodiment of the present invention;

[0093] Figure 15 This is a diagram illustrating the MLMP welding path generation effect of the V-groove weld in an embodiment of the present invention. Figure 15 (a) is a diagram showing the original data results of the weld; Figure 15 (b) is a diagram showing the weld seam segmentation results; Figure 15 (c) is a diagram showing the results of weld feature extraction; Figure 15 (d) is a schematic diagram of the MLMP welding path and the end position of the welding torch.

[0094] Figure 16 This is a diagram illustrating the MLMP welding path generation effect of the lap weld in an embodiment of the present invention. Figure 16 (a) is a diagram showing the original data results of the weld; Figure 16 (b) is a diagram showing the weld seam segmentation results; Figure 16 (c) is a diagram showing the results of weld feature extraction; Figure 16 (d) is a schematic diagram of the MLMP welding path and the end position of the welding torch.

[0095] Figure 17 This is a diagram illustrating the MLMP welding path generation effect of the S-shaped butt weld in an embodiment of the present invention. Figure 17 (a) is a diagram showing the original data results of the weld; Figure 17 (b) Figure 17 (c) is a diagram showing the weld seam segmentation results; Figure 17 (d) shows the result of weld feature extraction; Figure 17 (e) Figure 17 (f) is a schematic diagram of the MLMP welding path and the attitude of the welding torch end; Figure 17 (g) Figure 17 (h) Figure 17 (i) are respectively Figure 17 (d) Figure 17 (e) Figure 17 (f) is a magnified view of a portion of the image.

[0096] Figure 18 This is a comparison diagram of the simulated V-groove weld and the actual weld in an embodiment of the present invention. Figure 18 (a) is a drawing of the workpiece before welding; Figure 18 (b) is a schematic diagram of the simulated MLMP welding path; Figure 18 (c) is a picture of the workpiece after welding. Detailed Implementation

[0097] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention. Furthermore, it should be noted that, for ease of description, only the parts relevant to the present invention are shown in the drawings, not the entire structure.

[0098] Before discussing the exemplary embodiments in more detail, it should be noted that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe the steps as sequential processes, many of these steps can be performed in parallel, concurrently, or simultaneously. Furthermore, the order of the steps can be rearranged. The process can be terminated when its operation is complete, but may also have additional steps not included in the figures. The process can correspond to a method, function, procedure, subroutine, subroutine, etc.

[0099] In the field of medium and heavy plate welding, multi-layer, multi-pass welding technology, characterized by high precision and wide applicability, has attracted much attention. However, the traditional "teaching and playback" mode for welding robots relies on manual teaching, which is not only inefficient but also difficult to adapt to the real-time dynamic adjustment requirements of complex welds, especially lacking flexibility in multi-layer, multi-pass welding. Furthermore, current weld feature extraction technologies are mostly based on two-dimensional planar information, ignoring the spatial geometric characteristics of the weld, resulting in limited ability to express the weld shape during welding path generation. Simultaneously, existing multi-layer, multi-pass welding path generation methods are only applicable to V-groove welds, exhibiting poor adaptability to complex weld shapes or different types of welded workpieces. To address these issues, this embodiment proposes an adaptive multi-layer, multi-pass welding path generation method for robots based on 3D point clouds. This method can segment the 3D point cloud of the welded workpiece containing complex curved surfaces to extract the feature points of the weld, and then generate multi-layer, multi-pass welding paths using the weld feature points and the geometric information of the welded workpiece to solve the problems of low precision, slow speed, and limited applicability in medium and heavy plate welding.

[0100] An embodiment of an adaptive generation method and system for multi-layer, multi-pass welding paths in robots is provided below:

[0101] An adaptive generation method for multi-layer, multi-pass welding paths in robots, such as Figure 1 As shown, the method includes the following steps:

[0102] S1. A time-of-flight sensor is used to collect the three-dimensional point cloud of the weld seam, and the collected weld seam point cloud is preprocessed.

[0103] S2. The preprocessed weld point cloud is segmented by a density clustering-based random sample consistent weld segmentation algorithm to extract weld features, and the feature points of the weld are interpolated and fitted using a cubic spline interpolation method.

[0104] S3. Generate multi-layer, multi-pass welding paths based on the geometric information of the workpiece and the interpolated weld feature points.

[0105] S4. Based on multi-layer, multi-pass welding paths, the normal estimation method of weighted principal component analysis is used to calculate the welding torch end posture.

[0106] Specifically, the overall flowchart of the embodiment is as follows: Figure 2 As shown, a time-of-flight (TOF) sensor is used to acquire the 3D point cloud of the weld seam. The acquired weld seam point cloud is preprocessed to remove noise points and reduce computational complexity. Then, a density-based random sample consensus weld seam segmentation algorithm is used to segment the weld seam and extract its features. Cubic spline interpolation is then used to interpolate and fit the weld seam feature points. Furthermore, a multi-layer, multi-pass welding path is generated based on the geometric information of the workpiece and the weld seam feature points. Finally, the welding torch tip attitude is calculated using a weighted PCA normal estimation method.

[0107] In S1, the preprocessing of the acquired weld point cloud includes using a radius filtering method to remove outliers and using a voxel-based point cloud downsampling method to reduce the amount of point cloud data.

[0108] Point cloud preprocessing is a crucial part of the point cloud processing workflow, typically including two main steps: outlier removal and downsampling. These steps help to accurately analyze point cloud data and provide a more reliable foundation for subsequent processing and analysis. First, outlier removal aims to eliminate noise and outliers caused by sensor inaccuracies and the surrounding environment during the extraction of the 3D point cloud of the welded workpiece, ensuring the accuracy and stability of subsequent processing. To address this issue, radius filtering can be used to remove these outliers. This method calculates the number of points within the neighborhood of each point in the point cloud; if the number of neighborhood points is less than a set threshold, the point is marked as an outlier or noise point.

[0109] The point cloud downsampling method based on voxel filtering is employed to reduce the amount of point cloud data, simplify computational complexity, and preserve sufficient spatial structural features. This method divides the point cloud data into voxel units, calculates the center position of each voxel unit, and selects the point closest to the center as the representative point. This effectively reduces the amount of data, removes noise, and does not lose important spatial structural information.

[0110] The embodiment directly uses a TOF sensor to obtain the three-dimensional point cloud of the weld. Due to the low accuracy of the sensor, the extracted point cloud usually contains a lot of noise points. Figure 3 Several examples with significant noise were identified in the bidding. Therefore, the point cloud preprocessing method proposed above is used to denoise and downsample the 3D point cloud of the welded workpiece to enhance the robustness of the proposed method. The specific process principle is as follows: Figure 4 As shown.

