Self-adaptive welding method and device based on machine vision and self-adaptive welding system

Abstract

The invention relates to a self-adaptive welding method and device based on machine vision and a self-adaptive welding system. The welding method comprises the steps that a current welding seam image of a space curve welding seam of a welding workpiece is obtained, and the current welding seam image is preprocessed; acquiring a welding seam mathematical model of the space curve welding seam, and acquiring a current welding seam position of the space curve welding seam; according to the welding seam mathematical model of the space curve welding seam and the current welding seam position, a set welding path and set welding parameters of the space curve welding seam are obtained; and according to the set welding path and the set welding parameters of the space curve welding seam, the welding robot is controlled to weld the welding workpiece, and welding information of the space curve welding seam is monitored and adjusted in the welding process. The problems that when a traditional welding mode faces space curve welding seams, the welding quality is unstable, the welding precision is limited, the welding adaptability is insufficient, and complex industrial requirement scenes cannot be met can be solved.

Description

Technical Field

[0001] This invention relates to the field of welding technology, and in particular to an adaptive welding method, apparatus and adaptive welding system based on machine vision. Background Technology

[0002] In modern manufacturing, welding, as a key processing technology, is widely used in many fields such as machinery manufacturing, automobiles, shipbuilding, and aerospace. With the increasing complexity of product structures and the ever-increasing demands for welding quality, traditional welding technologies face numerous challenges. Taking boiler structures as an example, typical products are large in size, with diverse weld types, including numerous fillet welds, vertical fillet welds, and overhead welds, as well as spatially curved welds. Under these circumstances, achieving efficient and precise welding has become an urgent problem to be solved.

[0003] Traditional welding methods, such as manual welding, are not only inefficient when dealing with complex welds like spatial curved seams, but the weld quality is also greatly affected by the welder's skill level and working conditions. It is difficult to ensure a uniform and smooth weld, and visual defects such as arc shrinkage, porosity, undercut, and cracks are prone to occur. In addition, while ordinary welding robots can improve efficiency to some extent when dealing with complex welds like spatial curved seams, they can only weld along a predetermined trajectory and lack the ability to adapt to problems such as poor workpiece consistency, clamping deviations, or thermal deformation, thus failing to meet the mass production needs of complex products. Summary of the Invention

[0004] This invention provides an adaptive welding method, apparatus, and system based on machine vision, which can solve the problems of unstable welding quality, limited welding accuracy, and insufficient welding adaptability of traditional welding methods when facing complex welds such as spatial curved welds, thus failing to meet the complex industrial needs.

[0005] To address the aforementioned technical problems, this invention provides an adaptive welding method based on machine vision, applicable to an adaptive welding system; The adaptive welding system includes a multi-view image acquisition device and a welding robot connected to the multi-view image acquisition device. The welding method includes: The multi-view image acquisition device is controlled to acquire the current weld image of the spatial curve weld of the welded workpiece, and the current weld image of the spatial curve weld of the welded workpiece is preprocessed. The mathematical model of the spatial curve weld is obtained based on the preprocessed current weld image, and the current weld position of the spatial curve weld is obtained based on the weld mathematical model and the positional relationship between the multi-view image acquisition device and the welding robot. Based on the mathematical model of the space curve weld and the current weld position, obtain the set welding path and set welding parameters for the space curve weld; Based on the set welding path and parameters of the spatial curve weld, the welding robot is controlled to weld the workpiece, and the welding information of the spatial curve weld is monitored and adjusted during the welding process.

[0006] In addition, the present invention also proposes an adaptive welding device based on machine vision, which is applied to an adaptive welding system; The adaptive welding system includes a multi-view image acquisition device and a welding robot connected to the multi-view image acquisition device. The welding apparatus includes: The weld seam image acquisition module is used to control the multi-view image acquisition device to acquire the current weld seam image of the spatial curve weld seam of the welded workpiece, and to preprocess the current weld seam image of the spatial curve weld seam of the welded workpiece. The weld seam advance module is used to obtain the weld seam mathematical model of the spatial curve weld seam based on the preprocessed current weld seam image, and to obtain the current weld seam position of the spatial curve weld seam based on the weld seam mathematical model and the positional relationship between the multi-view image acquisition device and the welding robot. The welding parameter and path acquisition module is used to acquire the set welding path and set welding parameters of the space curve weld based on the weld mathematical model of the space curve weld and the current weld position. The welding execution and adjustment module is used to control the welding robot to weld the workpiece according to the set welding path and welding parameters of the spatial curve weld, and to monitor and adjust the welding information of the spatial curve weld during the welding process.

[0007] Furthermore, the present invention also proposes an adaptive welding system, comprising: Multi-view image acquisition device; A welding robot, connected to the multi-view image acquisition device; and The controller is connected to both the multi-view image acquisition device and the welding robot. The controller is used to implement the machine vision-based adaptive welding method described above.

[0008] Furthermore, the present invention also proposes a computer storage medium comprising: at least one instruction that, when executed by a computer, implements the method steps of the machine vision-based adaptive welding method as described above.

[0009] The beneficial effects of the technical solution provided by this invention include: A multi-view image acquisition device can acquire the current weld image of the spatial curve weld of the workpiece located at the welding station. By preprocessing and image processing the current weld image, the current weld position of the spatial curve weld can be accurately extracted and located. Furthermore, the mathematical model of the spatial curve weld can be acquired, thereby obtaining the set welding path and set welding parameters. Based on the set welding path and set welding parameters of the spatial curve weld, the welding robot can be controlled to weld the workpiece. Moreover, during the welding process, the real-time welding path and real-time welding parameters of the spatial curve weld are monitored. When there is a deviation between the real-time welding path and the set welding path, or a deviation between the real-time welding parameters and the set welding parameters, the real-time welding path and real-time welding parameters can be adjusted in a timely manner.

[0010] This invention employs a machine vision-based approach combined with a welding robot to achieve adaptive welding with spatial curve weld seam extraction and precise positioning. This improves weld seam positioning and welding accuracy, enhances the adaptability of the welding process, and ensures the stability of adaptive welding. Specifically, this solution utilizes multi-view image acquisition based on machine vision and advanced image processing algorithms to accurately extract and locate spatial curve weld seam features. This overcomes the limitations of traditional welding methods (where laser tracking-assisted welding technology suffers from limited weld seam positioning accuracy in complex environments and with workpieces made of special materials; and where fixed-program robot welding technology cannot adapt to manufacturing errors, clamping errors, and thermal deformation errors in the workpiece, making it difficult to guarantee welding accuracy) and achieves high-precision welding.

