Intelligent control method and system for laser welding equipment

Abstract

The invention discloses an intelligent control method and system for laser welding equipment. The method comprises the steps that real-time image data of a welding area are collected, and environment variable data are obtained; according to the real-time image data, image smoothing processing, noise suppression, filtering processing and denoising calculation are carried out, and clear contour data of the welding area are determined; according to the clear contour data, interference residual features are extracted, and integrated analysis is carried out in combination with the environment variable data to obtain the influence degree of interference on path planning; and according to the influence degree of the interference on path planning, dynamic parameter updating, interference compensation calculation, path deviation correction and welding track optimization are conducted, optimized track data are obtained, and a final welding operation sequence is determined according to the optimized track data. Accurate positioning of the laser welding equipment and dynamic optimization of the welding track can be achieved, and environmental interference is effectively compensated.

Description

Technical Field

[0001] This invention relates to the field of laser welding control technology, and in particular to an intelligent control method and system for laser welding equipment. Background Technology

[0002] Currently, laser welding technology has become a key process in modern manufacturing industries such as automotive and aerospace due to its high precision and efficiency. Welding quality directly affects product safety and reliability. Therefore, achieving intelligent control of the welding process, especially improving the environmental adaptability and accuracy of intelligent welding systems, remains a core objective pursued by the industry.

[0003] In existing technologies, many welding control methods rely heavily on preset process parameters and fixed control rules to guide the operation of welding equipment. However, in the complex actual welding environment, there are various interference factors such as light reflection and uneven material surfaces. These interferences can lead to a decrease in the accuracy of image feature recognition in the welding area, thus affecting subsequent path planning. More importantly, the intensity and characteristics of these interferences fluctuate in real time with the environment.

[0004] Therefore, existing technologies suffer from inaccurate identification of welding area features, which in turn affects the accuracy of path planning and control. Summary of the Invention

[0005] This invention provides an intelligent control method and system for laser welding equipment to solve the technical problem of inaccurate identification of welding area features, which in turn affects path planning and control accuracy.

[0006] In a first aspect, to solve the above-mentioned technical problems, the present invention provides an intelligent control method for laser welding equipment, comprising: Real-time image data of the welding area is acquired, and the real-time image data is smoothed to obtain a first filtered image; Based on the first filtered image, noise suppression and filtering are performed to obtain the second filtered image; For the second filtered image, denoising calculations are performed to determine the clear outline data of the welding area; From the clear contour data, residual interference features are extracted. If the residual interference features exceed the first interference threshold, the distribution characteristics of the residual interference features are analyzed to obtain environmental variable data. The distribution characteristics and the environmental variable data are then integrated and analyzed to obtain the degree of interference on path planning. Based on the degree of impact of the interference on path planning, dynamic parameter updates and interference compensation calculations are performed to obtain compensation data; Based on the compensation data, path deviation correction, welding trajectory optimization, and path planning scheme update are performed to obtain optimized trajectory data; Based on the optimized trajectory data, the drive control system performs precise positioning of the laser welding equipment and determines the final welding operation sequence.

[0007] In a second aspect, the present invention provides an intelligent control system for laser welding equipment, comprising: The image processing module is used to acquire real-time image data of the welding area and perform image smoothing processing on the real-time image data to obtain a first filtered image; The image filtering module is used to perform noise suppression and filtering processing on the first filtered image to obtain a second filtered image; The contour determination module is used to perform noise reduction calculations on the second filtered image to determine clear contour data of the welding area; The interference analysis module is used to extract residual interference features from the clear contour data. If the residual interference features exceed a first interference threshold, the distribution characteristics of the residual interference features are analyzed to obtain environmental variable data. The distribution characteristics and the environmental variable data are then integrated and analyzed to obtain the degree of interference on path planning. The compensation calculation module is used to perform dynamic parameter updates and interference compensation calculations based on the degree of impact of the interference on path planning, and obtain compensation data. The path optimization module is used to perform path deviation correction, welding trajectory optimization, and path planning scheme update on the compensation data to obtain optimized trajectory data. The execution control module is used to drive the control system to perform precise positioning of the laser welding equipment based on the optimized trajectory data, and to determine the final welding operation sequence.

[0008] Compared with the prior art, the present invention has the following beneficial effects: (1) This invention collects real-time image data of the welding area and performs image smoothing, noise suppression and filtering, and noise reduction calculation in sequence. This multi-level image processing technology can address and weaken image noise caused by factors such as light reflection and material inhomogeneity layer by layer. This process ultimately determines the clear contour data of the welding area, providing a high-quality and stable data foundation for subsequent interference analysis and accurate path planning, and solving the problem of decreased feature recognition accuracy due to interference in the prior art.

[0009] (2) This invention extracts residual interference features from contour data. When the features exceed the first interference threshold, it not only analyzes the distribution characteristics of the interference, but also acquires and integrates environmental variable data to conduct integrated analysis to obtain the degree of interference on path planning. This analysis method that integrates interference features with environmental variables enables it to get rid of the limitations of existing technologies that rely on preset parameters, and performs dynamic parameter updates and interference compensation calculations accordingly, thus realizing adaptive compensation for dynamic interference in complex environments.

[0010] (3) Based on the compensation data obtained by dynamic calculation, the present invention corrects and optimizes the path deviation of the welding trajectory, and drives the control system according to the optimized trajectory data, and performs precise positioning in combination with the equipment positioning accuracy improvement technology. This process ensures that the final welding path planning scheme can adapt to the environmental conditions in real time, significantly reduces the path deviation, improves the welding accuracy and efficiency, and can complete the high-quality welding task of specific micro parts. Attached Figure Description

[0011] Figure 1 This is a schematic diagram of the intelligent control method for laser welding equipment provided in the first embodiment of the present invention; Figure 2 This is a schematic diagram of the intelligent control system structure of a laser welding equipment provided in the second embodiment of the present invention. Detailed Implementation

[0012] 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, and 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.

[0013] Reference Figure 1 The first embodiment of the present invention provides an intelligent control method for laser welding equipment, comprising the following steps: S11, acquire real-time image data of the welding area, and perform image smoothing processing on the real-time image data to obtain a first filtered image; S12, based on the first filtered image, perform noise suppression and filtering processing to obtain the second filtered image; S13, For the second filtered image, perform noise reduction calculation to determine the clear outline data of the welding area; S14. Extract residual interference features from the clear contour data. If the residual interference features exceed the first interference threshold, analyze the distribution characteristics of the residual interference features, obtain environmental variable data, and integrate and analyze the distribution characteristics and the environmental variable data to obtain the degree of interference on path planning. S15, Based on the degree of impact of the interference on path planning, perform dynamic parameter updates and interference compensation calculations to obtain compensation data; S16, For the compensation data, perform path deviation correction, welding trajectory optimization and path planning scheme update to obtain optimized trajectory data; S17, Based on the optimized trajectory data, drive the control system to perform precise positioning of the laser welding equipment and determine the final welding operation sequence.