[0111] The random sample consistent weld seam segmentation algorithm based on density clustering in S2 specifically includes the following steps:

[0112] The iterations are performed using the following steps until the maximum number of iterations is reached:

[0113] Initialize three non-collinear points randomly from the point cloud set.

[0114] Calculate the plane equation based on the initialized points;

[0115] Calculate the distance from all points to the plane based on the plane equation, and obtain the total number of all interior points whose distance from a point to the plane is less than a threshold.

[0116] If the number of interior points is greater than the threshold, calculate the plane equation represented by each interior point as the input data for extracting weld feature points, and remove each interior point from the point cloud set.

[0117] After the iteration is complete, if there are any remaining point clouds in the point cloud set, perform the following steps on each of the remaining point clouds:

[0118] Using the point cloud as the center and the radius distance threshold as the radius, calculate the number of points inside the circle;

[0119] If the number of points inside the circle is greater than the threshold for the number of adjacent points, then mark the point cloud with the circle as the center, and randomly select another point cloud inside the circle until all the remaining point clouds have been marked.

[0120] The point clouds are classified according to the labels, and the final segmented point cloud results are obtained.

[0121] Specifically, after preprocessing the 3D point cloud of the welded workpiece, it needs to be segmented into different layers, and then the feature points of the weld are extracted. Random Sample Consensus (RANSAC) is a highly efficient, accurate, and robust model fitting algorithm. This method estimates model parameters based on the principle of hypothesis testing. Its core idea is to eliminate the influence of outliers through an iterative process when estimating model parameters, thereby obtaining a set of data samples unaffected by outliers. This method can perform planar fitting on the welded workpiece, thus segmenting it into different planes. However, in real welding environments, most welded workpieces also contain curved surface structures, and directly using a RANSAC-based segmentation algorithm cannot accurately segment the welded workpiece into different parts. Therefore, this embodiment presents a density-based clustering-based random sample consensus weld segmentation algorithm (DC-RANSAC), mainly divided into two parts: planar fitting and clustering. The specific process is shown in Algorithm 1:

[0122]

[0123] The proposed 3D point cloud segmentation method can effectively segment different surfaces of the welded workpiece and obtain the corresponding plane equations of the planar regions as input data for extracting weld feature points, such as... Figure 5 As shown.

[0124] Extracting weld features in S2 includes identifying and extracting the connecting lines between different surfaces within the weld area from the segmented point cloud, and further analyzing geometric features. The specific steps include: taking the point cloud closest to plane P1 on plane P2 as feature points, and these feature points constitute the weld path of the corner joint.

[0125] Before extracting weld features in S2, the region of interest is extracted from the point cloud, and a distance threshold from the point to the plane is set when extracting feature points.

[0126] Specifically, the features of a weld can be interpreted as the connecting lines between different surfaces within the weld region of a welded workpiece. Weld connecting lines can be considered as straight lines or curves connecting two adjacent surfaces in the welded workpiece. These connecting lines typically represent the geometry of the weld joint, including its length, shape, and orientation. Weld feature extraction refers to identifying and extracting these connecting lines from the point cloud and further analyzing their geometric features. To achieve robust weld feature extraction, we propose a distance-based weld feature extraction algorithm. For corner joints, such as... Figure 6 As shown, the point on plane P2 that is closest to plane P1 is considered as a feature point, and these points constitute the weld path of the corner joint. The distance from the point to the plane is calculated using equation (1).

[0127]

[0128] To mitigate the impact of noise on feature extraction, an appropriate threshold must be selected. The choice of threshold is crucial to the extraction results. If the threshold is too small, the number of feature points will decrease, failing to adequately represent the geometric features of the weld. Conversely, if the threshold is too large, the obtained feature points may include noise points surrounding the weld, which will negatively affect the accuracy of welding path planning when calculating the weld's geometric features.

[0129] Point cloud data typically contains a large number of data points, and directly calculating the distances between these points would significantly increase the computational load. To maintain computational efficiency, Regions of Interest (ROIs) are extracted from the point cloud. Similar to ROI extraction in images, the area around the weld can be designated as an ROI based on the appearance features of the welded workpiece. This reduces the number of points that need to be calculated, thereby improving computational efficiency. Figure 6 (b) shows the ROI regions of the point cloud and the results of ROI extraction. ROI extraction not only helps reduce computational burden but also allows focusing on the region of interest, thus enabling more accurate analysis and processing of weld data.

[0130] After acquiring the weld feature points, interpolation is required for the sparse and discontinuous data to ensure continuity and local smoothness in the transition areas between weld joints. This minimizes errors between data points, making the weld data as continuous and smooth as possible to facilitate subsequent welding path planning.

[0131] Cubic spline interpolation is a commonly used interpolation method. It ensures the interpolation curve passes through all nodes while maintaining the continuity of the first and second derivatives, providing a smooth curve shape. Simultaneously, it possesses good local controllability, effectively reducing overall error and avoiding drastic fluctuations. Therefore, cubic spline interpolation has the advantages of high accuracy and good stability. This embodiment uses this method to compensate for weld feature points; the specific principle is shown below.

[0132] The goal of cubic spline interpolation is to find multiple cubic polynomials to connect adjacent data points, making the entire curve smooth and continuous. Suppose we have a set of data points (x0, y0), (x1, y1), ..., (x... n y n A cubic polynomial is defined between every two adjacent points. This polynomial is defined as S. i (x)=a i x 3 +b i x 2 +c i x+d i Where i = 0, 1, ..., n-1, a total of n polynomials are needed. For each interval [x i x i+1 Construct a cubic polynomial:

[0133] S i (x)=a i (xx i ) 3 +b i (xx i ) 2 +c i (xx i )+d i (2)

[0134] Wherein, coefficient a i b i c i d i It is pending confirmation.

[0135] Each cubic spline interpolation function S i (x) must pass through two nodes of its defined interval, i.e., satisfy the following condition:

[0136]

[0137] Meanwhile, to ensure the smoothness of the interpolation curve, the derivatives of the cubic polynomial at the nodes in each interval must be continuous of first and second order:

[0138]

[0139] And the second derivatives at both ends are zero, that is, S″0(x0)=0, S″ n-1 (x n ) = 0.

[0140] Transforming the above conditions into a system of equations yields 4n unknowns. By using interpolation and continuity conditions, a total of 4n equations are constructed, thus uniquely determining all coefficients 'a'. i b i c i d i .