[0011] Moreover, compared to traditional fixed-program robot welding technology, which lacks flexibility and struggles to handle complex spatial curve welds and various workpiece deviations, this solution utilizes machine vision to acquire weld information in real time. This allows the welding robot to adaptively adjust the welding path and parameters based on the actual weld conditions, adapting to spatial curve welds of different shapes and sizes, as well as manufacturing errors, clamping errors, and thermal deformation errors in the welded workpiece, thus improving the adaptability of the welding process. Furthermore, through precise weld positioning, adaptive welding path, and welding parameter adjustment, this solution can call appropriate welding parameters and plan trajectories based on the workpiece conditions perceived by vision, ensuring a stable welding process, reducing welding defects, guaranteeing the uniformity of the overall weld quality, and achieving a stable improvement in welding quality. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 This is a schematic diagram of the steps of the adaptive welding method based on machine vision according to an embodiment of the present invention; Figure 2 This is a detailed flowchart illustrating step S200 of the machine vision-based adaptive welding method according to an embodiment of the present invention. Figure 3 This is a detailed flowchart illustrating step S400 of the machine vision-based adaptive welding method according to an embodiment of the present invention. Figure 4 This is a simplified structural block diagram of the machine vision-based adaptive welding device described in an embodiment of the present invention. Figure 5 This is a simplified block diagram illustrating the structure of the adaptive welding system described in an embodiment of the present invention. Detailed Implementation

[0014] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0015] While conventional robotic welding improves welding efficiency to some extent, it faces challenges when handling spatial curved welds. Because conventional robotic welding follows a fixed, pre-set trajectory, it cannot adapt to errors generated during workpiece manufacturing and clamping, or thermal deformation during welding. Furthermore, when workpiece deviations exist, conventional robotic welding struggles to accurately track the weld, leading to decreased welding precision and impacting product quality. Additionally, traditional weld tracking technologies (such as laser tracking) are susceptible to interference in complex welding environments. For example, strong arc light and spatter generated during welding can affect laser signal acquisition, resulting in inaccurate weld positioning and failing to meet the demands of high-precision welding. Moreover, laser tracking technology struggles to achieve precise weld tracking for workpieces with complex shapes and special surface materials. Moreover, traditional welding methods lack intelligent parameter adjustment; they cannot automatically optimize welding parameters based on real-time conditions for spatial curved welds of different shapes and positions, making it difficult to guarantee the uniformity of weld quality. To address these technical problems, this invention proposes an adaptive welding method, apparatus, and system based on machine vision.

[0016] Figure 1As shown, in some embodiments, the present invention proposes a machine vision-based adaptive welding method applied to an adaptive welding system 10. For example... Figure 5 As shown, the adaptive welding system 10 may include a multi-view image acquisition device 100 and a welding robot 200 connected to the multi-view image acquisition device 100. The multi-view image acquisition device 100 can acquire images of the spatial curve weld seam of the workpiece in real time, and the welding robot 200 can perform welding on the spatial curve weld seam of the workpiece.

[0017] Specifically, such as Figure 1 As shown, this machine vision-based adaptive welding method may include the following steps: S100: Control the multi-view image acquisition device 100 to acquire the current weld image of the spatial curve weld of the welded workpiece, and preprocess the current weld image of the spatial curve weld of the welded workpiece. S200: Obtain the mathematical model of the spatial curve weld based on the preprocessed current weld image, and obtain the current weld position of the spatial curve weld based on the weld mathematical model and the positional relationship between the multi-view image acquisition device 100 and the welding robot 200. S300. Based on the mathematical model of the space curve weld and the current weld position, obtain the set welding path and set welding parameters for the space curve weld. S400: Based on the set welding path and set welding parameters of the spatial curve weld, control the welding robot 200 to weld the workpiece, and monitor and adjust the welding information of the spatial curve weld during the welding process.

[0018] The multi-view image acquisition device 100 can acquire the current weld image of the spatial curve weld of the workpiece located at the welding station. By preprocessing and image processing the current weld image, the current weld position of the spatial curve weld can be accurately extracted and located. Furthermore, it can acquire the weld mathematical model of the spatial curve weld, thereby obtaining the set welding path and set welding parameters. Based on the set welding path and set welding parameters of the spatial curve weld, the welding robot 200 can be controlled to weld the workpiece. Moreover, during the welding process, the real-time welding path and real-time welding parameters of the spatial curve weld are monitored. When there is a deviation between the real-time welding path and the set welding path, or a deviation between the real-time welding parameters and the set welding parameters, the real-time welding path and real-time welding parameters can be adjusted in a timely manner.

[0019] This invention employs a solution combining machine vision and a welding robot 200 to achieve adaptive welding with spatial curve weld seam extraction and precise positioning. This improves weld seam positioning and welding accuracy, enhances the adaptability of the welding process, and ensures the stability of adaptive welding. Specifically, this invention uses multi-view image acquisition based on machine vision and advanced image processing algorithms to accurately extract and locate spatial curve weld seam features. This overcomes the impact of traditional welding methods (in traditional welding methods, laser tracking-assisted welding technology has limited weld seam positioning accuracy in complex environments and when dealing with welding workpieces of special materials; robot welding technology based on fixed programs cannot adapt to manufacturing errors, clamping errors, and thermal deformation errors of the welding workpiece, making it difficult to guarantee welding accuracy) on accuracy, thus achieving high-precision welding.

[0020] Moreover, compared to traditional fixed-program robot welding technology, which lacks flexibility and struggles to handle complex spatial curve welds and various workpiece deviations, this invention utilizes machine vision to acquire weld information in real time. This allows the welding robot 200 to adaptively adjust the welding path and parameters based on the actual weld conditions, adapting to spatial curve welds of different shapes and sizes, as well as manufacturing errors, clamping errors, and thermal deformation errors in the welded workpiece, thus improving the adaptability of the welding process. Furthermore, through precise weld positioning, adaptive welding path, and welding parameter adjustment, this invention can call appropriate welding parameters and plan trajectories based on the workpiece conditions perceived by vision, ensuring a stable welding process, reducing welding defects, guaranteeing the uniformity of the overall weld quality, and achieving a stable improvement in welding quality.

[0021] Furthermore, the multi-view image acquisition device 100 may include multiple industrial cameras 110 for placement around the welding work area, and an image acquisition card 120 connected to the multiple industrial cameras 110. Moreover, the industrial cameras 110 can be connected to the image acquisition card 120 via data cables. The industrial cameras 110 are used to capture images of the welded workpiece, and the image acquisition card 120 can acquire the captured images and transmit them to the controller 300 (such as a computer).

