Punching cutting path generation method and system combined with graphics optimization

Abstract

The present invention provides a method and system for generating a punching cutting path combined with graphic optimization, which relates to the field of path planning technology and includes: obtaining basic information of the punching cutting part; matching the punching cutting sample image and performing graphic semantic segmentation to obtain the punching cutting sample feature information, including geometric feature information and texture feature information; performing punching cutting path planning to generate an initial punching cutting path; performing graphic optimization to minimize the total path length, minimize the length of repeated paths, and minimize the number of path intersections to generate an optimized punching cutting path; performing pre-punching cutting and collecting pre-punching cutting feedback data, performing path feedback adjustment, and obtaining a target punching cutting path; and synchronizing to the punching cutting machine for punching cutting path guidance. The present invention solves the technical problem that traditional punching cutting path planning mostly uses simple linear or geometric methods, lacks effective path optimization, and results in low metal cutting efficiency and poor quality.

Description

Technical Field

[0001] The present invention relates to the technical field of path planning, and in particular to a method and system for generating a punching cutting path combined with graphic optimization. Background Art

[0002] Punching cutting technology is widely used in many fields, such as metal processing, electronic manufacturing, and the automotive industry. However, in the traditional punching cutting process, path planning often relies on manual experience, and the operator sets the cutting path based on experience. This method is not universal and is easily affected by human factors, especially when processing workpieces with complex shapes and high precision requirements. It is easy to lead to inefficient and inaccurate metal cutting paths; and although modern computer vision and deep learning technologies have made significant progress in image analysis, the existing punching cutting path generation methods have not fully utilized these technologies. The application of image semantic segmentation and feature extraction methods in path planning is still limited. There is a lack of in-depth analysis of complex punching cutting paths, and it is impossible to achieve accurate metal cutting path generation and optimization. Summary of the Invention

[0003] This application provides a punching cutting path generation method and system combined with graphic optimization, aiming to solve the technical problems that traditional punching cutting path planning mostly adopts simple linear or geometric methods, lacks effective optimization of the path, and leads to low metal cutting efficiency and poor quality.

[0004] The first aspect disclosed in the present application provides a method for generating a punching cutting path combined with graphic optimization, the method comprising: obtaining basic information of a punching cutting part, the basic information of the punching cutting part comprising model information of the punching cutting part and positioning information of the punching cutting part; matching a punching cutting sample image according to the model information of the punching cutting part, and performing graphic semantic segmentation on the punching cutting sample image to obtain feature information of the punching cutting sample, wherein the feature information of the punching cutting sample comprises geometric feature information and texture feature information; performing punching cutting path planning based on the positioning information of the punching cutting part and combining the feature information of the punching cutting sample to generate an initial punching cutting path; performing graphic optimization on the initial punching cutting path to generate an optimized punching cutting path by minimizing the total path length, minimizing the length of repeated paths, and minimizing the number of path intersections; performing pre-punching cutting using the optimized punching cutting path, collecting pre-punching cutting feedback data, performing path feedback adjustment according to the pre-punching cutting feedback data to obtain a target punching cutting path; synchronizing the target punching cutting path to a punching cutting machine for punching cutting path guidance.

[0005] The second aspect disclosed in the present application provides a punching cutting path generation system combined with graphic optimization, the system is used for the above-mentioned punching cutting path generation method combined with graphic optimization, the system includes: a basic information acquisition module for acquiring basic information of the punching cutting part, the basic information of the punching cutting part includes punching cutting part model information and punching cutting part positioning information; a graphic semantic segmentation module for matching the punching cutting sample image according to the punching cutting part model information, and performing graphic semantic segmentation on the punching cutting sample image to obtain punching cutting sample feature information, wherein the punching cutting sample feature information includes geometric feature information and texture feature information; a path planning module for Based on the punching cutting piece positioning information, the punching cutting path is planned in combination with the punching cutting sample feature information to generate an initial punching cutting path; a graphic optimization module is used to perform graphic optimization on the initial punching cutting path by minimizing the total path length, minimizing the repeated path length, and minimizing the number of path intersections to generate an optimized punching cutting path; a path feedback adjustment module is used to use the optimized punching cutting path for pre-punching cutting, and collect pre-punching cutting feedback data, and perform path feedback adjustment according to the pre-punching cutting feedback data to obtain a target punching cutting path; a path guidance module is used to synchronize the target punching cutting path to the punching cutting machine for punching cutting path guidance.