[0141] For each interval [x] i x i+1 The coefficients of ] can be expressed in the form of a tridiagonal system of equations:

[0142] Am=d (5)

[0143] Where A is a tridiagonal matrix, and m is an unknown second derivative vector, the second derivative value for each interval can be obtained by solving this equation, and then each S can be derived from it. i The coefficient of (x).

[0144] Comparison before and after interpolation Figure 7 As shown. For Figure 7 The S-shaped butt weld shown is typically welded using a swaying welding method. Swaying welding is a commonly used welding technique. During swaying welding, the welding robot oscillates left and right to complete the welding of the workpiece. Therefore, it is essential to ensure that the number of weld points at both ends of the swaying welding process is consistent to minimize unnecessary errors. Therefore, we use interpolation to control the number of feature points on both sides of the S-shaped butt weld to ensure the smooth progress of the swaying welding.

[0145] To ensure weld quality in multi-layer, multi-pass welding processes, the welding path generation must consider the position and orientation of the welding torch tip. The selection of the torch tip position needs to take into account the distance between the weld and the weld joint to ensure that the welding wire accurately and evenly fills the weld, avoiding problems such as incomplete welds or slag. The welding torch orientation allows for flexible movement of the welding path in three-dimensional space, ensuring that the welding wire is correctly delivered to the weld area, thereby achieving complete filling and a good bond in the weld joint. Based on the process parameters in MLMP welding, the cross-sectional area of ​​the weld bead can be calculated using the following formula:

[0146]

[0147] In the formula, D is the diameter of the welding wire, W f η is the wire feed speed, η is the welding cladding rate, and v is the welding speed.

[0148] Existing multi-layer, multi-pass welding path generation methods are mostly limited to V-groove welds and are difficult to apply to other types of welded workpieces. Therefore, this embodiment proposes a multi-layer, multi-pass welding path generation method based on the geometry of the welded workpiece.

[0149] (1) Fillet weld: The V-groove is a typical example of a fillet weld, and its weld cross-section diagram is shown below. Figure 8 As shown, a multi-layer, multi-pass welding path is planned based on the extracted weld edge features A, B, and C. The number of welding layers is determined by the groove depth and the thickness of each weld pass. To reduce errors and improve the accuracy of path planning, the groove depth is calculated by combining information from all extracted weld edge feature points. Therefore, the number of welding layers L′ is:

[0150]

[0151] In the formula, z Ai z Bi z Ci Let A, B, and C represent the depth coordinates of the i-th weld feature point in weld edge features A, B, and C, respectively; n represents the number of weld feature points; h is the predetermined weld bead thickness; and th is the set error threshold. To ensure the weld bead precisely covers the entire weld, the predetermined weld bead thickness needs to be adaptively adjusted. When the remaining depth is within the allowable error range, the number of weld bead layers is rounded down; otherwise, the number of weld bead layers L′ = [L] + 1, and the predetermined weld bead thickness is adjusted as follows:

[0152]

[0153] When using multi-layer, multi-pass welding for V-grooves, the initial welding is performed first to obtain new weld features D and E. Next, the first pass of each layer is welded starting from the right side of the weld. The number of weld passes should be determined by both the groove width of that layer and the width of the weld pass. Therefore, the number of weld passes per layer is P′. m As shown below:

[0154]

[0155] In the formula, z Ai z Bi z Di z Ei Let y represent the depth coordinates of the i-th weld feature point in weld edge features A, B, D, and E, respectively. Ai y Bi y Diy Ei Let w represent the width coordinates of the i-th weld feature point in weld edge features A, B, D, and E, respectively, where w is the predetermined weld bead width, m is the current weld layer number, and D... m Let P be the width of the m-th layer section. To ensure that the weld bead is exactly covered by the weld seam, the predetermined weld bead width needs to be adaptively adjusted. m -[P m When ] ≤ th, th is the set error threshold, representing the remaining width th·w after rounding to the nearest whole number (w). If the remaining width is within the allowable error range, then for P... m Round down to the nearest integer to ensure the number of weld passes is an integer. If P m -[P m ]>th indicates that the remaining width is too large, requiring an additional weld pass, P′ m =[P m +1, and adjust the predetermined weld width as follows:

[0156]

[0157] After calculating the number of weld layers and passes in each layer, the coordinate positions of each pass can be further determined, and the weld positions on each cross-section can be planned. The height between two adjacent layers is the corrected weld thickness h′, and the width between two adjacent passes in the same layer is the corrected weld width w′. Therefore, the coordinates W of the j-th pass in the i-th layer are... ij for:

[0158]

[0159] In the formula These represent the average width coordinates of weld features B and E, respectively. These represent the average depth coordinates of weld features B and E, respectively.

[0160] (2) Lap Weld: A lap weld is formed by joining two workpieces together. In a lap weld, one side of the welded component (usually called the cover) partially covers the other side (usually called the base). A cross-sectional diagram of a lap weld is shown below. Figure 9 As shown.

[0161] Weld features B and C are obtained based on the extracted weld feature point A and the preset welding height H. Then, the number of welding layers is calculated using the predetermined weld thickness h and error range th. The number of layers L′ is:

[0162]

[0163] The thickness of the weld bead is adjusted to obtain h′. The number of weld passes per layer is determined by the cross-sectional width of that layer and the width of the weld bead, where m is the current weld layer number, and D...m Let be the width of the m-th layer cross-section. The cross-section width can be calculated based on the similarity ratio, and then the number of weld passes can be calculated based on the error range th.

[0164]

[0165] Similarly, the width of the weld bead needs to be adaptively adjusted according to the error range, and the adjusted w′ is shown in equation (12). Then, it can be determined according to the number of weld layers L′ and the number of passes P′ per layer. m Further determine the starting coordinate position W of the j-th channel in the i-th layer. ij :

[0166]

[0167] In the formula, This represents the average width coordinate value of weld feature A. This represents the average depth coordinate value of weld feature A.

[0168] (3) Butt welds: For butt joints with irregular shapes, oscillation welding was used. Oscillation welding is a commonly used welding technique. During oscillation welding, the welding robot oscillates periodically left and right at a specific angle along the welding direction to control the formation of the weld pool and weld bead. Oscillation welding has the following advantages: First, it is suitable for various weld shapes and sizes, whether straight, curved, or irregular; second, by controlling welding parameters and operating skills, the stability and consistency of the welding process can be achieved, ensuring welding quality; in addition, oscillation welding can improve the uniformity and aesthetics of the weld, and reduce residual stress and deformation. A schematic diagram of a butt weld with an S-shaped curve is shown below. Figure 10 As shown.