[0022] Therefore, in step S100, controlling the multi-view image acquisition device 100 to acquire the current weld image of the spatial curve weld of the welded workpiece can further include the following steps: S110, control multiple industrial cameras 110 of the multi-view image acquisition device 100 set around the welding station to take pictures of the spatial curve weld of the welding workpiece pre-placed at the welding station from different directions, and obtain the current weld image of the spatial curve weld.

[0023] A welding platform can be set up in the welding work area, and multiple industrial cameras 110 can be installed around the welding station of the welding platform at specific angles and positions. These industrial cameras 110 are connected to an image acquisition card 120 to form a multi-view image acquisition device 100. These industrial cameras 110 can be distributed in different positions to fully cover the surface of the workpiece placed at the welding station, ensuring that all parts of the spatial curve weld of the workpiece can be captured. For example, industrial cameras 110 can be installed at key positions such as above and to the side of the workpiece to avoid blind spots.

[0024] Once the workpiece is placed at the welding station on the welding stand, multiple industrial cameras 110 can be activated simultaneously to capture images of the workpiece from multiple angles and positions. The shooting frequency of the industrial cameras 110 can be adjusted according to the welding speed and the complexity of the spatial curve weld, ensuring that a sufficient number of clear images of the current weld are obtained to accurately reflect the shape and position information of the spatial curve weld.

[0025] Furthermore, in this embodiment, the industrial camera 110 can be configured as a 3D structured light industrial camera 110, and appropriate camera parameters, such as resolution, field of view, and frame rate, can be selected based on the size of the field of view of the welding station and the shape of the welded workpiece. Moreover, the industrial camera 110 can be equipped with a high-quality optical lens to ensure the clarity and accuracy of the captured images.

[0026] In addition, the industrial camera 110 can also be configured as a high-resolution special optical camera suitable for the welding environment, such as a camera with strong light and heat radiation resistance, to replace the ordinary industrial camera 110 for image acquisition. Whether it is a special optical camera or an ordinary industrial camera 110, the purpose is to acquire images that can reflect the characteristics of the weld, so as to obtain clear and accurate weld image information (current weld image) for subsequent processing and analysis.

[0027] S120, the image acquisition card 120 of the multi-view image acquisition device 100, acquires the current weld seam images of the spatial curve weld seam captured by multiple industrial cameras 110.

[0028] When acquiring current weld images of a spatial curved weld seam using multiple industrial cameras 110, the image acquisition card 120 can be used to capture these images, which are then transmitted to the controller 300 for processing and analysis. Furthermore, the multiple industrial cameras 110 can each capture current weld images at multiple locations along the spatial curved weld seam. The image acquisition card 120 can then capture these multiple images, which can be transmitted to the controller 300 for fusion to form a complete current weld seam image of the spatial curved weld seam.

[0029] By deploying one or more industrial cameras 110 around the welding work area, images of the welded workpiece can be acquired from different angles, overcoming the limitations of traditional single-view acquisition and comprehensively obtaining spatial curve weld information. Through camera layout design, shooting parameter settings, and image fusion processing, the shape and position data of the spatial curve weld on the welded workpiece can be captured completely and accurately, providing a foundation for subsequent precise processing.

[0030] Furthermore, in step S100, preprocessing the current weld image of the spatial curve weld of the welded workpiece may further include the following steps: S130. Based on the current weld image of the spatial curve weld of the welded workpiece, the salt-and-pepper noise of the current weld image is removed by the median filtering algorithm to obtain the first preprocessed weld image.

[0031] The image acquisition card 120 acquires and transmits current weld seam images of the spatial curve weld seam of the welded workpiece, captured by multiple industrial cameras 110, to the controller 300. The controller 300 can first perform denoising processing on the current weld seam image of the spatial curve weld seam, specifically by using a median filtering algorithm to remove salt-and-pepper noise from the current weld seam image. This algorithm replaces the current pixel value by calculating the median value within the pixel's neighborhood, effectively eliminating isolated noise points while preserving the image's edge and detail information.

[0032] S140. The first preprocessed weld image is contrast-enhanced using histogram equalization to obtain the second preprocessed weld image. After removing salt-and-pepper noise from the current weld image using a median filtering algorithm, histogram equalization can be used to enhance the contrast of the image (the first preprocessed weld image). This involves statistically analyzing and redistributing the grayscale values ​​of the first preprocessed weld image to make its grayscale range more uniform. This highlights the difference between the weld area and the background, making the weld features more prominent and facilitating subsequent extraction operations.

[0033] S150. The second preprocessed weld image is filtered using a Gaussian filtering algorithm to obtain the preprocessed current weld image.

[0034] After enhancing the image contrast of the current weld image using histogram equalization, the image (i.e., the second preprocessed weld image) can be further filtered. Specifically, a Gaussian filtering algorithm can be used to further smooth the second preprocessed weld image, removing high-frequency noise and making the image clearer and more stable, thus providing high-quality image data for subsequent feature extraction.

[0035] In addition to median filtering, histogram equalization, and Gaussian filtering, bilateral filtering can also be used in image preprocessing to better preserve edge information while removing noise. Wavelet transform can also be used for denoising and feature extraction, enabling analysis at different scales and more effectively extracting detailed image features. All these filtering and transform methods can improve image quality, highlight weld features, and reduce the impact of noise and interference.

[0036] In summary, multi-view image acquisition in step S100 is fundamental to obtaining comprehensive weld information. Strategically deploying multiple industrial cameras 110 around the welding area to simultaneously capture images of the welded workpiece from different angles avoids missing weld information and fully captures the complex shape of the spatial curve weld. Image preprocessing is a crucial step in improving image quality and highlighting weld features. Median filtering removes salt-and-pepper noise while preserving weld edge details; histogram equalization enhances image contrast, making the weld more distinct from the background; and Gaussian filtering further smooths the image, reducing high-frequency noise interference and providing clear and stable image data for subsequent weld extraction and analysis.

[0037] In addition, such as Figure 2 As shown, in step S200, obtaining the weld mathematical model of the space curve weld based on the preprocessed current weld image may further include the following steps: S210. Based on the preprocessed current weld image, an edge detection algorithm is used to perform edge detection on the preprocessed current weld image to extract the preliminary weld edge image of the space curve weld.