[0006] One or more technical solutions provided in this application have at least the following beneficial effects:

[0007] By obtaining the basic information of the punching and cutting parts, including model information and positioning information, and combining the geometric and texture features of the sample image to plan the cutting path, the accuracy of the path is ensured, and the initial path that meets the requirements can be generated for different cutting tasks; through graphic optimization, including minimizing the total path length, reducing repeated paths, and avoiding path intersections, the efficiency of the cutting process can be greatly improved, unnecessary path travel can be reduced, cutting time can be reduced, and mechanical wear can be reduced; by using the optimized path for pre-punching cutting and collecting feedback data, problems in the path, such as pauses, intersections or repetitions, can be identified and adjusted to ensure that the target cutting path generated is more accurate and efficient; by synchronizing the target path to the cutting machine for path guidance, the punching and cutting machine can be directly guided to operate according to the optimized path, thereby improving the overall production efficiency of metal cutting, reducing errors and waste in the metal cutting process, and ensuring metal cutting quality and production stability.

[0008] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 A flow chart of a punching cutting path generation method combined with graphic optimization provided in an embodiment of the present application.

[0010] Figure 2 Schematic diagram of the structure of the punching cutting path generation system combined with graphic optimization provided in an embodiment of the present application.

[0011] Explanation of the accompanying drawings: basic information acquisition module 10, graphic semantic segmentation module 20, path planning module 30, graphic optimization module 40, path feedback adjustment module 50, path guidance module 60. DETAILED DESCRIPTION

[0012] The embodiments of the present application provide a punching cutting path generation method and system combined with graphic optimization, thereby solving the technical problem that traditional punching cutting path planning mostly adopts simple linear or geometric methods, lacks effective optimization of the path, and leads to low metal cutting efficiency and poor quality.

[0013] After introducing the basic principles of this application, various non-limiting embodiments of this application will be specifically described below in conjunction with the accompanying drawings. It should be understood that the specific embodiments described here are only used to explain this application and are not used to limit this application.

[0014] Example 1, as Figure 1 As shown, an embodiment of the present application provides a method for generating a punching cutting path combined with graphic optimization, the method comprising:

[0015] Obtain basic information of the punching and cutting parts, wherein the basic information of the punching and cutting parts includes punching and cutting part model information and punching and cutting part positioning information.

[0016] Obtain basic information about the punched and cut parts, which is used for subsequent path planning and optimization. The punched and cut part model information refers to the model or identifier of the punched and cut part, including the size, material, structure and other information of the cut part. By obtaining this information, the correct punched and cut sample image and processing method can be selected; the punched and cut part positioning information refers to the position data of the punched and cut part on the production equipment, such as the coordinates or assembly position of the cut part on the workbench. These positioning information helps to accurately plan the cutting path and ensure that the path planning matches the layout of the actual cut part.

[0017] According to the punching cut part model information, the punching cut sample image is matched, and the punching cut sample image is subjected to graphic semantic segmentation to obtain punching cut sample feature information, wherein the punching cut sample feature information includes geometric feature information and texture feature information.

[0018] The punching and cutting part model information is used to match sample images corresponding to the model. These sample images have been annotated and contain key information such as punching and cutting. The sample images are then processed with graphic semantic segmentation. Semantic segmentation is a technology in computer vision that can label each pixel in the image as a specific category, such as hole position, cutting path, etc. The goal is to extract meaningful areas from the image for further feature analysis.

[0019] After semantic segmentation of the graphics, feature information is extracted from the segmented area. Geometric feature information includes aperture size, hole location, cutting contour shape, and interface contact angle; texture feature information includes surface roughness, crack size, and the number of surface defects. Texture features help evaluate the quality and effect of punching and cutting, and affect subsequent path optimization and adjustment.

[0020] Based on the punching cutting piece positioning information and in combination with the punching cutting sample feature information, punching cutting path planning is performed to generate an initial punching cutting path.

[0021] Based on the positioning information of the punching and cutting parts, the location of the punching points is first determined. Each punching point represents the specific location where the punching machine needs to punch. These points will form a set for subsequent path planning; and the starting point and end point of the cutting path are determined. The starting point is the starting position of the punching and cutting machine, and the end point is the completion position of the cutting. With this information, the two end point sets of the cutting path are formed, providing basic data for subsequent path planning. Using the punching point set, the cutting path starting point set, the end point set, and the punching and cutting sample feature information, the initial cutting path is planned. The specific path planning process is detailed in the subsequent steps. This cutting path covers all punching points and has a reasonable route sequence.

[0022] The initial punching cutting path is graphically optimized to generate an optimized punching cutting path by minimizing the total path length, minimizing the repeated path length, and minimizing the number of path intersections.