[0014] In step S11, real-time image data of the welding area is acquired, and image smoothing processing is performed on the real-time image data to obtain a first filtered image, including: Real-time image data of the welding area is acquired to obtain the original image sequence; The original image sequence is subjected to noise reduction processing to obtain the first filtered image.

[0015] It should be noted that real-time image data of the welding area is acquired through an industrial camera deployed above the laser welding equipment. This industrial camera continuously captures or scans the welding area at a preset frame rate. This frame rate is set to ensure that the fastest dynamic changes during the welding process can be captured, especially light reflection and material inhomogeneity interference.

[0016] For example, in a welding task, an industrial camera is set to an acquisition rate of 100 frames per second and a resolution of 1280x720 pixels. The camera continuously captures images containing welding sparks, molten pool reflections, and material surface shadows. This set of continuously acquired, unprocessed image frames constitutes the raw image sequence.

[0017] It should be noted that noise reduction processing on the original image sequence aims to reduce random high-frequency noise in the image caused by sensor thermal noise or environmental electromagnetic interference. In this embodiment, this processing is implemented using a Gaussian filtering algorithm. The Gaussian filtering algorithm performs a convolution operation on each frame of the original image sequence using a filter kernel of a preset size, replacing the value of each pixel with the weighted average of the pixel values ​​in its neighborhood, with the weights determined by a Gaussian distribution. This operation can effectively smooth noise points while preserving key edge information of the welding area.

[0018] It is worth noting that the determination of the filter kernel size (e.g., 5x5) of the Gaussian filtering algorithm is based on signal-to-noise ratio (SNR) analysis of historical welding image datasets. Sample images under different welding process parameters are collected and processed using filter kernels of different sizes (e.g., 3x3, 5x5, 7x7). The SNR of the processed image and the mean square error compared to the manually labeled "real" contour are calculated. Finally, the kernel size that achieves the optimal balance between maximizing the SNR (i.e., the best denoising effect) and minimizing the mean square error (i.e., the best edge preservation) is selected as the preset value.

[0019] For example, an image frame in an original image sequence has obvious random noise at the edges of its melt pool. A 5x5 filter kernel is used to perform Gaussian filtering on this image frame. After processing, isolated noise points in the original image are effectively smoothed, and the overall visual quality of the image is significantly improved. This image, after preliminary denoising and smoothing, is the first filtered image.

[0020] It should be noted that all parameters in this invention based on historical data statistical analysis (such as filter kernel size, variance threshold, strengthening threshold, etc.) were determined by collecting a large amount of typical welding image data. The data sources include images under different lighting conditions, material surfaces, and welding processes, with a sample size of no less than 1000 groups. Standard methods were used for parameter optimization: signal-to-noise ratio (SNR) analysis was used for filter parameters, ROC curve analysis was used for threshold differentiation, and regression analysis was used for the environment mapping model. Those skilled in the art can adjust the data collection scope and optimization methods according to the actual scenario.

[0021] In step S12, noise suppression and filtering are performed on the first filtered image to obtain a second filtered image, including: The first filtered image is subjected to local noise suppression processing, and the intensity control parameters are adjusted to obtain intermediate image data; For the intermediate image data, edge feature information is extracted, and the edge feature information is enhanced to obtain a transition image with enhanced edges; Based on the brightness distribution characteristics of the edge-enhanced transition image, contrast equalization adjustment is performed to obtain the second filtered image.

[0022] It should be noted that the local noise suppression processing of the first filtered image employs an adaptive smoothing method. This method queries a preset mapping function based on the local neighborhood statistical characteristics (such as local variance) of each pixel in the first filtered image to determine a specific intensity control parameter for that pixel. This mapping function ensures that a larger smoothing weight is assigned to regions with gentle pixel value changes (small variance) to suppress noise; while a smaller smoothing weight is assigned to regions with strong light reflection or rich texture details (large variance) to protect edge and texture features.

[0023] It is worth noting that the intensity control parameters are determined based on a pre-established mapping function. First, historical welding images containing various lighting interferences and material textures are acquired and used as a training set. Then, for the images in the training set, the correspondence between different local variance values ​​and the optimal smoothing intensity (i.e., maximizing the signal-to-noise ratio while preserving the edges) is tested experimentally. Finally, these correspondences are fitted by a function (such as multinomial regression) to obtain a mapping function that can output the optimal intensity control parameters based on the real-time local variance values, and this function is then embedded in the system.

[0024] For example, in a first filtered image, a certain region has pixel grayscale values ​​concentrated between 220-255 due to strong light reflection, resulting in extremely high local variance. The adaptive smoothing method applies a very weak intensity control parameter (almost no processing) to this region based on the mapping function to avoid loss of detail. In another noisier background region, the algorithm applies a stronger parameter to smooth the noise. The processed image is the intermediate image data.

[0025] It should be noted that the extraction of edge feature information from the intermediate image data involves processing the intermediate image data using the Canny operator to calculate the gradient magnitude and direction of the image, generating a preliminary edge map. If the detected edge features are blurred, enhancement processing is initiated. This enhancement processing uses the Laplacian operator to perform a convolution operation on the edge regions in the edge feature information; the calculation formula is as follows: in These are the sharpened pixel values. The pixel value within the neighborhood. Here, k represents the weighting coefficients of the Laplace operator kernel, and k is the neighborhood radius determined by the kernel size (e.g., 3x3). This operation increases the contrast between edge pixels and surrounding pixels, making boundary details stand out more.

[0026] It is worth noting that the preset weighting coefficients of the Laplace operator kernel It is predetermined according to the mathematical definition of the Laplace operator. For example, a 3x3 4-neighborhood operator kernel, its... It can be set to a weight matrix with a center point of -4, neighboring points (top, bottom, left, and right) of 1, and all other points of 0. This operation improves the contrast between edge pixels and surrounding pixels, making boundary details more prominent; the Laplacian operator kernel can be selected with 4-neighborhood or 8-neighborhood standard weights as needed.

[0027] For example, in a welding task, the intermediate image data suffers from discontinuous boundary lines due to the roughness of the material surface. First, blurred edge feature information is extracted using the Canny operator, and then a 3x3 Laplacian sharpening kernel is applied to enhance these edge regions. After processing, the grayscale transitions at the boundaries are more pronounced, and the discontinuous lines are strengthened, resulting in the enhanced transition image.

[0028] It should be noted that contrast equalization adjustment is performed on the transition image after edge enhancement to further improve the overall dynamic range of the image and make the texture features of the welding area clearer. This embodiment employs a contrast-limited adaptive histogram equalization (CLAHE) algorithm. This algorithm divides the transition image into several small rectangular sub-blocks, calculates its histogram independently for each sub-block, limits the amplification of contrast based on a preset "shearing threshold," and finally smooths the boundaries between sub-blocks through bilinear interpolation, thereby effectively suppressing noise amplification while enhancing local contrast.