[0038] Edge detection algorithms are used to perform edge detection on the preprocessed weld seam image. For example, the Canny algorithm can be used to perform edge detection on the preprocessed image. Specifically, when processing the preprocessed image using the Canny algorithm, the image can first be smoothed by Gaussian filtering, then the image gradient can be calculated, and then non-maximum suppression and double threshold detection can be used to accurately extract the preliminary weld seam edge image of the space curve weld.

[0039] S220. Combine morphological operations to process the preliminary weld edge image of the spatial curve weld to obtain the secondary weld edge image of the spatial curve weld.

[0040] After extracting the preliminary weld edge image of the spatial curve weld, morphological operations can be used to process the preliminary weld edge image. Specifically, dilation operations can be used to enlarge the weld edge in the preliminary weld edge image and connect any broken edges in the weld edge; then, erosion operations can be used to remove the excess weld edge caused by dilation in the preliminary weld edge image and refine the weld edge to obtain a more accurate weld edge image (i.e., a secondary weld edge image).

[0041] S230. The secondary weld edge image is processed using color space conversion and texture analysis algorithms to locate and extract the weld region and its weld edge points in the secondary weld edge image.

[0042] After obtaining the precise secondary weld edge image of the spatial curve weld, the weld region in the secondary weld edge image can be further identified using color and texture features (i.e., color space conversion and texture analysis algorithms). Specifically, based on the differences in color and texture between the weld region and other parts of the welded workpiece in the secondary weld edge image, the weld region in the secondary weld edge image can be accurately located and extracted using color space conversion and texture analysis algorithms, such as gray-level co-occurrence matrix, to obtain the final weld edge image of the spatial curve weld. Based on the obtained final weld edge image, multiple weld edge points of the weld region in the image can be obtained.

[0043] S240. The extracted weld edge points are fitted with a spline curve fitting method to construct a spline curve model of the spatial curve weld and obtain the mathematical model of the spatial curve weld.

[0044] After extracting the weld edge points from the weld edge image, curve fitting can be performed on the extracted weld edge points. A spline curve fitting method can be used to construct a suitable spline curve model based on the distribution characteristics of the weld edge points in the weld area, accurately describing the weld shape and position of the spatial curve weld, thus obtaining the mathematical model of the spatial curve weld.

[0045] In addition to the Canny algorithm, other edge detection algorithms such as Sobel and Prewitt can be used in weld edge detection. These are gradient-based edge detection algorithms that determine the weld edge by calculating the gradient of pixels in the image. Furthermore, for establishing the mathematical model of the weld, besides spline curve fitting, polynomial fitting, Bézier curve fitting, and other methods can be used, selecting the appropriate fitting method based on the shape characteristics of the spatial curve weld.

[0046] In summary, after acquiring the original image of the spatial curve weld of the welded workpiece (i.e., the current weld image), preprocessing operations such as denoising, contrast enhancement, and filtering can be performed on the image. Then, algorithms such as edge detection, morphological operations, and color and texture feature analysis are combined to extract weld feature information (weld edge points, weld shape, and weld position) from the image, constructing a mathematical model of the spatial curve weld. This ensures accurate extraction of weld feature information from the complex background of the image, achieving a precise digital description of the spatial curve weld.

[0047] Furthermore, in step 200, based on the weld mathematical model and the positional relationship between the multi-view image acquisition device 100 and the welding robot 200, the current weld position of the spatial curve weld is obtained, which may further include the following steps: S250, the camera coordinate system of the industrial camera 110 of the multi-view image acquisition device 100 and the working coordinate system of the welding robot 200.

[0048] S260. Using camera calibration technology, establish the coordinate transformation relationship between the image coordinate system where the mathematical model of the spatial curve weld is located and the working coordinate system where the welding robot 200 is located. S270. Based on the coordinate transformation relationship, map the weld point coordinates of the weld mathematical model to the working coordinate system of the welding robot 200 to obtain the current weld position of the spatial curve weld in the working coordinate system.

[0049] In industrial vision systems, the image coordinate system needs to be coordinated with the camera coordinate system and the world coordinate system (i.e., the working coordinate system of the welding robot 200) to achieve a complete mapping from the image to the physical world. Through pre-calibrated camera work, a positional relationship model (i.e., coordinate transformation relationship) between the industrial camera 110 and the welding robot 200 is established. Specifically, the intrinsic and extrinsic parameter matrices of the industrial camera 110 can be obtained using feature points on a calibration plate, realizing the transformation from the image coordinate system to the working coordinate system of the welding robot 200. Then, the image coordinate information of the weld points in the extracted spatial curve weld mathematical model, located in the image coordinate system, can be accurately mapped to the robot's working coordinate system through the coordinate transformation model. This yields the weld coordinate information of the weld points in the working coordinate system, thus providing the current weld position of the spatial curve weld in the working coordinate system.

[0050] Furthermore, in coordinate transformation, besides traditional camera calibration methods, deep learning-based camera calibration methods can also be used. These methods train neural networks to learn the mapping relationship between image coordinates and world coordinates (also known as coordinate transformation relationships). Then, this mapping relationship can be used to obtain the current weld position of the spatial curve weld in the working coordinate system from the image coordinate information of the weld points in the extracted mathematical model of the weld in the image coordinate system.

[0051] In summary, the above steps are all aimed at accurately extracting the weld features of the spatial curve weld and converting the position information of the spatial curve weld in the image into coordinate information that the welding robot 200 can understand and manipulate, so as to achieve precise positioning of the spatial curve weld. By employing different edge detection algorithms, fitting methods, and coordinate transformation methods, these tasks can be completed more accurately and efficiently, adapting to different weld shapes and welding environments.

[0052] Furthermore, in step S270, obtaining the current weld position of the spatial curve weld in the working coordinate system may further include the following steps: S272. Based on the manufacturing error, clamping deviation, and camera measurement error of the welded workpiece, establish a weld positioning error compensation model. S274. Based on the weld positioning error compensation model, the current weld position of the spatial curve weld in the working coordinate system is corrected to obtain the corrected current weld position.

[0053] Considering factors such as manufacturing errors, clamping deviations, and camera measurement errors in the welded workpiece, error compensation can be applied to the weld positioning results of spatial curve welds. Specifically, a weld positioning error compensation model can be established through the measurement and analysis of a large number of samples. Based on different error sources and magnitudes, corresponding compensation coefficients and correction algorithms are introduced to correct the weld positioning coordinates (i.e., the weld coordinate information of the spatial curve weld in the working coordinate system), thereby improving the accuracy of weld positioning. This ensures that the welding robot 200 can overcome various error interferences and accurately align the weld during subsequent welding processes, thus improving welding precision.