[0169] The number of layers required for welding is calculated using the extracted weld feature connection lines A, B, C, and D, as well as the predetermined weld thickness h and allowable error range th. The number of layers L′ is:

[0170]

[0171] In the formula, z Ai z Bi z Ci z Di Let represent the depth coordinates of the i-th weld feature point in weld edge features A, B, C, and D, respectively, and n represent the number of weld feature points.

[0172] Adaptive adjustment of weld bead thickness:

[0173]

[0174] After calculating the number of welding layers, it is necessary to calculate the coordinates of the start and end points of the welding path, and then perform lap welding from the start to the end point. The starting coordinates of the j-th pass in the i-th layer are... and endpoint coordinates for:

[0175]

[0176] In the formula, y Cj y Dj Let z represent the width coordinates of the j-th weld feature point in weld features C and D, respectively. Cj z Dj These represent the depth coordinates of the j-th weld feature point in weld features C and D, respectively.

[0177] (4) Vertical Plate Welding: During the welding process, liquid droplets of metal form at the end of the welding wire. This molten metal is usually called a droplet, and the phenomenon of it falling from the end of the welding wire onto the workpiece metal under the influence of surface tension and gravity is called the droplet dripping phenomenon. In actual welding, for vertically placed welding workpieces, the droplets are more likely to drip under the influence of gravity. This phenomenon will change the shape of the workpiece surface, causing the actual welding path to deviate from the predetermined path, affecting the welding accuracy.

[0178] The stress analysis of the molten droplets generated during the welding process of the vertical plate is as follows: Figure 11 As shown. Figure 11 (a) represents the force situation of the molten droplet in the middle of the V-groove weld, which is mainly subjected to gravity G, surface tension F of the V-groove workpiece, and arc force P of the welding torch.

[0179] The gravitational force G acting on the molten droplet is vertically downward, exacerbating the droplet's downward flow. The surface tension F acting on the droplet is mainly due to the restraining effect of the V-groove workpiece's sidewalls and the condensed weld metal, which can alleviate the droplet's downward flow to some extent. The direction of the arc force acting on the droplet is controlled by the welding torch's tilt angle. When the torch is tilted upward, the arc force can be decomposed into vertically upward and horizontal components; the vertically upward component can alleviate the droplet's downward flow to some extent. The magnitude of the arc force is determined by the welding speed. Too slow a welding speed will cause the droplet to overflow from the weld, damaging the workpiece structure. Too fast a welding speed will result in uneven weld cladding, affecting the weld's formation quality.

[0180] Figure 11(b) illustrates the forces acting on the molten droplet at both ends of the V-groove weld. Besides the forces of gravity (G), the surface tension (F) exerted by the V-groove workpiece, and the arc force (P) of the welding torch, the droplet is also subjected to a supporting force (N) from the lower end of the workpiece. Although the arc force of the welding torch must be tilted downwards to avoid collision between the torch and the workpiece, thus exacerbating the droplet's downward flow, the supporting force (N) remains vertically upwards. This ensures that the net force on the droplet in the vertical direction is zero, thereby mitigating the droplet's downward flow caused by gravity.

[0181] In actual welding processes, welding speed often needs to be set based on parameters such as the materials of the workpiece and the solder, and the weld cladding rate. Therefore, for example... Figure 11 In (a), the molten droplets in the middle of the V-groove weld cannot be simply controlled by adjusting the welding speed to prevent the droplets from flowing downwards due to gravity. Therefore, we... Figure 11 (b) The droplet stress condition was designed as follows: Figure 12 The welding path shown is used to address the droplet dripping phenomenon. Extraction is as follows: Figure 12 The weld feature points A, B, and C are shown. These feature points are located in the XOY plane, with the CB direction being the positive X-axis direction.

[0182] The included angle θ of the V-groove weld is:

[0183]

[0184] Based on the known weld thickness h and weld angle θ, the number of weld passes P can be calculated as follows:

[0185]

[0186] In the formula, y Ai y Bi y Ci represents the width coordinate value of the i-th weld feature point in weld features A, B, and C, respectively; n represents the number of weld feature points; and th is the set error threshold.

[0187] The adaptive adjustment of weld bead thickness h′ is as follows:

[0188]

[0189] After calculating the number of weld passes, it is necessary to calculate the coordinates of the start, intermediate, and end points of the weld path, and then perform traversal welding from the start point to the intermediate point and then to the end point. The coordinates of the start, intermediate, and end points of the j-th pass in the i-th layer are given. and for:

[0190]

[0191] In the formula, x Aj x Bjx Cj These represent the length coordinates of the j-th weld feature point in weld features A, B, and C, respectively. These represent the average width coordinates of weld features A, B, and C, respectively. Using this bottom-up welding path significantly reduces droplet transfer caused by gravity.

[0192] The welding torch posture mainly refers to the tilt angle of the welding torch, which needs to be perpendicular to the plane of the welding path, i.e., the normal vector of the plane containing the welding path. A common method for calculating the plane normal vector is the principal component analysis (PCA) algorithm. Its principle is to estimate the local plane by using the least squares method for the points in the neighborhood of the current point, and regard the principal direction of the plane as the normal vector of that point.

[0193] Traditional PCA methods have certain limitations, as they assign equal weights to points in the neighborhood of the normal vector estimation. However, when the neighborhood contains many noisy points and points not belonging to the plane (e.g., at the junction of a weld bevel and weld bead), the estimation accuracy of the normal vector is not high. Figure 13 As shown. Therefore, in order to improve the accuracy of normal vector estimation, it is necessary to estimate the neighborhood point set S. i point p in i Different weights are assigned. As the distance between neighboring points and the center point increases, the importance of tangent plane projection estimation gradually decreases. The specific algorithm is shown below:

[0194] (1) For each point p i Select its surrounding neighborhood point set S i .

[0195] (2) Calculate the neighborhood point set S i point p in i Euclidean distance d to neighborhood center point c i d i for:

[0196] d i =||p i -c|| (26)

[0197] (3) Define the weights of different points based on the Euclidean distance of points in the neighborhood from the center. The formula for the weight w is:

[0198]

[0199] Where d is the average distance between neighboring points and the center point. th is a manually set distance threshold that decreases as the density of points in the neighborhood increases.

[0200] (4) Set the neighborhood point set S iThe coordinates of the points in the matrix are subtracted from the coordinates of the center point c to achieve mean centering of the coordinates, and then the weighted covariance matrix is ​​minimized.