[0029] It is worth noting that the determination of the shearing threshold in the CLAHE algorithm is based on contrast analysis of a historical welding image dataset. A large number of welding images under different lighting conditions are collected, and CLAHE processing is performed using different shearing thresholds (e.g., from 1.0 to 4.0). The processing results are then blindly scored using an image quality evaluation function (such as BRISQUE or NIQE). Finally, the threshold that yields the highest average quality score for the image set (e.g., 2.5) is selected as the preset value.

[0030] For example, in a transitional image after edge enhancement, the grayscale values ​​are mainly concentrated in a narrow range of 100-180, resulting in invisible texture details. By applying the CLAHE algorithm (with a clipping threshold of 2.5 and a sub-block size of 8x8) to this transitional image, the grayscale values ​​are redistributed to the full range of 0-255, making texture features and boundary details clearly visible. This contrast-enhanced grayscale image is the second filtered image.

[0031] In step S13, denoising calculations are performed on the second filtered image to determine the clear contour data of the welding area, including: The second filtered image is subjected to local smoothing and range adjustment processing to obtain a smooth transition image; For the smooth transition image, boundary region detail extraction and local enhancement are performed to obtain an enhanced boundary transition image; Based on the brightness distribution characteristics of the enhanced boundary transition image, contrast adjustment and interference region optimization are performed to obtain an optimized comparison image. Based on the optimized comparison image, the contour of the welding area is finely marked to determine the clear contour data.

[0032] It should be noted that the local smoothing and range adjustment processing of the second filtered image is performed using an adaptive window smoothing method. This method iterates through each pixel in the second filtered image, calculates the pixel distribution characteristics (such as variance) of its local neighborhood, and dynamically adjusts the range of pixel values ​​(i.e., window size) used for smoothing calculation based on these characteristics.

[0033] It is worth noting that the window size adjustment logic is determined based on a preset variance threshold. This variance threshold is set by statistically analyzing the variance values ​​of "flat regions" and "edge regions" in a large number of historical welding images, and selecting the variance value that best distinguishes these two types of regions (e.g., through ROC curve analysis). When the local variance of a pixel is lower than the threshold (determined as a flat region), a larger smoothing window (e.g., 7x7) is used to strongly remove residual noise; when the local variance is higher than the threshold (determined as an edge region), a smaller window (e.g., 3x3) is used or no smoothing is performed to preserve contour details.

[0034] For example, in the second filtered image, a background region with stable grayscale values ​​(local variance of 5.2, lower than the preset variance threshold of 10.0, which is set based on statistical analysis of flat and edge regions in historical images) is smoothed using a 7x7 window. Meanwhile, a contour edge (local variance of 28.0, higher than the threshold) is slightly smoothed using a 3x3 window. The resulting image is the smoothed transition image.

[0035] It should be noted that the boundary region detail extraction and local enhancement of the smooth transition image are achieved in two steps. First, a gradient operator (such as the Sobel operator) is used to convolve the smooth transition image to calculate the gradient magnitude of each pixel, thereby extracting details from the boundary region. Second, the extracted boundary region is locally enhanced. This enhancement operation is implemented through a non-linear enhancement function, which adjusts the pixel value of pixels with gradient magnitudes higher than a preset enhancement threshold. according to Enhancement is performed, where α is a preset enhancement coefficient greater than 1.

[0036] It is worth noting that the preset enhancement threshold is determined based on a statistical analysis of the gradient magnitude distribution of "flat regions" (i.e., non-edge noise regions) in the historical image dataset. By analyzing the gradient values ​​of a large number of flat regions, the 99th percentile of its distribution is calculated and set as the noise threshold. The enhancement threshold is set slightly higher than (e.g., 120%) this noise threshold. This setting ensures that only pixels identified as real edges (rather than noise) are enhanced, avoiding excessive amplification of noise.

[0037] In one alternative implementation, the preset enhancement coefficient α (e.g., 1.2) is determined based on the optimization of the enhancement effect on historical weak edge images. By comparing images processed with different α values, an image quality evaluation function (such as BRISQUE) is used to score the images, and the α value that maximizes the edge contrast score without introducing significant noise is selected as the preset value.

[0038] For example, after calculation using the Sobel operator, the gradient magnitude of a realistic but blurred weld edge is 40. This value is higher than the preset enhancement threshold of 25. The local enhancement operation then increases the pixel contrast of this edge by 30%, making it more prominent in the image. The processed image is the enhanced boundary transition image.

[0039] It should be noted that the contrast adjustment and interference region optimization of the enhanced boundary transition image aim to improve the overall dynamic range of the image and eliminate isolated noise points that may be introduced during the enhancement process. First, a histogram normalization method is used to remap the pixel grayscale values ​​of the enhanced boundary transition image, making their distribution closer to a Gaussian distribution to stretch the overall contrast. Then, a median filter is used to process the image to optimize the interference region, i.e., to remove salt-and-pepper noise that may be generated by sharpening.

[0040] It is worth noting that the determination of the window size of the median filter (e.g., 3x3) is based on statistical analysis of the typical noise speckle sizes generated after historical images are enhanced. Selecting a minimum window that covers the vast majority (e.g., 95%) of noise speckle sizes can effectively filter out isolated noise points while minimizing blurring of the contour data.

[0041] For example, an enhanced boundary transition image is generally dark, with several isolated bright pixels (salt-and-pepper noise) appearing near the edges. First, histogram normalization is used to expand its grayscale range from [30, 150] to [10, 240], making the contours clearly visible. Then, a 3x3 median filter is applied to traverse the image, replacing these isolated bright pixels with the median of their neighborhood, effectively removing the interference. The processed image is the optimized comparison image.

[0042] It should be noted that the refined marking of the welding area contour in the optimized comparison image is achieved through adaptive thresholding segmentation and morphological closing operations. First, the optimized comparison image is processed using the Otsu method. This method iterates through all possible grayscale values, automatically calculates and determines the grayscale value that maximizes the inter-class variance between the foreground (contour) and background pixels as the optimal segmentation threshold, and binarizes the image accordingly. Subsequently, morphological closing operations are applied to the binarized image to fill in any small holes or broken, nearby edges within the contour.

[0043] It is worth noting that the determination of the structuring element size (e.g., 5x5) for the morphological closing operation is based on a statistical analysis of the "typical gap" size of contour breaks in historical binarized images. Choosing a structuring element size slightly larger than the maximum typical gap ensures effective connection of most broken contours while avoiding the erroneous merging of adjacent independent contours.