[0054] In summary, in step S200, edge detection using the Canny algorithm accurately delineates the weld edge contour of the spatial curve weld. Combined with morphological dilation and erosion operations, the edges are refined and optimized, and weld edge features are precisely extracted. Simultaneously, analyzing color and texture features further distinguishes the weld area from other parts of the welded workpiece, improving the accuracy of weld area identification. Moreover, by fitting weld edge points with spline curves, a precise mathematical model of the weld is established, enabling a quantitative description of the shape and position of the spatial curve weld. Camera calibration establishes a transformation relationship between the image coordinate system and the working coordinate system of the welding robot 200, mapping the coordinates of the spatial curve weld in the image (current weld image) to the working space of the welding robot 200. Considering various error factors, a compensation model and correction algorithm are used to correct the positioning results of the spatial curve weld, ensuring that the welding robot 200 can accurately align with the spatial curve weld for subsequent welding.

[0055] Furthermore, in step S300, based on the mathematical model of the space curve weld and the current weld position, the set welding path and set welding parameters for the space curve weld are obtained, which may further include the following steps: S310. Based on the mathematical model of the space curve weld, obtain the current weld shape of the space curve weld.

[0056] In the above steps, the process of extracting weld feature information (weld edge points, weld shape, weld position) from the current weld image of the spatial curve weld to construct the weld mathematical model of the spatial curve weld, and the coordinate transformation relationship between the image coordinate system and the working coordinate system of the welding robot 200, can obtain the current weld shape and current weld position of the spatial curve weld in the working coordinate system.

[0057] S320. Based on the current weld shape of the spatial curve weld, combined with the kinematic model of the welding robot 200 and the welding process requirements, obtain the set welding parameters for the spatial curve weld.

[0058] S330. Based on the current weld position of the spatial curve weld, combined with the kinematic model of the welding robot 200 and the welding process requirements, the set welding path of the spatial curve weld is obtained by using a path planning method based on optimization algorithm.

[0059] Based on the precisely located weld position and shape, combined with the kinematic model of the welding robot 200 and the welding process requirements, the welding path (setting the welding path) and welding parameters of the welding robot 200 are planned. That is, the current weld shape and position of the spatial curved weld are matched with the welding process parameter library (i.e., welding process requirements) to calculate the set welding parameters (including setting the welding current, setting the welding voltage, setting the welding speed, etc.) suitable for the spatial curved weld, and the optimal welding path (i.e., setting the welding path) of the welding robot 200 is planned.

[0060] Specifically, precise welding parameters such as welding current, welding voltage, welding speed, wire feed speed, welding torch 220 oscillation amplitude, and frequency can be calculated using a machine learning-based parameter optimization algorithm, combined with geometric features such as the current weld shape and position of the spatial curve weld, and a support vector machine regression algorithm. Furthermore, path planning methods based on optimization algorithms, such as genetic algorithms, can be employed to find the optimal motion path (i.e., the set welding path) for the welding robot 200 while meeting welding quality requirements, thereby reducing the robot's motion time and energy consumption and improving welding efficiency.

[0061] In addition, in the path planning of the welding robot 200, besides the genetic algorithm, intelligent optimization algorithms such as ant colony optimization and particle swarm optimization can also be used, which can find the optimal or near-optimal path (i.e., set the welding path) in the complex search space.

[0062] In addition, the welding robot 200 may include a robot body 210, a welding power source 230, a wire feeding mechanism 240, and a welding torch 220 mounted on the robot body 210, which is electrically connected to the welding power source 230. The robot body 210 can be connected to the controller 300 via a control line, receiving motion commands from the controller 300 to control the movement of the robot body 210. The welding torch 220 is connected to the robot body 210 and follows the movement of the robot body 210 to perform welding operations. The welding torch 220 is connected to the welding power source 230 via a cable to obtain the electrical energy required for welding.

[0063] Furthermore, the robot body 210 has multiple degrees of freedom, enabling it to move flexibly within the workspace of the welding station to meet the needs of welding welds in different positions and shapes. The welding torch 220 is used to perform the welding operation and is connected to the welding power source 230, providing a stable welding arc according to the set welding parameters. The welding power source 230 provides the necessary electrical energy for the welding process and can precisely control parameters such as welding current and welding voltage. The wire feeding mechanism 240 can transport the welding wire during the welding process of the welding torch 220 and can precisely control the wire feeding speed.

[0064] Therefore, as Figure 3 As shown, in step S400, the welding robot 200 is controlled to weld the workpiece according to the set welding path and set welding parameters of the space curve weld. This may further include the following steps: S410. Based on the set welding path of the obtained spatial curve weld, control the robot body 210 to drive the welding torch 220 to move according to the set welding path. S420. During the movement of the welding torch 220, the welding torch 220 and the wire feeding mechanism 240 are controlled to weld the workpiece according to the set welding parameters of the space curve weld.

[0065] The welding robot 200 calculates the motion parameters of each joint of the robot body 210 based on the set welding path and set welding parameters of the obtained spatial curve weld, combined with the kinematic model of the welding robot 200, and controls the movement of the robot body 210 so that the welding torch 220 performs welding according to the planned path.

[0066] Furthermore, during the welding process, multiple welding robots 200 can work collaboratively to complete the welding task of the spatial curve weld. Based on the set welding path of the spatial curve weld on the workpiece, the welding paths of each of the multiple welding robots 200 can be planned to collaboratively complete the welding of the entire weld. During the welding process, the various welding robots 200 maintain real-time data interaction and coordination through the controller 300, and by rationally planning the robot's working area and motion trajectory, collisions and interference between robots are avoided.

[0067] Furthermore, in step S400, monitoring and adjusting the welding information of the spatial curve weld during the welding process may further include the following steps: S430. During the welding process of the workpiece, the real-time welding path, real-time welding parameters, and real-time welding environment information of the spatial curve weld are obtained through the sensor detection device 400 and the multi-view image acquisition device 100. S440. Based on the set welding path, set welding parameters, and real-time welding environment information of the spatial curve weld, the welding information is adjusted in combination with the real-time parameter changes of the spatial curve weld.

[0068] During the welding process, welding information (including real-time welding parameters and real-time welding path) and environmental information can be monitored in real time. Based on this data and preset control algorithms, the welding information can be adaptively adjusted to ensure the stability of the welding process and the consistency of welding quality.