[0023] The total path length refers to the total movement distance of the punching and cutting machine from the starting point to the end point. In path optimization, the movement distance of the machine is minimized as much as possible by adjusting the direction of the path, thereby saving cutting time. This goal is achieved through optimization algorithms, such as the approximate algorithm of the traveling salesman problem. During the cutting process, some paths may be repeated due to improper planning. Repeated paths not only waste time but also increase the burden on the equipment. Therefore, during optimization, it is necessary to eliminate or reduce the length of repeated paths to ensure that the path of each punching point and cutting position is passed only once as much as possible. Path intersection means that the cutting machine will encounter intersections with other paths during movement, which may cause the cutting machine to pause, turn or additional positioning time. The optimization process needs to minimize path intersections and adjust the direction of the path to prevent intersections or conflicts between paths.

[0024] In order to achieve the above optimization goals, optimization algorithms such as greedy algorithms, genetic algorithms, and simulated annealing are used. By continuously iterating the optimization path, the generated optimized punching and cutting path will minimize time, improve cutting efficiency, and reduce equipment friction and loss, thereby improving production efficiency and cutting quality.

[0025] The optimized punching cutting path is used for pre-punching cutting, and pre-punching cutting feedback data is collected. Path feedback adjustment is performed according to the pre-punching cutting feedback data to obtain a target punching cutting path.

[0026] The optimized punching cutting path is transmitted to the actual punching cutting machine for pre-cutting operation. This is a trial operation stage, the main purpose of which is to verify the effectiveness of path planning. During the pre-punching cutting process, the punching cutting machine cuts the material according to the path and collects data related to the cutting process in real time to obtain pre-punching cutting feedback data, including processing motion pause data, path intersection and repetition data, etc.

[0027] Path adjustments are made based on the collected feedback data. Specifically, if the feedback data indicates path intersections or duplications, the path is adjusted to ensure the machine avoids these issues during the next cut. If long pauses are recorded, the path needs to be replanned to avoid excessive machine repositioning. The goal of path feedback adjustment is to use feedback data from the pre-punching cutting process to improve path planning, making the final target cutting path more accurate and efficient.

[0028] The target punching and cutting path is synchronized to the punching and cutting machine for punching and cutting path guidance.

[0029] The final target punching and cutting path is synchronized to the punching and cutting machine. The cutting machine automatically guides the cutting operation according to the path data. The cutting machine executes according to each punching point and cutting path on the path, ensuring that all punching and cutting actions meet the requirements, making the punching and cutting process more accurate and efficient.

[0030] Furthermore, the method of performing graphic semantic segmentation on the punching cut sample image to obtain punching cut sample feature information includes:

[0031] Perform graphic semantic segmentation on the punching cutting sample image to obtain the punching cutting mark area; perform geometric feature extraction on the punching cutting mark area to obtain the geometric feature information, wherein the geometric feature information includes aperture size features, aperture position features, cutting contour shape features, and interface contact angle features; perform texture feature extraction on the punching cutting mark area to obtain the texture feature information, wherein the texture feature information includes surface roughness features, crack size features, and surface defect quantity features.

[0032] Image semantic segmentation is a computer vision technology used to label each pixel in an image as a different category. The goal is to identify punching areas, cutting paths, and other related areas from punching and cutting sample images. During the image semantic segmentation process, the punching areas and cutting paths are marked as different areas to obtain punching and cutting identification areas. These areas will serve as the basis for subsequent feature extraction.

[0033] The geometric features of the punching and cutting mark areas in the image are extracted to obtain geometric feature information used to describe punching and cutting. Among them, the aperture size feature describes the diameter or other size features of the punching; the aperture position feature describes the position of the hole, including its coordinates relative to the cutting piece or its position on the workpiece; the cutting contour shape feature describes the contour shape of the cutting, such as the linearity, curvature, and angle of the cutting path; the interface contact angle feature describes the contact angle between the cutting path and other areas. For example, where two cutting paths intersect, the size of the contact angle will affect the cutting quality and efficiency.

[0034] Texture features are extracted from the punching and cutting mark areas in the image to obtain texture feature information used to judge the cutting quality and surface flatness. The surface roughness feature describes the smoothness or roughness of the cut surface, which is usually obtained by measuring the details of the surface texture, such as the amplitude of fluctuations, wavelength, etc. A rough surface means poor cutting quality or improper cutting parameters. The crack size feature describes whether there are cracks on the surface and their size characteristics. Cracks are usually related to factors such as material properties, cutting conditions, and tool wear. The appearance of cracks may affect the final cutting quality. The surface defect quantity feature describes the number of defects on the surface, including cracks, pores, scratches, etc. The number of surface defects usually reflects instability or equipment problems in the cutting process.