[0044] For example, the Otsu method was used to calculate the optimal segmentation threshold for the optimized comparison image, determining it to be 142. After binarization based on this threshold, the weld contour was basically clear, but several tiny breaks with a width of 2-3 pixels existed. Subsequently, a morphological closing operation (dilation followed by erosion) was performed on the image using a 5x5 structuring element, successfully connecting these break points and forming a continuous and complete contour. This final binarized contour image data is the clear contour data.

[0045] In step S14, residual interference features are extracted from the clear contour data. If the residual interference features exceed a first interference threshold, the distribution characteristics of the residual interference features are analyzed to obtain environmental variable data. The distribution characteristics and the environmental variable data are then integrated and analyzed to obtain the degree of interference's impact on path planning, including: Obtain the environment variable data; From the clear contour data, interference residues are extracted to obtain the interference residue features; if the interference residue features exceed the first interference threshold, the interference residue features are scanned in layers to obtain a preliminary mapping image of the interference distribution; The preliminary mapping image is correlated and matched with the environmental variable data to obtain a comprehensive distribution map of interference and environmental factors; if the interference source in the comprehensive distribution map overlaps with the preset path planning area, a local avoidance calculation is performed to obtain adjusted path simulation data. Based on the adjusted path simulation data, the clear contour data is calibrated a second time to obtain the degree of impact of the interference on path planning.

[0046] It should be noted that acquiring the aforementioned environmental variable data refers to collecting real-time physical environment readings through non-visual sensors deployed around the welding equipment, such as light intensity meters. Specifically, this refers to light intensity values ​​(unit: lux). Environmental variable data includes, but is not limited to, light intensity, temperature, humidity, and airflow speed, collected through multi-sensor fusion. This data provides independent criteria for subsequent analysis of interference sources (e.g., distinguishing between excessive ambient light and equipment vibration).

[0047] It should be noted that the extraction of interference residues from the clear contour data is achieved through a two-stage extraction operation. The first stage is intensity threshold extraction, which traverses the clear contour data and marks all pixels with an intensity lower than a preset contour intensity as "low-intensity candidate points". The second stage is morphological feature extraction, which first obtains a preset continuous weld feature, which defines morphological parameters such as the minimum length and maximum curvature that a standard weld should have. Subsequently, this operation performs connected component analysis on the clear contour data (after excluding "low-intensity candidate points") using a two-pass scan-line based method. This method first traverses all pixels, assigns initial labels based on the connectivity of their 8-neighborhoods (i.e., top, bottom, left, right, and diagonally adjacent pixels) and records the equivalence relationships between the labels; then, a second scan is performed to merge equivalent labels, assigning a unique cluster ID to all pixels belonging to the same connected component, thereby identifying all independent pixel clusters. The length (i.e., the total number of pixels within the cluster) and curvature of each cluster are then calculated. Clusters of pixels that do not morphologically conform to the preset continuous weld seam characteristics (e.g., length less than the minimum length or curvature greater than the maximum curvature) are marked as "morphological interference points". Finally, the set of "low-intensity candidate points" and the set of "morphological interference points" are merged to obtain the interference residual feature.

[0048] It is worth noting that the preset continuous weld features (e.g., minimum length 50 pixels, maximum curvature 0.1) and the preset contour strength (e.g., grayscale value 100) are determined based on statistical analysis of clear contour data of historical "standard" welding images (i.e., images acquired under ideal conditions with extremely low interference), and the average value of their shape and strength is selected as the benchmark.

[0049] For example, assume a preset contour strength of 100 and a preset minimum length of 50 pixels for continuous weld features. Morphological analysis traverses the clear contour data, first marking all pixels with an intensity below 100 as "low-intensity candidate points." Then, connected component analysis is performed on the remaining pixels, identifying an isolated pixel cluster with a length of only 10 pixels. Since 10 pixels is less than the minimum length of 50 pixels, this cluster is marked as a "morphological interference point." Finally, all "low-intensity candidate points" and "morphological interference points" are merged to obtain the interference residual feature.

[0050] It is worth noting that the determination of the first interference threshold is based on statistical analysis of clear contour data from historical "standard" welding images (i.e., images acquired under ideal conditions with extremely low interference). The distribution of the intensity and density of "interference residue features" in these standard samples is calculated, and the 99th percentile of this distribution is selected as the first interference threshold. This setting ensures that subsequent complex analysis procedures are only triggered when the interference residue significantly exceeds the normal fluctuation range.

[0051] It should be noted that if the residual interference features exceed the first interference threshold, the residual interference features are then subjected to a layered scan. This scan operation classifies the interference signals according to their intensity. For example, pixels with intensity values ​​between 50 and 80 are marked as "low-intensity interference layer," and those with intensity values ​​greater than 80 are marked as "high-intensity interference layer," thereby visually displaying the distribution range of the interference and obtaining a preliminary mapping image of the interference distribution.

[0052] For example, in the clear contour data, the average intensity of the detected residual interference features was 90, exceeding the first interference threshold of 85. The residual interference features were then divided into two layers according to intensity, generating a preliminary mapping image of the interference distribution, which showed that high-intensity interference was concentrated in the right-hand region of the image.

[0053] It should be noted that the correlation matching between the preliminary mapping image and the environmental variable data involves mapping the spatial coordinates of the interference in the preliminary mapping image to the physical deployment locations of the sensors in the environmental variable data. For example, if the preliminary mapping image shows that the interference is concentrated in the right-hand area of ​​the image, and the environmental variable data shows that the reading of the light intensity meter deployed on the right side of that physical location (e.g., 200 lux) is significantly higher than in other areas (e.g., 100 lux), then the source of the interference is determined to be strong light from that physical location. If the source of interference identified in the comprehensive distribution map overlaps with a preset path planning area (this area is the target welding path coordinates preset according to the CAD model of the welded workpiece), it indicates that the interference has actually affected the welding path, and at this time, local avoidance calculation is initiated. This calculation applies a virtual repulsive force field to the preset path planning area based on the density and intensity of the interference distribution. The magnitude of the force is proportional to the intensity of the interference. An avoidance vector orthogonal to this repulsive force field is calculated. For the coordinates of the original path With modifications, its calculation formula can be simplified to: The calculation simulates a new path that bypasses the high-interference area, which is the adjusted path simulation data.

[0054] For example, the initial mapping image shows that the interference is concentrated on the right side. Simultaneously, the acquired environmental variable data shows that the light intensity meter reading deployed on the right side reaches 200 lux, significantly higher than the 100 lux on the left. Through correlation matching, a comprehensive distribution map of the interference and environmental factors is obtained, determining that the source of the interference is uneven lighting on the right side. Since this high-interference area overlaps with the preset welding path, local avoidance calculation is triggered, simulating a new path that shifts the original path 10 pixels to the left, resulting in the adjusted path simulation data.