[0201]

[0202] In the formula, N represents the number of points in the neighborhood, and w i p is the weight of the i-th point. i Let represent the spatial coordinates of the i-th point, c represent the spatial coordinates of the neighborhood center point, and n be the normal vector. The eigenvector corresponding to the smallest eigenvalue is obtained through singular value decomposition (SVD) of the weighted covariance matrix, and this eigenvector is used as the normal vector for that point.

[0203] Another embodiment illustrates a robotic multi-layer, multi-pass welding path adaptive generation system 140, such as... Figure 14 As shown, system 140 includes:

[0204] The weld seam point cloud acquisition module 141 is used to acquire the three-dimensional point cloud of the weld seam using a time-of-flight sensor and to preprocess the acquired weld seam point cloud.

[0205] The weld feature point acquisition module 142 is used to segment the preprocessed weld point cloud by using a density clustering-based random sample consistent weld segmentation algorithm to extract weld features, and to use a cubic spline interpolation method to interpolate and fit the weld feature points.

[0206] The multi-layer multi-pass welding path generation module 143 is used to generate multi-layer multi-pass welding paths based on the geometric information of the welded workpiece and the weld feature points after interpolation fitting.

[0207] The welding torch end attitude acquisition module 144 is used to calculate the welding torch end attitude based on a multi-layer, multi-pass welding path using a weighted principal component analysis normal estimation method.

[0208] In addition to the above module, system 140 may also include other components; however, since these components are not relevant to the content of this disclosure, their illustrations and descriptions are omitted here.

[0209] Other specific working processes of the robot multi-layer multi-pass welding path adaptive generation system 140 are described in the above-described embodiment of the robot multi-layer multi-pass welding path adaptive generation method, and will not be repeated here.

[0210] To verify the effectiveness of the method of the present invention, a welding robot system was built using a commercially available welding robot. The welding robot system mainly consists of two parts: an industrial robot system and a sensor system. The robot system includes a robotic arm, a welding controller, and a workpiece. The sensor system includes a Time-of-Flight (TOF) sensor and a vision controller.

[0211] To verify the effectiveness of the proposed algorithm, a series of experiments were conducted, focusing on verifying the accuracy and efficiency of its generation of multi-layer, multi-pass welding paths in different types of welds. These experiments included tests on corner welds, lap welds, and butt welds.

[0212] Welding path generation

[0213] (1) Simulation Experiments: Three-dimensional simulation experiments were conducted for three typical weld types (V-groove weld, lap weld, and S-shaped butt weld). First, the welded workpieces were modeled using 3D modeling technology to obtain their 3D point cloud data. These point cloud data contain detailed information about the weld and its surrounding area, but also contain a certain amount of noise. To ensure the accuracy of subsequent analysis, a radius-based denoising method was used to remove isolated noise points in the point cloud, and voxel filtering was combined with point cloud downsampling to further reduce data redundancy. Next, the DC-RANSAC algorithm was used to accurately segment the weld point cloud. This algorithm segments the welded workpiece point cloud into different geometric regions, providing clear input data for subsequent weld feature point extraction. In the segmentation experiment of the V-groove weld, the DC-RANSAC algorithm showed good stability and robustness, and could effectively separate the weld area from the non-weld area. The segmentation results of the V-groove weld are as follows: Figure 15 As shown in (b), during the weld feature point extraction stage, based on the distribution characteristics of the weld geometry in the point cloud data, a distance-based feature extraction method is used to accurately locate the weld feature points. To improve the continuity and smoothness of the weld trajectory, a cubic spline interpolation method is further used to generate a continuous trajectory curve for the weld. This curve can fully reflect the spatial orientation of the weld, providing an accurate reference for subsequent welding path planning. The feature point extraction results for the V-groove weld are shown in Figure [image missing]. Figure 15 As shown in (c). Finally, based on the multi-layer, multi-pass welding path generation algorithm, the multi-layer, multi-pass welding path and the welding torch end posture of the V-groove weld were calculated. The multi-layer, multi-pass welding path and the welding torch end posture are as follows: Figure 15 As shown in (d).

[0214] The experimental results for lap welds and S-type butt welds are as follows: Figure 16 and Figure 17 As shown, the DC-RANSAC algorithm can accurately segment point clouds of V-groove welds, and is also applicable to the segmentation of lap welds and complex S-shaped butt welds. Experimental results show that this method has accurate segmentation results and high robustness, and can flexibly adapt to complex working conditions with different weld shapes.

[0215] (2) Actual Welding: To verify the practical application effect of the proposed path generation method, a V-groove weld was selected for actual welding experiments. During the experiment, the welding process parameters were set as follows: wire diameter D was 1 mm, and wire feed speed W... f The welding speed is 6 m / min, the welding cladding rate η is 0.92, and the welding speed v is 6 mm / s. Based on these parameters, the cross-sectional area of ​​the weld bead can be calculated as 12.04 mm² according to equation (7). In addition, the bottom edge width of the V-groove welded workpiece is 0.9 mm. The groove height is 35 mm, the number of welding layers is 7, the number of welding passes is 36, and the allowable error th is set to 1.5 mm. The final welding result is as follows. Figure 18 As shown in (c). Figure 18 This image shows a comparison between the simulated welding path and the actual welding of a V-groove weld. The image demonstrates that the simulated MLMP welding path can be effectively applied to actual welding, with the generated trajectory highly consistent with the actual weld morphology, meeting the path accuracy requirements of MLMP welding. Furthermore, the weld filling effect shows that the actual weld is uniform and continuous in shape, with a smooth surface and no obvious defects, meeting the quality requirements of industrial welding. The consistency between the simulated path and the actual welding results verifies the effectiveness and robustness of the simulation model, proving that the method proposed in this invention is not only applicable to path planning in experimental environments but can also be successfully extended to actual industrial scenarios, providing an important basis for the development of intelligent welding systems.

[0216] Overall, the experimental results demonstrate that the adaptive generation method for multi-layer, multi-pass welding paths based on 3D point clouds exhibits significant adaptability, accuracy, and stability. This method can adaptively estimate the number of layers and passes required for welding based on the characteristics of different weld shapes, and generate accurate welding paths and welding torch end postures for each weld pass. The experiments also show that the algorithm demonstrates good adaptability and robustness across different weld types, such as V-groove welds, lap welds, and S-shaped butt welds, providing technical support for the stable application of complex welding processes. These experiments indicate that the adaptive generation method for multi-layer, multi-pass welding paths based on 3D point clouds provides a reliable welding path generation scheme for the engineering application of multi-layer, multi-pass welding processes, effectively supporting various welding needs in actual production.