[0035] Furthermore, the method of performing graphic semantic segmentation on the punching cut sample image to obtain the punching cut mark area includes:

[0036] Based on the punching cutting detection records, a historical punching cutting image set is obtained; the punching cutting areas in the historical punching cutting image set are identified and their shapes are marked to obtain a sample punching cutting identification result set; based on graphic semantic segmentation, an encoder and a decoder are constructed, and the historical punching cutting image set and the sample punching cutting identification result set are used for training to obtain a punching cutting area identifier; the punching cutting area identifier is used to identify the punching cutting area of ​​the punching cutting sample image to obtain the punching cutting identification area.

[0037] Obtain punching and cutting inspection records, which contain historical punching and cutting image sets of previously executed punching and cutting tasks. The historical punching and cutting image sets contain different cutting areas, including punching positions, cutting paths and other information. These images can be collected from previous cutting tasks or obtained through automatic inspection systems. These image sets will serve as basic data for subsequent pattern recognition and training.

[0038] Each image in the historical image set is analyzed to identify the punching cut areas. For each punching cut area, it is marked according to shape features, such as circle, ellipse, etc. The identified punching cut areas and their shape information will constitute the sample punching cut recognition result set. These recognition results will be used as annotation data for subsequent model training.

[0039] Semantic segmentation typically employs convolutional neural networks, particularly U-Net and FCN architectures, which effectively segment images. Based on graph semantic segmentation, an encoder and decoder are constructed. The encoder extracts feature information from the image. Through multiple convolutional layers, the encoder converts the input image into a higher-level feature representation, extracting the image's deep semantic information. The decoder then converts the features extracted by the encoder back into an image, accurately classifying them at the pixel level and labeling each region in the image (such as punching areas and cutting paths). The combined efforts of the encoder and decoder result in more accurate image segmentation, enabling better identification and labeling of punching and cutting areas.

[0040] Training is performed using a historical punching cutting image set and a sample punching cutting recognition result set. The sample recognition result set provides image annotation data. The model learns through this data and gradually adjusts its internal parameters. The training process requires multiple rounds of iterations. Through continuous forward propagation and backpropagation, the model's loss function is optimized, thereby improving the model's accuracy in punching cutting area recognition. Through the above training, a punching cutting area identifier is finally obtained. This identifier can accurately identify punching areas, cutting paths and other related areas based on the input punching cutting sample images.

[0041] The punching cutting sample image is input into the punching cutting area identifier. The identifier processes the input image based on the previous training results and identifies target areas such as the punching area and the cutting path. During the image recognition process, the identifier assigns a label to each pixel in the image to identify whether it belongs to the punching area, cutting path area or other area. In this way, the punching cutting identification area of ​​the entire image will be extracted to form the punching cutting identification area. These identification areas provide important data for subsequent path planning and optimization.

[0042] Furthermore, the punching cutting path planning is performed based on the punching cutting piece positioning information and in combination with the punching cutting sample feature information to generate an initial punching cutting path, and the method includes:

[0043] Based on the positioning information of the punching cutting piece, a punching point set, a cutting path starting point set, and a cutting path end point set are extracted; the first punching point, the first cutting path starting point, and the corresponding first cutting path end point are randomly extracted; based on the first punching point, the first cutting path starting point, and the first cutting path end point, the first punching cutting record data is collected, wherein the first punching cutting record data includes multiple groups of punching cutting paths and multiple groups of punching cutting images; based on the geometric feature information and texture feature information of the punching cutting sample feature information, the multiple groups of punching cutting images are screened to obtain a punching cutting image screening result; the punching cutting path corresponding to the punching cutting image screening result is extracted as the initial punching cutting path.

[0044] The positioning information of the punching and cutting parts includes the position data of the punching and cutting parts in the equipment. These positioning information include the exact coordinates of each punching point and cutting point on the workpiece. According to the positioning information, all punching points are extracted from the punching and cutting parts. Each punching point represents the position where the punching operation needs to be performed in the cutting path; according to the positioning information, the starting point set of the cutting path and the end point set of the cutting path are extracted. The starting point of the cutting path refers to the starting position of the cutting path, and the end point of the cutting path refers to the end position of the cutting path. Each cutting path requires a clear starting point and end point to ensure the completion of the cutting and the accuracy of the position.