[0069] Specifically, the sensor detection device 400 can monitor real-time welding parameters such as welding current, welding voltage, and welding speed. Furthermore, during the welding process, the multi-view image acquisition device 100 can acquire real-time images of the current welding portion of the spatial curve weld seam using the welding tool. The acquired real-time weld images are then analyzed using the aforementioned processing method to obtain weld information such as the real-time weld shape and position of the current welding portion of the spatial curve weld seam, and based on this, the real-time welding path for the current welding portion is determined. Moreover, the welding parameters and welding path can be automatically adjusted according to the actual weld conditions to achieve dynamic optimization of the welding process and ensure consistent weld quality. For example, during the welding process, the real-time welding path and parameters may change relative to the set welding path or parameters due to changes in the structural characteristics of the current weld portion or the influence of environmental characteristics, requiring adjustment of the original set welding path or parameters. Alternatively, during the welding process, the real-time welding path and parameters may shift relative to the set welding path or parameters for some reason, requiring adjustment of the real-time welding path and parameters.

[0070] Furthermore, deep learning algorithms can be used to learn and train on a large amount of weld seam image data to construct a deep neural network model specifically for weld seam recognition and localization. First, a rich variety of sample images of spatial curve welds are collected, including different types of welded workpieces, weld seam shapes, welding processes, and various possible interference situations. These sample images are labeled to clarify weld seam information such as weld seam location and characteristics. Then, the labeled image data is input into the deep neural network model for training; commonly used network structures include convolutional neural networks (CNNs) and their variants. During training, the deep neural network model can continuously adjust its parameters and learn the weld seam information of spatial curve welds. When actual welding operations are performed, the industrial camera 110 acquires weld seam images of the welded workpiece and inputs them into the trained deep neural network model. Through analysis and processing of the weld seam images, the model quickly and accurately identifies weld seam information such as weld seam location, weld seam shape, and weld seam direction, and converts this weld seam information into instructions that the welding robot 200 can understand, guiding the welding robot 200 to plan the welding path, generate welding parameters, and perform the welding operation. Meanwhile, during the welding process, real-time image data is continuously fed back to the deep neural network model. The deep neural network model dynamically adjusts the welding path and welding parameters based on the new welding information to adapt to changes in the weld.

[0071] Furthermore, the sensor detection device 400 includes a welding parameter sensor 410 and an environmental sensor 420. The welding parameter sensor 410 can be used to monitor the real-time welding parameters of the spatial curve weld, including real-time welding current, real-time welding voltage, real-time welding speed, real-time wire feed speed, etc. The environmental sensor 420 can be used to monitor the real-time welding environment information of the spatial curve weld, including real-time ambient temperature, real-time ambient humidity, real-time arc intensity, etc., providing a reference for the control and adjustment of the welding process.

[0072] In summary, the adaptive welding system operates continuously during the welding process, constantly acquiring images, processing data, controlling the movement of the welding robot 200, and adjusting welding parameters to achieve efficient and precise welding of spatial curved welds. The multi-view image acquisition device 100, the welding robot 200, and the sensor detection device 400 work closely together during welding. The multi-view image acquisition device 100 acquires and transmits weld images in real time, processes the weld images to obtain set welding information, and transmits it to the welding robot 200 to control its welding of the spatial curved weld. The multi-view image acquisition device 100 and the sensor detection device 400 monitor the real-time welding information and adjust the welding information of the welding robot 200 accordingly, ensuring smooth data interaction between the devices and guaranteeing the orderly progress of the entire welding process. Based on the real-time information monitoring feedback from the multi-view image acquisition device 100 and the sensor detection device 400, the motion trajectory (i.e., welding path) and welding parameters of the welding robot 200 can be continuously adjusted, forming a closed-loop control. By continuously monitoring and adjusting, the welding process is kept in optimal condition at all times, adapting to minor changes in the workpiece and fluctuations in the welding environment, thus ensuring the stability of welding quality.

[0073] In addition, such as Figure 4 As shown, in other embodiments, the present invention also proposes a machine vision-based adaptive welding device 1000, applied to an adaptive welding system 10. For example... Figure 5 As shown, the adaptive welding system includes a multi-view image acquisition device 100 and a welding robot 200 connected to the multi-view image acquisition device 100. The multi-view image acquisition device 100 can acquire images of the spatial curve weld seam of the workpiece in real time, and the welding robot 200 can perform welding on the spatial curve weld seam of the workpiece.

[0074] Specifically, such as Figure 4 As shown, the machine vision-based adaptive welding device 1000 may include: The weld seam image acquisition module 1100 is used to control the multi-view image acquisition device 100 to acquire the current weld seam image of the spatial curve weld seam of the welded workpiece, and to preprocess the current weld seam image of the spatial curve weld seam of the welded workpiece. The weld seam advance module 1200 is used to obtain the weld seam mathematical model of the spatial curve weld seam based on the preprocessed current weld seam image, and to obtain the current weld seam position of the spatial curve weld seam based on the weld seam mathematical model and the positional relationship between the multi-view image acquisition device 100 and the welding robot 200. The welding parameter and path acquisition module 1300 is used to acquire the set welding path and set welding parameters of the space curve weld based on the weld mathematical model of the space curve weld and the current weld position. The welding execution and adjustment module 1400 is used to control the welding robot 200 to weld the workpiece according to the set welding path and set welding parameters of the spatial curve weld, and to monitor and adjust the welding information of the spatial curve weld during the welding process.

[0075] The machine vision-based adaptive welding device 1000 described in this embodiment corresponds to the machine vision-based adaptive welding method described above. The functions of each module in the machine vision-based adaptive welding device 1000 in this embodiment are detailed in the corresponding method embodiments and will not be repeated here. Furthermore, the specific structure of the adaptive welding system 10 in this embodiment is also specifically described in the above embodiments and will not be repeated here. In addition, as... Figure 5 As shown, in some other embodiments, the present invention also proposes an adaptive welding system 10, including a multi-view image acquisition device 100, a welding robot 200 connected to the multi-view image acquisition device 100, and a controller 300 connected to both the multi-view image acquisition device 100 and the welding robot 200. The controller 300 can control the multi-view image acquisition device 100 to acquire weld images of the workpiece, process the weld images to obtain welding information such as welding parameters and welding paths, and control the welding robot 200 to weld the workpiece according to the welding parameters and welding paths. Furthermore, the controller can adjust and control the welding parameters and welding paths in real time during the welding process.

[0076] Furthermore, the controller 300 can be used to implement each step of the aforementioned machine vision-based adaptive welding method. The specific implementation method can be found in the details of the aforementioned machine vision-based adaptive welding method, and will not be repeated here. Moreover, the specific structure of the adaptive welding system 10 in this embodiment has also been specifically described in the aforementioned embodiments, and will not be repeated here.