[0055] It should be noted that the secondary calibration of the clear contour data based on the adjusted path simulation data is achieved through boundary optimization. This operation uses the adjusted path simulation data as a guiding benchmark and applies a B-spline interpolation algorithm to smooth the clear contour data (which may have spurs or offsets due to interference). A cubic B-spline curve can be used, with control points generated by equally spaced sampling of the adjusted path simulation data. The recursive calculation uses the De Boor algorithm. First, a new set of control points is determined based on the adjusted path simulation data. Then, these control points are weighted and averaged using B-spline basis functions to recursively calculate a smooth curve that maintains the general trend of the original path while continuously changing its curvature, thus fitting its boundary features to the simulated new path. Finally, by comparing the calibrated contour data with the uncalibrated contour data, the average deviation between the two is calculated. This quantification result represents the degree of influence of the interference on path planning.

[0056] For example, the clear contour data exhibits irregular jagged edges due to interference. A boundary optimization operation smooths these jagged edges using B-spline interpolation based on the adjusted path simulation data (shifted 10 pixels to the left), resulting in a calibrated contour. Comparison reveals that 30% of the original path segments are located in high-interference areas, while only 5% of the calibrated path is in such areas. The quantification of this discrepancy (e.g., a report describing path offset and risk reduction) represents the degree of impact of the interference on path planning.

[0057] In step S15, based on the degree of impact of the interference on path planning, dynamic parameter updates and interference compensation calculations are performed to obtain compensation data, including: From the degree of impact of the interference on path planning, the distribution characteristics of path deviation and the planning adjustment area are extracted, and the distribution characteristics of path deviation are scanned in layers to obtain a preliminary mapping map of the deviation. Acquire relevant data on environmental changes, and perform correlation matching between the preliminary mapping map and the relevant data on environmental changes to obtain a comprehensive distribution feature map; If the deviation area shown in the comprehensive distribution feature map overlaps with the planning adjustment area, then a local correction calculation is performed to obtain the adjusted simulation data; Based on the adjusted simulation data, boundary feature optimization processing is performed to obtain the compensation data.

[0058] It should be noted that before performing step S15, the impact of the interference on path planning obtained in step S14 (which includes pixel coordinate-based deviation data, such as "10 pixel units" in the S14 example) needs to be converted into data in the actual physical coordinate system based on a pre-acquired camera calibration result. Therefore, all subsequent calculations in step S15 are performed in the physical coordinate system (e.g., 1.5 mm), thereby ensuring that the calculated compensation value can be directly used to drive the physical device.

[0059] It is worth noting that the pre-acquired camera calibration results are obtained through a calibration operation during the device initialization phase. This operation uses a standard calibration board (such as a checkerboard) to collect image coordinates and corresponding physical coordinates measured by the device coordinate system (e.g., by manually controlling the laser head) at multiple different positions and orientations of the laser welding equipment's worktable. Finally, by applying a camera calibration algorithm (such as the Zhang Zhengyou calibration method), these coordinate pairs are calculated and fitted to obtain a transformation matrix or mapping function that can uniquely map any pixel coordinate to physical coordinates (unit: millimeters). This result is stored for later retrieval.

[0060] It should be noted that extracting the distribution characteristics of path deviation and the planning adjustment area from the degree of impact of the interference on path planning is achieved by parsing the structured data output in step S14. This data already includes the quantified value of the path offset caused by the interference (i.e., the distribution characteristics of the path deviation) and the coordinates of the affected area (i.e., the planning adjustment area). The subsequent layered scanning operation classifies the distribution characteristics of the path deviation according to the magnitude of the offset. For example, areas with offset values ​​between 1.5 mm and 3 mm are marked as "moderate deviation layer," and areas with offset values ​​greater than 3 mm are marked as "high deviation layer," thereby obtaining a preliminary mapping map of the deviation.

[0061] It should be noted that acquiring the relevant data on environmental changes refers to real-time collection of physical parameters at the welding site using non-visual sensors (such as hygrometers and airflow meters), specifically airflow velocity data (unit: m / s). The correlation and matching between the preliminary mapping map and the relevant data on environmental changes is achieved using a spatiotemporal correlation analysis method. This method first synchronizes the preliminary mapping map (containing spatial coordinates) and the airflow velocity data (containing sensor coordinates and timestamps) in time; then, the space is divided into n grid regions (n ​​can be dynamically adjusted according to image resolution and sensor density), and two values ​​are extracted for each region i: the average deviation intensity of that region. (From the preliminary mapping) and the average airflow velocity in the region (Relevant data from the environmental changes); Finally, calculate these two vectors. and The Pearson correlation coefficient r between the Z-score normalized values ​​is calculated using the following formula: ,in and These are the average values ​​of the two vectors. If the correlation coefficient r exceeds a preset correlation threshold, a causal relationship is determined between the two vectors, aiming to analyze the causal relationship between path deviation and environmental factors.

[0062] It is worth noting that the determination of the correlation threshold (e.g., 0.8) is based on regression analysis of historical data. By analyzing a large number of known "path deviation" and "environmental factors" sample pairs, their Pearson correlation coefficients r are calculated, and the coefficient value that can confirm the causal relationship with the highest confidence level (e.g., 95%) is selected as the preset value.

[0063] For example, the preliminary mapping of the deviation shows that the "height deviation layer" is concentrated above the path. Meanwhile, relevant environmental change data shows that the airflow velocity in this area is 3 m / s, significantly higher than the 1 m / s in other areas. Through spatiotemporal correlation analysis, the correlation coefficient between the two is calculated to be 0.85, exceeding the correlation threshold of 0.8, indicating that the path deviation may be caused by airflow disturbance, thus obtaining the comprehensive distribution feature map.

[0064] It should be noted that if the deviation area shown in the comprehensive distribution feature map overlaps with the planned adjustment area, it indicates that environmental changes have had a substantial impact on the welding path, and in this case, a local correction calculation is initiated. This calculation in this embodiment is implemented using the Artificial Potential Field method. This method converts the deviation density in the comprehensive distribution feature map into a repulsive potential field. The target point of the planned adjustment area is set as the gravitational center to generate a gravitational potential field. By calculating the total potential field negative gradient A new path is planned that simultaneously approaches the target point and avoids the deviation area (repulsion source), and this path is the adjusted simulation data.

[0065] For example, the integrated distribution feature map shows that the "height deviation layer" above the path overlaps with the planned adjustment area. Local correction calculation is triggered, and the artificial potential field method simulates a new path that shifts the original path downward by 3 mm based on the strong repulsive force generated by the "height deviation layer" to ensure that areas with deviation values ​​greater than 3 mm are avoided, thus obtaining the adjusted simulation data.