[0217] Accuracy and stability verification

[0218] To further verify the accuracy and stability of the multi-layer, multi-pass welding path generation method proposed in this invention, three-dimensional models of three different types of welds were created, including V-groove welds, lap welds, and S-shaped butt welds. These three-dimensional models were then sampled to obtain their three-dimensional point cloud data. Based on the point cloud data, Gaussian noise was added to simulate the complex conditions in actual welding environments, thereby constructing several representative point cloud datasets.

[0219] Next, an adaptive multi-layer, multi-pass welding path generation method based on 3D point clouds was used to generate multi-layer, multi-pass welding paths for both point clouds with added Gaussian noise and the original point clouds. On the generated welding paths, the maximum error (ME) and root mean square error (RMSE) in the X, Y, and Z directions, as well as their average values, were measured as performance evaluation metrics to quantitatively assess the algorithm's accuracy and stability. These error metrics intuitively reflect the degree of deviation between the generated path and the ideal path, thus helping to evaluate the algorithm's performance. The specific ME and RMSE are shown in Table 3; this error represents the difference between the welding path generated from the point cloud with Gaussian noise and the welding path generated from the original point cloud. Experimental results show that the proposed method generates welding paths with high accuracy and strong robustness.

[0220] Table 3. MLMP Welding Path Errors for V-groove Welds, Lap Welds, and S-type Butt Welds

[0221]

[0222] As shown in Table 3, the maximum error (ME) of the welding path is generally no higher than 1.103 mm, and the maximum root mean square error (RMSE) is 0.677 mm, indicating that the path generation method has high accuracy in practical applications. In terms of error distribution, the error in the X direction is generally smaller than the errors in the Y and Z directions. This may be because the mean values ​​of the y and z coordinates of the weld feature points are used to estimate the y and z coordinates of the welding path in the path calculation. However, there may be a certain number of outliers in the weld feature point data. The sensitivity of these outliers to the mean value causes the calculation results to deviate from the true values, thus increasing the errors in the Y and Z directions.

[0223] Furthermore, the performance of the proposed method was tested under different noise intensities. Specifically, the error between the welding path generated by the algorithm and the expected welding position in the V-groove weld was analyzed after adding Gaussian noise with different variances. The error results are shown in Table 4.

[0224] As shown in Table 4, the maximum error (ME) and root mean square error (RMSE) of this method under different Gaussian noise environments are controlled within 0.194 mm and 0.028 mm, respectively, with a small error fluctuation range. This indicates that the proposed method can maintain good stability and accuracy under complex noise conditions, reliably extract weld features, and generate high-quality welding paths. Even under strong noise conditions, the algorithm can effectively cope with environmental interference, and the generated path maintains a high degree of consistency with the expected path.

[0225] Table 4. Experimental results of MLMP welding path generation under different noise intensities.

[0226]

[0227] This superior performance is primarily attributed to the algorithm's design features. First, in the point cloud denoising and segmentation stage, the algorithm comprehensively considers the local density characteristics of the point cloud data and combines this with a robust outlier handling strategy to effectively filter out interference from noise points in weld feature extraction. Second, in the multi-layer, multi-pass welding path generation stage, the algorithm dynamically adjusts path planning parameters, improving its tolerance to noise-induced deviations and making the path generation process highly robust to point cloud errors.

[0228] Furthermore, experimental results show that even under high noise levels, the generated welding path still meets the precision requirements of industrial production. This further verifies the potential and practicality of the proposed method in real-world industrial applications. In summary, the stability and noise resistance of this method provide a reliable solution to the path planning problem in complex welding environments, laying a solid technical foundation for the development and application of intelligent welding systems.

[0229] Algorithm efficiency analysis

[0230] Furthermore, the efficiency of the algorithm was verified in detail. The running time of each part of the proposed algorithm was analyzed and calculated to comprehensively evaluate its efficiency performance in practical applications. For this purpose, experiments were specifically conducted on V-groove welds, and the experimental results are shown in Table 5.

[0231] Currently, industrial applications of multi-layer, multi-pass welding typically rely on a "teach and playback" model for setting welding paths. In this model, operators need to manually teach the welding path and record weld information, a process that takes approximately 3-5 minutes, and even longer for complex welds. In contrast, a robotic multi-layer, multi-pass welding path adaptive generation method based on 3D point clouds can automatically generate multi-layer, multi-pass welding paths through algorithms, significantly reducing manual intervention.

[0232] Efficiency analysis was performed on each step of the proposed algorithm, including point cloud preprocessing, weld segmentation, weld feature extraction, weld feature point interpolation, MLMP welding path generation, and normal vector estimation. The runtime of each step is recorded in Table 5. The results in Table 5 show that the computation time for point cloud segmentation is relatively long. This is mainly because the DC-RANSAC algorithm needs to iteratively calculate the distance from each point to the plane equation multiple times to determine whether the point belongs to the plane containing the weld. When the point cloud data volume is large, the computational burden of this process increases significantly, resulting in longer processing time.

[0233] Table 5 shows the running time of each part of the algorithm.

[0234]

[0235] Nevertheless, compared to the overall time consumption of the traditional "teach and playback" mode, the computation time of this algorithm is still significantly reduced, especially in terms of high efficiency in the automatic generation of welding paths and rapid planning of welding postures. Through automated point cloud segmentation processing, the shape and position of the weld can be more accurately identified, thereby generating the welding path precisely. This method not only greatly shortens the welding path teaching time but also significantly reduces labor costs, making it particularly suitable for mass production in multi-layer, multi-pass welding scenarios.

[0236] Overall, the adaptive generation method for multi-layer, multi-pass welding paths in robots based on 3D point clouds effectively solves the problem of cumbersome teaching in traditional methods and demonstrates a significant advantage in path generation efficiency. Experimental results show that the algorithm significantly improves production efficiency while ensuring the accuracy of welding path generation, and can well meet the efficiency requirements of industrial production for multi-layer, multi-pass welding processes, providing strong support for the intelligent and automated development of robotic welding technology.

[0237] Comparison with existing welding path generation methods

[0238] To further verify the advantages of the proposed method in multi-layer, multi-pass welding path generation, two existing welding path generation methods were selected for comparison. These two methods represent typical path generation strategies based on 3D reconstruction technology and 2D feature point extraction, respectively. Method 1 is a welding path generation method based on 3D point cloud reconstruction. It uses digital stripe projection technology to reconstruct the welding workpiece in 3D, obtains 3D point cloud data through the 3D structure of the workpiece, and then segments and processes the point cloud data using the RANSAC algorithm to extract weld features and generate the welding path. Method 2 is a welding path generation method based on 2D feature point extraction. It first extracts key feature points of the weld on a 2D plane using image processing technology, and then uses laser plane calibration technology to map the 2D feature points to 3D space to plan multi-layer, multi-pass welding paths. Since Method 2 is only applicable to path generation for V-groove welds, V-groove welds were selected as the comparative experimental object to ensure the comparability and fairness of the experiment. The error data of the welding paths generated by the three methods are shown in Table 6.