[0077] The machine vision-based adaptive welding solution proposed in this invention exhibits significant advantages in welding accuracy, environmental adaptability, production efficiency, welding quality, and system intelligence, representing a major breakthrough in modern welding technology. It achieves high-precision welding through precise weld seam positioning and error compensation mechanisms; it demonstrates strong adaptability by adapting to different welding workpieces and working conditions and coping with complex welding environments; it achieves efficient production through optimized welding path planning and real-time parameter adjustment; and it realizes intelligent and automated welding through intelligent control and easy integration and expansion. Specifically, it is manifested in: By employing multi-view industrial cameras to acquire weld seam images and combining them with advanced image processing algorithms, such as the Canny algorithm, morphological operations, and color and texture feature analysis, the weld seam edge contour of spatial curve welds can be accurately extracted, and an accurate mathematical model of the weld seam can be obtained through spline curve fitting. In welding complex workpieces, even with weld seam shapes that are complex and varied, the positioning accuracy can be controlled within a very small range. Compared to traditional laser tracking technology, it is unaffected by the surface material characteristics of the welded workpiece and interference from ambient light, significantly improving the accuracy of weld seam positioning. Taking into full account manufacturing errors, clamping deviations, and camera measurement errors of the welded workpiece, the weld seam positioning results are corrected through a pre-established error compensation model and correction algorithm. This allows the welding robot to more accurately follow the weld seam trajectory during welding, reducing welding deviations, ensuring the dimensional and positional accuracy of the weld seam, effectively avoiding welding defects caused by error accumulation, and improving overall welding accuracy.

[0078] Regardless of variations in the material, shape, and size of the workpiece, this solution can quickly acquire weld information through machine vision and adjust the welding robot's path and parameters based on the actual conditions of the spatial curve weld. For different batches of workpieces, even with manufacturing errors, reprogramming is unnecessary; the welding robot can adaptively adjust the welding process, improving production flexibility and versatility. In harsh welding environments with strong arc light and high spatter, the machine vision solution, with its stable image acquisition and processing capabilities, can still accurately extract weld features from spatial curve welds. Unlike laser tracking technology, it is less affected by arc light reflection and spatter obstruction, ensuring the stability and continuity of the welding process in complex environments and broadening the application scenarios of welding technology.

[0079] By utilizing optimization algorithms such as genetic algorithms to plan the welding path of welding robots, the robots can move along the optimal path, reducing unnecessary movements and movement time, and improving welding efficiency. When welding large and complex structural components, this can significantly shorten the welding cycle and increase output per unit time. During the welding process, welding parameters such as welding current, welding voltage, and welding speed can be monitored in real time, and these parameters can be automatically adjusted according to the actual conditions of the spatial curve weld. This real-time adjustment mechanism avoids welding interruptions and rework caused by unsuitable welding parameters, ensuring the efficient execution of the welding process and further improving production efficiency.

[0080] Through precise weld seam positioning and adaptive parameter adjustment, the weld width, depth, and forming quality of the spatial curve weld remain consistent throughout the entire welding process. When welding spatial curve welds, regardless of changes in weld curvature, the system ensures stable weld quality, reduces welding defects, and improves product quality and reliability. The system records a large amount of welding data during the welding process, including image information, welding parameters, and the welding robot's motion trajectory. This data provides a basis for weld quality traceability, facilitating rapid identification of causes when quality problems occur and enabling corresponding improvement measures to further enhance weld quality.

[0081] The entire welding process is highly automated, with image acquisition, weld seam extraction, weld seam positioning, welding path planning, and welding parameter adjustment all completed automatically by the system. No manual intervention is required, reducing the impact of human factors on welding quality and also lowering the workload of operators. The various components of this solution, such as the multi-view image acquisition device, welding robot, and sensor detection device, possess excellent openness and compatibility, easily integrating with other automated production equipment to form a complete automated production line. Furthermore, with technological advancements, the system can be easily expanded and upgraded to meet ever-increasing production demands. Furthermore, in other embodiments, the present invention proposes a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement all or part of the method steps of the machine vision-based adaptive welding method as described above.

[0082] The present invention can implement all or part of the processes in the above methods, or it can be accomplished by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when the computer program is executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content contained in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0083] Based on the same inventive concept, embodiments of the present invention also provide an electronic device, including a memory and a processor, wherein the memory stores a computer program that runs on the processor, and the processor executes the computer program to implement all or part of the method steps described above.

[0084] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the computer device, connecting all parts of the computer device through various interfaces and lines.

[0085] Memory can be used to store computer programs and / or models. The processor performs various functions of the computer device by running or executing the computer programs and / or models stored in the memory, and by accessing data stored in the memory. Memory can primarily include a program storage area and a data storage area. The program storage area can store the operating system and at least one application program required for a function (e.g., sound playback, image playback, etc.); the data storage area can store data created based on the use of the mobile phone (e.g., audio data, video data, etc.). Furthermore, memory can include high-speed random access memory, and can also include non-volatile memory, such as hard disks, RAM, plug-in hard disks, SmartMedia Cards (SMC), Secure Digital (SD) cards, Flash Cards, at least one disk storage device, flash memory device, or other volatile solid-state storage devices.

[0086] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, servers, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage and optical storage) containing computer-usable program code.

[0087] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), servers, and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0088] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0089] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0090] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A machine vision-based adaptive welding method, applied to an adaptive welding system; The adaptive welding system includes a multi-view image acquisition device and a welding robot connected to the multi-view image acquisition device. Its features are, The welding method includes: The multi-view image acquisition device is controlled to acquire the current weld image of the spatial curve weld of the welded workpiece, and the current weld image of the spatial curve weld of the welded workpiece is preprocessed. The mathematical model of the spatial curve weld is obtained based on the preprocessed current weld image, and the current weld position of the spatial curve weld is obtained based on the weld mathematical model and the positional relationship between the multi-view image acquisition device and the welding robot. Based on the mathematical model of the space curve weld and the current weld position, obtain the set welding path and set welding parameters for the space curve weld; Based on the set welding path and parameters of the spatial curve weld, the welding robot is controlled to weld the workpiece, and the welding information of the spatial curve weld is monitored and adjusted during the welding process.