[0066] It should be noted that the boundary feature optimization processing based on the adjusted simulation data is achieved by applying a B-spline interpolation algorithm to the original boundary points of the planned adjustment area for smooth adjustment. This process first calculates the curvature $k$ of each point on the original boundary path and defines points with curvature $k$ exceeding a preset curvature threshold as "irregular fluctuation points". During the B-spline interpolation calculation, the algorithm introduces the coordinates of the adjusted simulation data (i.e., the new path) as strongly constrained control points, while simultaneously reducing the weight of the "irregular fluctuation points" in the interpolation calculation, thereby generating an optimized boundary that is smooth, has continuous curvature, and accurately fits the new path. This optimized and calibrated final path correction data, which can be used to compensate for the original path, is the compensation data.

[0067] In step S16, for the compensation data, path deviation correction, welding trajectory optimization, and path planning scheme update are performed to obtain optimized trajectory data, including: Obtain the initial welding path planning scheme and real-time environmental conditions; The compensation data is processed in layers to obtain deviation correction parameters; Based on the deviation correction parameters, the initial welding path planning scheme is locally smoothed to obtain a smoothed path planning scheme. The smoothed path planning scheme is compared and analyzed with the real-time environmental conditions to obtain the adjustment coefficient for environmental adaptation. Based on the adjustment coefficient for environmental adaptation, the smoothed path planning scheme is dynamically updated to obtain the optimized trajectory data.

[0068] It should be noted that obtaining the initial welding path planning scheme refers to retrieving a preset ideal welding path coordinate sequence from the database, without considering real-time interference. Obtaining the real-time environmental conditions refers to collecting environmental parameters of the welding area in real time, i.e., temperature and humidity data, through devices such as temperature sensors.

[0069] It is worth noting that the determination of the preset ideal welding path coordinate sequence is based on the three-dimensional CAD (Computer-Aided Design) model of the welding workpiece and is generated in advance in combination with welding process requirements (such as weld type and depth).

[0070] It should be noted that the hierarchical processing of the compensation data is achieved through data integration. This process classifies the compensation data (i.e., path correction data) output by S15 according to the deviation intensity and extracts its key information. For example, it calculates the average offset vector of the "height deviation layer". This calculation is achieved by taking the arithmetic mean of the X offset component and Y offset component of all data points in the layer, and uses it as the deviation correction parameter.

[0071] For example, suppose the compensation data contains a region labeled "height deviation layer" with 100 data points. By summing the X offsets of these 100 points and dividing by 100, the average X offset is +0.2 mm; by summing the Y offsets and dividing by 100, the average Y offset is -2.8 mm. Therefore, the resulting deviation correction parameter is a vector of (+0.2, -2.8) mm.

[0072] It should be noted that the local smoothing of the initial welding path planning scheme based on the deviation correction parameters is achieved through a trajectory adjustment operation. This operation applies the deviation correction parameters, for example, a vector of (+0.2, -2.8) mm, to the affected sections of the initial welding path planning scheme, i.e., performing vector addition on each coordinate point of that section. To avoid abrupt inflection points, this processing employs a B-spline interpolation algorithm for smoothing. The specific implementation of this B-spline interpolation algorithm is consistent with that described in step S14. By calculating new control points on the adjusted coordinate points, a trajectory with continuous curvature is generated, ensuring the continuity of the path and the controllability of curvature changes, thus obtaining the smoothed path planning scheme.

[0073] It should be noted that the comparison and analysis between the smoothed path planning scheme and the real-time environmental conditions aims to perform secondary fine-tuning of the path to adapt to other environmental factors. This analysis compares the smoothed path planning scheme with the real-time environmental conditions (such as temperature changes). If a deviation is found between the path correction value and the real-time environmental conditions (for example, increased temperature may cause thermal expansion of materials, requiring additional compensation), an adjustment coefficient for environmental adaptation is calculated for fine-tuning.

[0074] It is worth noting that the adjustment coefficient for environmental adaptation is determined based on a preset "environment-compensation" mapping model. By using historical experimental data, the nonlinear relationship between different real-time environmental conditions (such as temperature and humidity) and the required compensation value is statistically analyzed. These relationships are then fitted through regression analysis to obtain a predictive model that can output the corresponding adjustment coefficient based on the real-time environmental input.

[0075] For example, the smoothed path planning scheme already includes a -3 mm Y-axis compensation. At this time, the acquired real-time environmental conditions show that the welding point temperature has increased by 15°C. By querying the "environment-compensation" mapping model, it is determined that this temperature change requires an additional +0.2 mm compensation. Therefore, the environmental adaptation adjustment factor is obtained as +0.2 mm.

[0076] It should be noted that, according to the adjustment coefficient for environmental adaptation, the smoothed path planning scheme is dynamically updated by superimposing the adjustment coefficient onto the smoothed path planning scheme. For example, the Y-axis compensation is updated from -3 mm to -2.8 mm (-3 mm + 0.2 mm), thereby generating a final path data that simultaneously considers S14 interference analysis and S16 environmental adaptation, which is the optimized trajectory data.

[0077] In step S17, based on the optimized trajectory data, the drive control system performs precise positioning of the laser welding equipment to determine the final welding operation sequence, including: The optimized trajectory data is parsed to obtain key positioning point parameters; The parameters of the key positioning points are detected in real time to obtain the detection results; If the detection result exceeds the preset position threshold, the position of the laser welding equipment is adjusted to obtain calibrated position data; if the detection result does not exceed the preset position threshold, the key positioning point parameters are determined as the calibrated position data. Obtain the welding sequence plan, and perform correlation processing with the calibrated position data to obtain the appropriate execution instructions; According to the execution instructions, the laser welding equipment is guided to perform precise positioning to determine the final welding operation sequence.

[0078] It should be noted that data parsing of the optimized trajectory data involves extracting all coordinate points associated with subsequent welding actions (such as laser activation, power adjustment, and laser deactivation) from the optimized trajectory data (usually a sequence of coordinate points) and using them as the key positioning point parameters.

[0079] For example, the optimized trajectory data is a sequence containing 1000 coordinate points. The data parsing operation traverses this sequence and, according to the preset welding sequence plan, extracts the 1st point (as the "laser on" point), the 300th point (as the "power adjusted to 90%" point), the 700th point (as the "power adjusted to 80%" point), and the 1000th point (as the "laser off" point). These four coordinate points are then determined as the key positioning point parameters.

[0080] It should be noted that the real-time detection of the key positioning point parameters involves the control system first sending an absolute positioning command to the laser welding equipment, causing it to move to the theoretical target position of the key positioning point parameters; after the equipment is in place, the actual physical coordinates of the equipment are read through the equipment's built-in position feedback system (such as an encoder), and the deviation vector between the actual physical coordinates and the theoretical target position is calculated. This deviation vector is the detection result.

[0081] For example, the control system sends an absolute positioning command, with the target position being the key positioning point parameters (X: 100.50, Y: 50.00). After the laser welding equipment moves into position, its encoder reports the actual physical coordinates as (X: 100.52, Y: 49.99). By calculating the difference between the actual coordinates and the target coordinates, the detection result (i.e., the deviation vector) is obtained as (+0.02, -0.01).