[0045] A randomly selected punch point from the extracted punch point set is used as the first punch point for analysis. Similarly, a randomly selected starting point from the set of cutting path starting points is matched with the corresponding end point. Together, the starting and end points define a complete cutting path. This process aims to introduce some randomness to avoid rigid path planning and increase path diversity and flexibility.

[0046] According to the first punching point, the starting point of the first cutting path, and the end point of the first cutting path, the correspondence between them is analyzed, and the first punching cutting record data obtained by the actual or simulated cutting operation is obtained according to the corresponding matching, wherein each punching cutting operation will generate a specific cutting path, and by executing different paths multiple times, multiple punching cutting path data are collected. These path data include information such as the path length and cutting sequence of each cutting; in addition to the path data, corresponding multiple groups of punching cutting images are also collected to record the effect after each cutting. These images can show the punching area, cutting path, and cutting quality of the material.

[0047] For each set of punching and cutting images, screening is performed based on the geometric and texture features of the images. The difference between each set of images and the predetermined standards is calculated, and images that meet the conditions are selected based on the degree of matching of these features. After screening, the punching and cutting image screening results are obtained, which include images that meet the requirements. This screening result will provide accurate basic data for subsequent path extraction and optimization.

[0048] The corresponding punching cutting path is extracted from each qualified image. These paths represent the motion trajectory of the cutting machine when performing punching cutting, including the direction, sequence and possible intersection points of the cutting path. The extracted punching cutting path is used as the initial punching cutting path. This initial path is generated based on the filtered image data and reflects the actual cutting conditions at different punching points and cutting paths.

[0049] Furthermore, the method of screening the plurality of groups of punching cut images based on the geometric feature information and the texture feature information of the punching cut sample feature information to obtain the punching cut image screening results includes:

[0050] Extract multiple groups of comparison geometric feature information and multiple groups of comparison texture feature information of the multiple groups of punching and cutting images; calculate multiple groups of geometric feature deviations between the multiple groups of comparison geometric feature information and the geometric feature information; calculate multiple groups of texture feature deviations between the multiple groups of comparison texture feature information and the texture feature information; based on the multiple groups of geometric feature deviations and the multiple groups of texture feature deviations, perform image screening of the multiple groups of punching and cutting images to obtain the punching and cutting image screening results.

[0051] The punching cut area identifier constructed in the previous steps is used to identify the punching cut areas in multiple sets of punching cut images. Based on the identification results, geometric and texture features are extracted to obtain multiple sets of comparison geometric feature information and multiple sets of comparison texture feature information. The specific feature extraction process has been detailed in the previous steps and is not repeated here. The comparison geometric feature information refers to the geometric features extracted from the comparison image and is used for comparison with the geometric features in the current image; the comparison texture feature information refers to the texture features extracted from the comparison image and serves as the basis for comparison with the texture features of the current image.

[0052] The geometric features of the current image are compared with the geometric features extracted from the reference image, and the deviation between them is calculated. The deviation can be a numerical difference, an angular difference, or a shape similarity. Common calculation methods include Euclidean distance and cosine similarity. The difference between each pair of geometric features is calculated, for example, the distance between two punching positions, the angular difference between two cutting paths, etc. The geometric features in multiple reference images are compared with the geometric features of the current image to obtain multiple groups of geometric feature deviations. Through these deviations, the gap between the current image and multiple historical sample images can be understood.

[0053] Similar to geometric features, the deviation between the texture features of the current image and the texture features of the control image is calculated. Texture features can include surface roughness, crack size, number of defects, etc. The deviation calculation is based on the numerical differences of these texture features, and multiple groups of texture feature deviations are obtained. These data can help determine whether there are obvious quality differences in the images, such as surface cracks, uneven roughness, etc.

[0054] Based on the calculated deviation results, multiple groups of punching and cutting images are screened. By comparison, those images with smaller deviations in geometric and texture features are selected, and they are considered to be more consistent with the target cutting path. If the deviations of some images are too large, they will be excluded because they may represent cutting results that do not meet the quality standards. After screening, the punching and cutting image screening results are finally obtained. These images meet the set standards and can be used for subsequent path extraction and optimization.

[0055] Furthermore, the method of performing image screening on the multiple sets of punching and cutting images based on the multiple sets of geometric feature deviations and the multiple sets of texture feature deviations to obtain the punching and cutting image screening results includes:

[0056] Obtain a geometric feature deviation threshold and a texture feature deviation threshold; eliminate the punching cut images that do not meet the geometric feature deviation threshold in the multiple sets of geometric feature deviations to obtain a first image screening result; eliminate the punching cut images that do not meet the texture feature deviation threshold in the multiple sets of texture feature deviations to obtain a second image screening result; integrate the first image screening result and the second image screening result to obtain the punching cut image screening result.