2. The adaptive welding method based on machine vision according to claim 1, characterized in that, The multi-view image acquisition device includes multiple industrial cameras for placement around the welding work area, and an image acquisition card connected to the multiple industrial cameras; The control multi-view image acquisition device acquires the current weld seam image of the spatial curve weld seam of the welded workpiece, including: Multiple industrial cameras of a multi-view image acquisition device set around the welding station are controlled to take pictures of the spatial curve weld of the workpiece pre-positioned at the welding station from different angles, and the current weld image of the spatial curve weld is obtained. The image acquisition card of the multi-view image acquisition device is used to acquire current weld seam images of spatial curve weld seams captured by multiple industrial cameras.

3. The adaptive welding method based on machine vision according to claim 2, characterized in that, The preprocessing of the current weld image of the spatial curve weld of the welded workpiece includes: Based on the current weld image of the spatial curve weld of the welded workpiece, the median filtering algorithm is used to remove the salt-and-pepper noise of the current weld image to obtain the first preprocessed weld image. The first preprocessed weld image was contrast-enhanced using histogram equalization to obtain the second preprocessed weld image. The Gaussian filtering algorithm is used to filter the second preprocessed weld image to obtain the preprocessed current weld image.

4. The adaptive welding method based on machine vision according to claim 1, characterized in that, The step of obtaining the mathematical model of the spatial curve weld based on the preprocessed current weld image includes: Based on the preprocessed current weld image, an edge detection algorithm is used to perform edge detection on the preprocessed current weld image to extract the preliminary weld edge image of the space curve weld. The preliminary weld edge image of the spatial curve weld is processed by combining morphological operations to obtain the secondary weld edge image of the spatial curve weld. Color space conversion and texture analysis algorithms are used to process the secondary weld edge image to locate and extract the weld region and its edge points. The spline curve fitting method is used to fit the extracted weld edge points to construct a spline curve model of the spatial curve weld, thus obtaining the mathematical model of the spatial curve weld.

5. The machine vision-based adaptive welding method according to claim 4, characterized in that, The step of obtaining the current weld position of the spatial curve weld based on the weld mathematical model and the positional relationship between the multi-view image acquisition device and the welding robot includes: To obtain the camera coordinate system of the industrial camera in the multi-view image acquisition device and the working coordinate system of the welding robot; The coordinate transformation relationship between the image coordinate system of the mathematical model of the spatial curve weld and the working coordinate system of the welding robot is established using camera calibration technology. Based on the coordinate transformation relationship, the coordinates of the weld point in the mathematical model of the weld are mapped to the working coordinate system of the welding robot to obtain the current weld position of the spatial curve weld in the working coordinate system.

6. The machine vision-based adaptive welding method according to claim 5, characterized in that, The process of obtaining the current weld position of the spatial curve weld in the working coordinate system includes: A weld positioning error compensation model is established based on the manufacturing error, clamping deviation, and camera measurement error of the welded workpiece. Based on the weld positioning error compensation model, the current weld position of the spatial curve weld in the working coordinate system is corrected to obtain the corrected current weld position.

7. The adaptive welding method based on machine vision according to claim 1, characterized in that, The step of obtaining the set welding path and set welding parameters for the space curve weld based on the weld mathematical model and the current weld position includes: Based on the mathematical model of the space curve weld, obtain the current weld shape of the space curve weld; Based on the current weld shape of the spatial curve weld, combined with the kinematic model of the welding robot and the welding process requirements, the set welding parameters for the spatial curve weld are obtained. Based on the current weld position of the spatial curve weld, combined with the kinematic model of the welding robot and the welding process requirements, a path planning method based on optimization algorithms is used to obtain the set welding path for the spatial curve weld.

8. The adaptive welding method based on machine vision according to claim 1, characterized in that, The welding robot includes a robot body, a welding power source, a wire feeding mechanism, and a welding torch mounted on the robot body, wherein the welding torch is electrically connected to the welding power source. The process of controlling a welding robot to weld workpieces based on a set welding path and parameters for a spatial curve weld seam includes: Based on the set welding path of the obtained spatial curve weld, the robot body is controlled to drive the welding torch to move along the set welding path. During the movement of the welding torch, the welding torch and wire feeding mechanism are controlled to weld the workpiece according to the set welding parameters of the spatial curve weld.

9. The machine vision-based adaptive welding method according to claim 8, characterized in that, The monitoring and adjustment of welding information for spatial curve welds during the welding process includes: During the welding process, the real-time welding path, real-time welding parameters, and real-time welding environment information of the spatial curve weld are obtained through sensor detection devices and multi-view image acquisition devices. The welding information is adjusted based on the set welding path, set welding parameters, and real-time welding environment information of the spatial curve weld, combined with the real-time parameter changes of the spatial curve weld.

10. The machine vision-based adaptive welding method according to claim 9, characterized in that, The sensor detection device includes a welding parameter sensor and an environmental sensor; The welding parameter sensor is used to monitor the real-time welding parameters of the spatial curve weld, including real-time welding current, real-time welding voltage, and real-time welding speed. The environmental sensor is used to monitor the real-time welding environment information of the spatial curve weld seam, including real-time ambient temperature, real-time ambient humidity, and real-time arc intensity.

11. An adaptive welding device based on machine vision, applied to an adaptive welding system; The adaptive welding system includes a multi-view image acquisition device and a welding robot connected to the multi-view image acquisition device. Its features are, The welding apparatus includes: The weld seam image acquisition module is used to control the multi-view image acquisition device to acquire the current weld seam image of the spatial curve weld seam of the welded workpiece, and to preprocess the current weld seam image of the spatial curve weld seam of the welded workpiece. The weld seam advance module is used to obtain the weld seam mathematical model of the spatial curve weld seam based on the preprocessed current weld seam image, and to obtain the current weld seam position of the spatial curve weld seam based on the weld seam mathematical model and the positional relationship between the multi-view image acquisition device and the welding robot. The welding parameter and path acquisition module is used to acquire the set welding path and set welding parameters of the space curve weld based on the weld mathematical model of the space curve weld and the current weld position. The welding execution and adjustment module is used to control the welding robot to weld the workpiece according to the set welding path and welding parameters of the spatial curve weld, and to monitor and adjust the welding information of the spatial curve weld during the welding process.

12. An adaptive welding system, characterized in that, include: Multi-view image acquisition device; A welding robot is connected to the multi-view image acquisition device; as well as The controller is connected to both the multi-view image acquisition device and the welding robot. The controller is used to implement the machine vision-based adaptive welding method as described in any one of claims 1-10.

13. A computer storage medium, characterized in that, The computer storage medium includes at least one instruction that, when executed by a computer, implements the method steps of the machine vision-based adaptive welding method as described in any one of claims 1-10.

Images