[0082] It is worth noting that the preset position threshold (e.g., 0.1 mm) is determined based on the precision requirements of the welding process and the repeatability of the equipment itself. This threshold is determined by statistically analyzing the pass rate data of historical welding tasks, selecting the highest tolerable deviation value that can guarantee a product pass rate of over 99.7% (3-Sigma) as the preset value.

[0083] It should be noted that if the magnitude of the detection result (i.e., the deviation vector) exceeds the preset position threshold, it indicates that there is a deviation in the device positioning, and an adjustment operation is triggered. This operation sends a relative displacement command to the device's servo control system, and the compensation vector of this command is the negative value of the deviation vector. The servo system drives the motor to perform this compensation movement and repeatedly executes the second step (real-time detection) and the fourth step (adjustment operation) of S17 in this embodiment until the detection result falls within the threshold range. At this point, the final actual position is determined as the calibrated position data. If the detection result does not exceed the preset position threshold, the key positioning point parameter (theoretical target position) is directly determined as the calibrated position data.

[0084] For example, a key positioning point parameter is (X: 100.50, Y: 50.00). The deviation vector calculated in real-time is (+0.12, +0.01), whose magnitude exceeds the preset position threshold of 0.1 mm. At this time, adjustment is triggered, and the system receives a relative displacement command (-0.12, -0.01). After compensating for the movement, real-time detection is performed again, and the calculated new deviation vector is (-0.01, 0.00), whose magnitude does not exceed the threshold. The actual position at this time (X: 100.49, Y: 50.00) is determined as the calibrated position data.

[0085] It should be noted that obtaining the welding sequence plan refers to retrieving a preset set of operation instructions associated with the current task from memory. This plan defines "what to do" at each positioning point (e.g., laser on, laser off, power adjustment, wire feed speed).

[0086] It is worth noting that the construction of the preset operation instruction set is completed offline. This construction process is based on the three-dimensional CAD model of the welded workpiece and the welding process card. Technicians pre-calibrate all key coordinate points (i.e., the source of the key positioning point parameters) and bind a corresponding process action (such as "laser activation") to each coordinate point, thereby forming an instruction sequence containing [coordinate, action] key-value pairs.

[0087] It should be noted that the association processing between the calibrated position data and the welding sequence planning is implemented using a lookup table matching operation. This operation uses the calibrated position data (i.e., a confirmed coordinate) as the lookup key, searches within the welding sequence planning (a sequence of [coordinate, action] key-value pairs), and retrieves the action value associated with that coordinate key. Finally, this process combines the "position" and "action" into a complete, executable machine instruction, which is the adapted execution instruction. If the deviation between the calibrated position data and the coordinates in the welding sequence planning is less than a tolerance threshold (e.g., 0.5 mm), the nearest neighbor coordinate is matched; otherwise, re-path planning is triggered. All processing steps are executed in a real-time loop with a cycle not exceeding 100 ms, ensuring timely compensation for dynamic interference.

[0088] It should be noted that guiding the laser welding equipment to precise positioning according to the adapted execution instructions means sending the adapted execution instructions to the motion controller and the laser controller. The motion controller drives the equipment to the coordinate position specified in the instructions, and upon reaching that position, the laser controller triggers the corresponding welding action (such as activating the laser). The set of all these sequentially executed instructions constitutes the final welding operation sequence.

[0089] In summary, this invention constructs a full-chain intelligent control process, from real-time image acquisition and multi-level filtering and denoising (image smoothing and noise suppression) to determine a clear outline, to fusing interference residual features and environmental variable data to dynamically analyze the degree of interference impact, and then to performing dynamic parameter updates, interference compensation calculations, and welding trajectory optimization. This achieves dynamic adaptive compensation and precise control of the welding path of laser welding equipment, solving the technical problems of low path planning accuracy and difficulty in real-time deviation correction in complex interference environments in existing technologies. It significantly improves the positioning accuracy of the equipment and the stability of the welding trajectory in variable environments.

[0090] Reference Figure 2 The second embodiment of the present invention provides an intelligent control system for laser welding equipment, comprising: The image processing module is used to acquire real-time image data of the welding area and perform image smoothing processing on the real-time image data to obtain a first filtered image; The image filtering module is used to perform noise suppression and filtering processing on the first filtered image to obtain a second filtered image; The contour determination module is used to perform noise reduction calculations on the second filtered image to determine clear contour data of the welding area; The interference analysis module is used to extract residual interference features from the clear contour data. If the residual interference features exceed a first interference threshold, the distribution characteristics of the residual interference features are analyzed to obtain environmental variable data. The distribution characteristics and the environmental variable data are then integrated and analyzed to obtain the degree of interference on path planning. The compensation calculation module is used to perform dynamic parameter updates and interference compensation calculations based on the degree of impact of the interference on path planning, and obtain compensation data. The path optimization module is used to perform path deviation correction, welding trajectory optimization, and path planning scheme update on the compensation data to obtain optimized trajectory data. The execution control module is used to drive the control system to perform precise positioning of the laser welding equipment based on the optimized trajectory data, and to determine the final welding operation sequence.

[0091] It should be noted that the intelligent control system for laser welding equipment provided in this embodiment of the invention is used to execute all the process steps of the intelligent control method for laser welding equipment in the above embodiment. The working principles and beneficial effects of the two are one-to-one, so they will not be described again.

[0092] This invention also provides an electronic device. The electronic device includes a processor, a memory, and a computer program stored in the memory and executable on the processor, such as an intelligent control program for laser welding equipment. When the processor executes the computer program, it implements the steps described in the embodiments of the intelligent control method for laser welding equipment, for example... Figure 1 The step S11 shown. Alternatively, when the processor executes the computer program, it implements the functions of each module / unit in the above system embodiments, such as the image processing module.

[0093] For example, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the electronic device.

[0094] The electronic device may be a desktop computer, laptop, handheld computer, or smart tablet, etc. The electronic device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that the above components are merely examples of electronic devices and do not constitute a limitation on the electronic device. It may include more or fewer components than described above, or combine certain components, or different components. For example, the electronic device may also include input / output devices, network access devices, buses, etc.

[0095] 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 electronic device, connecting all parts of the electronic device via various interfaces and lines.

[0096] The memory can be used to store the computer programs and / or modules. The processor implements various functions of the electronic device by running or executing the computer programs and / or modules stored in the memory and by calling data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0097] If the modules / units integrated into the electronic device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when 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 system capable of carrying the 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 included 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, computer-readable media do not include electrical carrier signals and telecommunication signals.