[0057] The geometric feature deviation threshold refers to the maximum difference allowed during the geometric feature comparison process, for example, the maximum allowable distance between two punching positions, the maximum allowable deviation of the cutting path angle, etc. If the difference between the two is greater than the set threshold, then the image will be considered unqualified. The threshold is set according to the requirements of the specific application. For example, for larger or complex punching and cutting tasks, the threshold can be relatively loose, while for tasks with higher precision requirements, the threshold may need to be stricter.

[0058] The texture feature deviation threshold refers to the maximum allowable surface quality difference, for example, the maximum allowable deviation of surface roughness, the maximum allowable difference of crack size, etc. If the deviation of the texture feature is greater than this threshold, the image will be considered to have unsatisfactory cutting quality and should be rejected. This threshold is set based on the surface quality standard of the workpiece. For tasks requiring high-quality surfaces, the threshold should be set lower.

[0059] For each group of punching and cutting images, the calculated geometric feature deviation is compared with the geometric feature deviation threshold. If the deviation is greater than the set threshold, it means that the geometric features of the image do not meet the requirements, and it will be eliminated from the screening results. Those images with deviations within the threshold are retained to obtain the first image screening results. The geometric features of this group of images meet the requirements and can continue to be used for subsequent texture feature screening.

[0060] For each group of punching and cutting images, the calculated texture feature deviation is compared with the texture feature deviation threshold. If the texture feature deviation is greater than the set threshold, it means that the texture quality of the image does not meet the requirements, and it will be eliminated from the screening results. Those images whose texture features meet the standards are retained to obtain the second image screening results. The texture features of these images meet the standards and can be used for subsequent punching and cutting path extraction.

[0061] The first image screening results are combined with the second image screening results to ensure that the final screened image meets the requirements of both geometric features and texture features. Through the integrated screening results, a punching cutting image that meets all requirements is finally obtained. These images can serve as the basic data for subsequent cutting path extraction and optimization.

[0062] Furthermore, the method of performing path feedback adjustment based on the pre-punching cutting feedback data to obtain a target punching cutting path includes:

[0063] Extract processing motion pause data, path intersection and repetition data based on the pre-punching cutting feedback data; perform processing path defect marking based on the processing motion pause data, the path intersection and repetition data, and perform path feedback adjustment based on the processing path defect marking results.

[0064] Processing motion pause refers to the pause or pause state of the punching and cutting machine when executing the cutting path. The pause may occur for multiple reasons, such as path adjustment, equipment error, long punching point interval, slow cutting speed, etc. The time and location of all pause events, as well as the duration of each pause, the location where it occurred, and the reason for the pause are extracted from the feedback data.

[0065] Path intersection occurs when different parts of the cutting path meet, which may cause incorrect movement or inefficiency of the cutting machine. Path intersection is usually caused by unreasonable path planning or the failure to effectively allocate multiple cutting paths. Extracting path intersection data indicates at which locations and under what circumstances the path intersection occurs, as well as the specific coordinates of the intersection points. This data helps to optimize the path planning later.

[0066] Path repetition refers to the situation where certain paths are repeated when the cutting machine is cutting. This may be due to inaccurate path planning or failure to remove redundant paths. The repeated path information is extracted from the feedback data to indicate which paths are repeated, as well as the length and location of the repeated paths. This data helps to identify and remove unnecessary repeated cutting.

[0067] By analyzing machining motion pauses, path intersections, and repetitions, defective areas in the path are marked, including pause points, intersections, and repeated paths. The marking results will help identify non-optimal parts of the cutting process and provide a basis for subsequent path adjustments. Based on the results of machining path defect marking, path adjustments are made, including removing unnecessary pauses, avoiding path intersections, and eliminating repeated paths. The purpose of path feedback adjustment is to improve cutting efficiency, reduce errors and unnecessary motion, and ultimately generate a more efficient and accurate cutting path.