[0239] As can be seen from the data in Table 6, Method 2 outperforms Method 1 in terms of path generation accuracy, with maximum ME and RMSE values ​​of 1.092 mm and 0.667 mm, respectively, significantly lower than those of Method 1. This indicates that Method 2 has relatively higher accuracy and stability. However, the applicability of Method 2 is limited, primarily designed for V-groove welds, and it is difficult to extend to other weld types commonly found in industrial environments, such as welded workpieces with complex shapes containing lap and butt welds. Therefore, the application scope of Method 2 is significantly restricted.

[0240] Table 6 Comparison of MLMP welding path generation experimental results

[0241]

[0242] The method proposed in this invention exhibits good accuracy, with a maximum ME and RMSE of 1.103 mm and 0.663 mm, respectively, which is essentially on par with the accuracy of Method 2. Furthermore, the average RMSE value of the proposed method in the XYZ directions is lower than that of Method 2, demonstrating stronger robustness. This means that the proposed method can effectively reduce the influence of noise points and outliers on the welding path generation process, significantly reducing the error between the actual generated path and the ideal welding position. In addition, the proposed method is not only applicable to V-groove welds but also to various weld types such as lap and butt joints, basically covering common weld forms in industrial applications, thus possessing wider applicability.

[0243] The experimental results above demonstrate that the proposed adaptive generation method for multi-layer, multi-pass welding paths based on 3D point clouds has significant advantages over existing methods in terms of accuracy and applicability. This method maintains high welding path accuracy while adapting to various weld shapes, exhibiting excellent robustness and broad application potential.

[0244] Based on the technical solutions provided in the above embodiments, a method and system for adaptive generation of multi-layer, multi-pass welding paths for robots based on 3D point clouds is proposed. This method acquires 3D point cloud data of the welding surface using a TOF sensor and combines it with the geometric feature information of the workpiece to automatically plan the number of layers and passes required for multi-layer, multi-pass welding, and then adaptively generates the welding path. This method not only improves the accuracy and efficiency of welding but also enhances the system's adaptability to different welding tasks. By adapting to various welding workpiece types, the system can achieve efficient and high-precision automated operation in medium-thick plate welding, greatly reducing the need for manual intervention, reducing errors and defects, and thus effectively improving welding efficiency and quality.

[0245] The specific advantages are as follows:

[0246] 1. This invention proposes a random sample consistent weld seam segmentation algorithm based on density clustering. This algorithm can not only handle simple planar structures, but also cope with complex curved surface structures, and can be effectively applied to the curved surface segmentation of medium and thick plate welded workpieces.

[0247] 2. This invention fully considers the diversity and complexity of welded workpieces in industrial production, and designs multi-layer, multi-pass welding paths for various common medium and thick plate welds. It is widely applicable to welded workpieces of various shapes and sizes in industrial production and can be efficiently applied to multi-layer, multi-pass welding technology.

[0248] 3. Experiments were conducted on various welded workpieces using this invention, and the maximum error (ME) and root mean square error (RMSE) of the generated multi-layer, multi-pass weld paths were controlled below 1.103 mm and 0.677 mm, respectively. Furthermore, the time required is significantly reduced compared to the traditional "teach and playback" mode. This achieves an efficient and stable weld path generation process.

[0249] In this document, the terms “comprising,” “including,” or any other variations thereof are intended to cover non-exclusive inclusion, such that a step or method that comprises a list of elements includes not only those elements but also other elements not expressly listed or inherent to such a step or method.

[0250] The above description, in conjunction with specific preferred embodiments, provides a further detailed explanation of the present invention. It should not be construed that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, various simple deductions or substitutions can be made without departing from the concept of the present invention, and all such modifications and substitutions should be considered within the scope of protection of the present invention.