[0098] It should be noted that the system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the system embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0099] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A method for intelligent control of laser welding equipment, characterized in that, include: Real-time image data of the welding area is acquired, and the real-time image data is smoothed to obtain a first filtered image; Based on the first filtered image, noise suppression and filtering are performed to obtain the second filtered image; For the second filtered image, denoising calculations are performed to determine the clear outline data of the welding area; From the clear contour data, residual interference features are extracted. If the residual interference features exceed the first interference threshold, the distribution characteristics of the residual interference features are analyzed to obtain environmental variable data. The distribution characteristics and the environmental variable data are then integrated and analyzed to obtain the degree of interference on path planning. Based on the degree of impact of the interference on path planning, dynamic parameter updates and interference compensation calculations are performed to obtain compensation data; Based on the compensation data, path deviation correction, welding trajectory optimization, and path planning scheme update are performed to obtain optimized trajectory data; Based on the optimized trajectory data, the drive control system performs precise positioning of the laser welding equipment and determines the final welding operation sequence.

2. The intelligent control method for laser welding equipment according to claim 1, characterized in that, The process of acquiring real-time image data of the welding area and performing image smoothing processing on the real-time image data to obtain a first filtered image includes: Real-time image data of the welding area is acquired to obtain the original image sequence; The original image sequence is subjected to noise reduction processing to obtain the first filtered image.

3. The intelligent control method for laser welding equipment according to claim 1, characterized in that, The step of performing noise suppression and filtering processing on the first filtered image to obtain a second filtered image includes: The first filtered image is subjected to local noise suppression processing, and the intensity control parameters are adjusted to obtain intermediate image data; For the intermediate image data, edge feature information is extracted, and the edge feature information is enhanced to obtain a transition image with enhanced edges; Based on the brightness distribution characteristics of the edge-enhanced transition image, contrast equalization adjustment is performed to obtain the second filtered image.

4. The intelligent control method for laser welding equipment according to claim 1, characterized in that, The step of performing noise reduction calculations on the second filtered image to determine the clear contour data of the welding area includes: The second filtered image is subjected to local smoothing and range adjustment processing to obtain a smooth transition image; For the smooth transition image, boundary region detail extraction and local enhancement are performed to obtain an enhanced boundary transition image; Based on the brightness distribution characteristics of the enhanced boundary transition image, contrast adjustment and interference region optimization are performed to obtain an optimized comparison image. Based on the optimized comparison image, the contour of the welding area is finely marked to determine the clear contour data.

5. The intelligent control method for laser welding equipment according to claim 1, characterized in that, The process involves extracting residual interference features from the clear contour data. If the residual interference features exceed a first interference threshold, the distribution characteristics of the residual interference features are analyzed to obtain environmental variable data. The distribution characteristics and the environmental variable data are then integrated and analyzed to determine the degree of interference's impact on path planning, including: Obtain the environment variable data; From the clear contour data, interference residues are extracted to obtain the interference residue features; if the interference residue features exceed the first interference threshold, the interference residue features are scanned in layers to obtain a preliminary mapping image of the interference distribution; The preliminary mapping image is correlated and matched with the environmental variable data to obtain a comprehensive distribution map of interference and environmental factors; if the interference source in the comprehensive distribution map overlaps with the preset path planning area, a local avoidance calculation is performed to obtain adjusted path simulation data. Based on the adjusted path simulation data, the clear contour data is calibrated a second time to obtain the degree of impact of the interference on path planning.

6. The intelligent control method for laser welding equipment according to claim 1, characterized in that, The step involves dynamically updating parameters and calculating interference compensation based on the degree of impact of the interference on path planning, to obtain compensation data, including: From the degree of impact of the interference on path planning, the distribution characteristics of path deviation and the planning adjustment area are extracted, and the distribution characteristics of path deviation are scanned in layers to obtain a preliminary mapping map of the deviation. Acquire relevant data on environmental changes, and perform correlation matching between the preliminary mapping map and the relevant data on environmental changes to obtain a comprehensive distribution feature map; If the deviation area shown in the comprehensive distribution feature map overlaps with the planning adjustment area, then a local correction calculation is performed to obtain the adjusted simulation data; Based on the adjusted simulation data, boundary feature optimization processing is performed to obtain the compensation data.

7. The intelligent control method for laser welding equipment according to claim 1, characterized in that, The process involves correcting path deviations, optimizing welding trajectories, and updating path planning schemes based on the compensation data to obtain optimized trajectory data, including: Obtain the initial welding path planning scheme and real-time environmental conditions; The compensation data is processed in layers to obtain deviation correction parameters; Based on the deviation correction parameters, the initial welding path planning scheme is locally smoothed to obtain a smoothed path planning scheme. The smoothed path planning scheme is compared and analyzed with the real-time environmental conditions to obtain the adjustment coefficient for environmental adaptation. Based on the adjustment coefficient for environmental adaptation, the smoothed path planning scheme is dynamically updated to obtain the optimized trajectory data.

8. The intelligent control method for laser welding equipment according to claim 1, characterized in that, The step of driving the control system to perform precise positioning of the laser welding equipment based on the optimized trajectory data and determining the final welding operation sequence includes: The optimized trajectory data is parsed to obtain key positioning point parameters; The parameters of the key positioning points are detected in real time to obtain the detection results; If the detection result exceeds the preset position threshold, the position of the laser welding equipment is adjusted to obtain calibrated position data; if the detection result does not exceed the preset position threshold, the key positioning point parameters are determined as the calibrated position data. Obtain the welding sequence plan, and perform correlation processing with the calibrated position data to obtain the appropriate execution instructions; According to the execution instructions, the laser welding equipment is guided to perform precise positioning to determine the final welding operation sequence.

9. An intelligent control system for laser welding equipment, characterized in that, include: The image processing module is used to acquire real-time image data of the welding area and perform image smoothing processing on the real-time image data to obtain a first filtered image; The image filtering module is used to perform noise suppression and filtering processing on the first filtered image to obtain a second filtered image; The contour determination module is used to perform noise reduction calculations on the second filtered image to determine clear contour data of the welding area; The interference analysis module is used to extract residual interference features from the clear contour data. If the residual interference features exceed a first interference threshold, the distribution characteristics of the residual interference features are analyzed to obtain environmental variable data. The distribution characteristics and the environmental variable data are then integrated and analyzed to obtain the degree of interference on path planning. The compensation calculation module is used to perform dynamic parameter updates and interference compensation calculations based on the degree of impact of the interference on path planning, and obtain compensation data. The path optimization module is used to perform path deviation correction, welding trajectory optimization, and path planning scheme update on the compensation data to obtain optimized trajectory data. The execution control module is used to drive the control system to perform precise positioning of the laser welding equipment based on the optimized trajectory data, and to determine the final welding operation sequence.

10. An electronic device comprising a processor, a memory, and a computer program stored in the memory and executable on the processor; characterized in that: The computer program is used to execute the intelligent control method for laser welding equipment as described in any one of claims 1-8.

Images