[0068] In summary, the punching cutting path generation method combined with graphic optimization provided in the embodiments of the present application has the following technical effects:

[0069] By obtaining the basic information of the punching and cutting parts, including model information and positioning information, and combining the geometric and texture features of the sample image to plan the cutting path, the accuracy of the path is ensured, and the initial path that meets the requirements can be generated for different cutting tasks; through graphic optimization, including minimizing the total path length, reducing repeated paths, and avoiding path intersections, the efficiency of the cutting process can be greatly improved, unnecessary path travel can be reduced, cutting time can be reduced, and mechanical wear can be reduced; by using the optimized path for pre-punching cutting and collecting feedback data, problems in the path, such as pauses, intersections or repetitions, can be identified and adjusted to ensure that the target cutting path generated is more accurate and efficient; by synchronizing the target path to the cutting machine for path guidance, the punching and cutting machine can be directly guided to operate according to the optimized path, thereby improving the overall production efficiency of metal cutting, reducing errors and waste in the metal cutting process, and ensuring metal cutting quality and production stability.

[0070] Example 2, based on the same inventive concept as the punching cutting path generation method combined with graphic optimization in the previous embodiment, Figure 2 As shown, an embodiment of the present application provides a punching cutting path generation system combined with graphic optimization, the system comprising:

[0071] The basic information acquisition module 10 is used to acquire basic information of the punching and cutting parts, wherein the basic information of the punching and cutting parts includes the model information of the punching and cutting parts and the positioning information of the punching and cutting parts.

[0072] The graphic semantic segmentation module 20 is used to match the punching cut sample image according to the punching cut part model information, and perform graphic semantic segmentation on the punching cut sample image to obtain punching cut sample feature information, wherein the punching cut sample feature information includes geometric feature information and texture feature information.

[0073] The path planning module 30 is used to plan a punching cutting path based on the punching cutting piece positioning information and the punching cutting sample feature information to generate an initial punching cutting path.

[0074] The graphic optimization module 40 is used to perform graphic optimization on the initial punching cutting path by minimizing the total path length, minimizing the repeated path length, and minimizing the number of path intersections, and generate an optimized punching cutting path.

[0075] The path feedback adjustment module 50 is used to perform pre-punching cutting using the optimized punching cutting path, collect pre-punching cutting feedback data, and perform path feedback adjustment based on the pre-punching cutting feedback data to obtain a target punching cutting path.

[0076] The path guiding module 60 is used to synchronize the target punching and cutting path to the punching and cutting machine for punching and cutting path guidance.

[0077] Furthermore, the graphic semantic segmentation module 20 includes:

[0078] A semantic segmentation unit is used to perform graphic semantic segmentation on the punching and cutting sample image to obtain a punching and cutting mark area; a geometric feature extraction unit is used to perform geometric feature extraction on the punching and cutting mark area to obtain the geometric feature information, wherein the geometric feature information includes aperture size features, aperture position features, cutting contour shape features, and interface contact angle features; a texture feature extraction unit is used to perform texture feature extraction on the punching and cutting mark area to obtain the texture feature information, wherein the texture feature information includes surface roughness features, crack size features, and surface defect quantity features.

[0079] Furthermore, the graphic semantic segmentation module 20 includes:

[0080] A historical image acquisition unit is used to acquire a historical punching cut image set based on the punching cut detection record; a shape identification unit is used to identify and shape-identify the punching cut area in the historical punching cut image set to obtain a sample punching cut identification result set; a training unit is used to construct an encoder and a decoder based on graphic semantic segmentation, and use the historical punching cut image set and the sample punching cut identification result set for training to obtain a punching cut area identifier; an area identification unit is used to use the punching cut area identifier to identify the punching cut area of ​​the punching cut sample image to obtain the punching cut identification area.

[0081] Furthermore, the path planning module 30 includes:

[0082] A point extraction unit is used to extract a punching point set, a cutting path starting point set, and a cutting path end point set based on the punching cutting part positioning information; a first punching point extraction unit is used to randomly extract a first punching point, a first cutting path starting point, and a corresponding first cutting path end point; a cutting record data extraction unit is used to collect first punching cutting record data based on the first punching point, the first cutting path starting point, and the first cutting path end point, wherein the first punching cutting record data includes multiple groups of punching cutting paths and multiple groups of punching cutting images; a screening unit is used to screen the multiple groups of punching cutting images based on the geometric feature information and texture feature information of the punching cutting sample feature information to obtain a punching cutting image screening result; an initial punching cutting path acquisition unit is used to extract the punching cutting path corresponding to the punching cutting image screening result as the initial punching cutting path.

[0083] Furthermore, the path planning module 30 includes:

[0084] A comparison geometric feature information extraction unit is used to extract multiple groups of comparison geometric feature information and multiple groups of comparison texture feature information of the multiple groups of punching and cutting images; a geometric feature deviation calculation unit is used to calculate multiple groups of geometric feature deviations between the multiple groups of comparison geometric feature information and the geometric feature information; a texture feature deviation calculation unit is used to calculate multiple groups of texture feature deviations between the multiple groups of comparison texture feature information and the texture feature information; an image screening unit is used to perform image screening of the multiple groups of punching and cutting images based on the multiple groups of geometric feature deviations and the multiple groups of texture feature deviations to obtain the punching and cutting image screening results.