Claims

1. A robot multi-layer multi-pass welding path adaptive generation method, characterized in that, The method comprises the following steps: The three-dimensional point cloud of the weld is collected by using a time flight sensor, and the collected weld point cloud is preprocessed; The preprocessed weld point cloud is segmented by using a random sample consensus weld segmentation algorithm based on density clustering to extract the features of the weld, and the cubic spline interpolation method is used to interpolate and fit the weld feature points; According to the geometric information of the welded workpiece and the interpolated and fitted weld feature points, a multi-layer multi-pass welding path is generated; Based on the multi-layer multi-pass welding path, the welding gun end posture is calculated by using a normal estimation method based on weighted principal component analysis; The method for generating the multi-layer multi-pass welding path is selected from at least one of the following methods (1) to (3): (1) For an angle weld, the multi-layer multi-pass welding path is generated, specifically comprising the following steps: According to the extracted weld features A, B and C, the expression of the welding layer number L' is: wherein z Ai , z Bi , z Ci respectively represent the depth coordinate value of the i-th weld feature point in weld features A, B, C, n represents the number of weld feature points, h is the predetermined weld pass thickness, and th is the set error threshold value; in order to make the weld pass cover the entire weld, the predetermined weld pass thickness h is adaptively adjusted to h', when the remaining depth is within the error allowable range, the weld pass layer number is rounded down, otherwise the weld pass layer number L' = [L] + 1, and the predetermined weld pass thickness is adjusted as follows: After the initial welding to obtain new weld features D and E, each layer starts welding from the right side of the weld with the first weld pass; the number of passes is determined by the groove width of each layer and the width of the pass, and the number of passes P' of each layer is determined by the following formula: m As shown below: wherein z Ai , z Bi , z Di , z Ei respectively represent the depth coordinate value of the i-th weld feature point in weld features A, B, D, E, y Ai , y Bi , y Di , y Ei respectively represent the width coordinate value of the i-th weld feature point in weld features A, B, D, E, w is a predetermined weld width, D m is the width of the m-th layer section, and m is the number of the layer currently welded. To ensure that the weld bead is precisely covered by the weld seam, the predetermined weld bead width w is adaptively adjusted to w′, when P m -[P m When ]≤th, th is the set error threshold, representing the remaining width th·w after rounding to the nearest whole number w; if the remaining width is within the allowable error range, then for P m Round down to the nearest integer to ensure the number of weld passes is an integer; if P m -[P m ]>th indicates that the remaining width is too large, requiring an additional weld pass, P′ m =[P m +1, and adjust the predetermined weld width as follows: Based on the number of welding layers and the number of welding beads of each layer, the coordinate positions of each bead number are further determined, and the welding bead positions on each section are respectively planned. The height between the adjacent two layers is the corrected welding bead thickness h', and the width between the adjacent two beads in the same layer is the corrected welding bead width w'. Therefore, the coordinate W of the jth bead of the ith layer is: ij W = (i-1)h' + jw' wherein respectively represent average width coordinate values of the weld features B, E, respectively represent average depth coordinate values of the weld features B, E; (2) For a lap weld formed by connecting two workpieces together, the multi-layer multi-pass welding path is generated, specifically comprising the following steps: According to the extracted weld feature A and the preset welding height H, the weld features B and C are obtained, and then the welding layer number is calculated by using the predetermined weld pass thickness h and the error range th, and the layer number L' is: The weld pass thickness h is adjusted to obtain h', and the welding pass number of each layer is determined by the cross-sectional width of the layer and the width of the weld pass, the cross-sectional width can be obtained according to the similarity ratio, and the welding pass number can be calculated according to the error range th: m is the current layer number of welding, D m is the cross-sectional width of the mth layer, and w is the welding width. The adjusted w′ is According to the layer number L' and the pass number P' of each layer m Further determine the starting coordinate position W of the jth pass of the ith layer ij : a mean width coordinate value representing a weld feature A, a mean depth coordinate value representing a weld feature A, for a butt weld, generating a multi-pass weld path, comprising the following steps: The welding layer number required by welding is calculated by using the extracted weld feature connecting lines A, B, C and D, the predetermined weld pass thickness h and the allowable error range th, and the layer number L' is: z Ai 、z Bi 、z Ci 、z Di respectively represent the depth coordinate value of the i-th weld feature point in weld features A, B, C, D, and n represents the number of weld feature points. The adaptive adjustment is performed on the weld thickness: After the number of welding layers is calculated, the starting and ending coordinates of the welding path need to be calculated, and then the welding is performed from the starting point to the ending point; the starting and ending coordinates and are: where y Cj , y Dj , respectively, represent the width coordinate values of the jth weld feature point in the weld features C, D, respectively, and z Cj , z Dj , respectively, represent the depth coordinate values of the jth weld feature point in the weld features C, D, respectively. (3) For vertical plate welding, the multi-layer multi-pass welding path is generated, specifically comprising the following steps: According to the extracted weld features A, B and C, the feature points are located in the XOY plane, and the CB direction is the positive direction of the X axis; The V-shaped groove weld included angle θ is: According to the known weld pass thickness h and the weld included angle θ, the welding pass number P can be obtained: y Ai , y Bi , y Ci respectively represent the width coordinate value of the ith weld feature point in weld features A, B, and C, n represents the number of weld feature points, th is a set error threshold value, and the adaptive adjustment of the weld thickness h is: The coordinates of the starting point, intermediate point and ending point of the welding path are calculated based on the number of welding passes, and then the welding is performed from the starting point to the intermediate point and then to the ending point, wherein the coordinates of the starting point, intermediate point and ending point are and ​ x Aj 、x Bj 、x Cj respectively represent the length coordinate value of the jth weld feature point in the weld feature A, B, C, respectively represent the average width coordinate value of the weld feature A, B, C.

2. The method of claim 1, wherein, The preprocessing of the collected weld point cloud comprises removing outliers by using a radius filtering method and reducing the number of point cloud data by using a point cloud downsampling method based on voxel filtering.

3. The method of claim 1, wherein, The random sample consensus weld segmentation algorithm based on density clustering comprises the following steps: Iterate the following steps until the maximum iteration number: Randomly initialize three non-collinear points from the point cloud set; Calculate the plane equation according to the initialized points; Calculate the distance of all points to the plane based on the plane equation, and obtain the number of all inliers less than the point-to-plane distance threshold; If the number of inliers is greater than the inlier number threshold, calculate the plane equation represented by each inlier as the input data of the extracted weld feature points, and remove each inlier from the point cloud set; After iteration, if there are remaining point clouds in the point cloud set, the following steps are performed on each remaining point cloud: Calculate the number of points in the circle with the point cloud as the center and the radius distance threshold as the radius; If the number of points in the circle is greater than the adjacent point number threshold, mark the point cloud as the center, and randomly select another point cloud in the circle until all the remaining point clouds are marked; Classify the point clouds according to the marking, and finally obtain the segmented point cloud result.

4. The method of claim 1, wherein, Before extracting the weld feature, the point cloud is first extracted in the region of interest, and the distance threshold of the point to the plane is set when the feature point is extracted.

5. The method of claim 1, wherein, The normal estimation method of weighted principal component analysis is used to calculate the pose of the welding torch tip, which specifically includes the following steps: For a point p i , a neighborhood point set S i around it is selected; Compute the Euclidean distance d of a point p in the neighborhood point set S to the neighborhood center point c i i i , d i is: d i = ||p i -c||;​​ The weight of different points is defined according to the Euclidean distance of the points in the neighborhood from the center, and the formula of the weight w is: wherein is the average distance between the neighborhood points and the center point, and tj is an artificially set distance threshold. Subtracting the coordinates of the center point c from the point coordinates in the neighborhood set S i to achieve mean centering of the coordinates, then minimizing its weighted covariance matrix: N represents the number of points within the neighborhood, w i is the weight of the i-th point, p i represents the spatial coordinates of the i-th point, c represents the spatial coordinates of the neighborhood center point, and n is the normal vector. By singular value decomposition of the weighted covariance matrix, the eigenvector corresponding to the smallest eigenvalue is obtained as the normal vector of the point p i .

6. A robotic multi-layer multi-pass welding path adaptive generation system, comprising: The system is used to realize the adaptive generation method of the multi-layer multi-pass welding path of the robot according to any one of claims 1 to 5, comprising: The weld point cloud acquisition module is used to acquire the three-dimensional point cloud of the weld by using a time flight sensor, and to preprocess the acquired weld point cloud; The weld feature point acquisition module is used to segment the preprocessed weld point cloud by using a random sample consensus weld segmentation algorithm based on density clustering to extract the features of the weld, and to use a cubic spline interpolation method to interpolate and fit the weld feature points; The multi-layer multi-pass welding path generation module is used to generate a multi-layer multi-pass welding path according to the geometric information of the welding workpiece and the interpolated and fitted weld feature points; The welding torch tip pose acquisition module is used to calculate the pose of the welding torch tip based on the multi-layer multi-pass welding path by using the normal estimation method of weighted principal component analysis.

Images