[0085] Furthermore, the path planning module 30 includes:

[0086] A threshold acquisition unit is used to obtain a geometric feature deviation threshold and a texture feature deviation threshold; a first elimination unit is used to eliminate the punching cut images that do not meet the geometric feature deviation threshold in the multiple groups of geometric feature deviations to obtain a first image screening result; a second elimination unit is used to eliminate the punching cut images that do not meet the texture feature deviation threshold in the multiple groups of texture feature deviations to obtain a second image screening result; a screening result acquisition unit is used to integrate the first image screening result and the second image screening result to obtain the punching cut image screening result.

[0087] Furthermore, the path feedback adjustment module 50 includes:

[0088] A data extraction unit is used to extract processing motion pause data, path intersection and repetition data based on the pre-punching cutting feedback data; a feedback adjustment unit is used to mark processing path defects based on the processing motion pause data, the path intersection and repetition data, and to adjust the path feedback according to the processing path defect marking results.

[0089] Through the above detailed description of the punching cutting path generation method combined with graphic optimization in this specification, those skilled in the art can clearly understand the punching cutting path generation system combined with graphic optimization in this embodiment. Since it corresponds to the method disclosed in the embodiment, the description is relatively simple. For relevant matters, please refer to the method part description.

[0090] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present application. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application is not limited to the embodiments shown herein, but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A punching cutting path generation method combined with graphic optimization, characterized in that: The method comprises: Obtaining basic information of the punching and cutting parts, wherein the basic information of the punching and cutting parts includes model information of the punching and cutting parts and positioning information of the punching and cutting parts; Matching punching cut sample images according to the punching cut part model information, and performing graphic semantic segmentation on the punching cut sample images to obtain punching cut sample feature information, wherein the punching cut sample feature information includes geometric feature information and texture feature information; Based on the punching cutting piece positioning information and in combination with the punching cutting sample feature information, a punching cutting path is planned to generate an initial punching cutting path; Performing graphical optimization on the initial punching cutting path to generate an optimized punching cutting path by minimizing the total path length, minimizing the repeated path length, and minimizing the number of path intersections; Perform pre-punching cutting using the optimized punching cutting path, collect pre-punching cutting feedback data, perform path feedback adjustment based on the pre-punching cutting feedback data, and obtain a target punching cutting path; The target punching and cutting path is synchronized to the punching and cutting machine for punching and cutting path guidance.

2. The punching cutting path generation method combined with graphic optimization according to claim 1, characterized in that: The method of performing graphic semantic segmentation on the punching and cutting sample image to obtain punching and cutting sample feature information includes: Performing graphic semantic segmentation on the punching and cutting sample image to obtain a punching and cutting mark area; Extracting geometric features of the punching and cutting mark area to obtain the geometric feature information, wherein the geometric feature information includes aperture size features, aperture position features, cutting contour shape features, and interface contact angle features; Texture feature extraction is performed on the punching and cutting mark area to obtain the texture feature information, wherein the texture feature information includes surface roughness features, crack size features, and surface defect quantity features.

3. The method for generating a punching cutting path combined with graphic optimization according to claim 2, wherein: The method of performing graphic semantic segmentation on the punching and cutting sample image to obtain the punching and cutting mark area includes: Based on the punching and cutting detection records, a historical punching and cutting image set is obtained; Identifying and shape-marking the punching cut areas in the historical punching cut image set to obtain a sample punching cut recognition result set; Based on the semantic segmentation of the graph, an encoder and a decoder are constructed, and the historical punching cut image set and the sample punching cut recognition result set are used for training to obtain a punching cut area identifier; The punching cutting area identifier is used to perform punching cutting area identification on the punching cutting sample image to obtain the punching cutting mark area.

4. The method for generating a punching cutting path combined with graphic optimization according to claim 1, wherein: The punching cutting path planning is performed based on the punching cutting piece positioning information and in combination with the punching cutting sample feature information to generate an initial punching cutting path, the method comprising: Extracting a punching point set, a cutting path starting point set, and a cutting path ending point set based on the punching and cutting piece positioning information; Randomly extracting a first punching point, a first cutting path starting point, and a corresponding first cutting path end point; Based on the first punching point, the first cutting path starting point, and the first cutting path end point, collecting first punching cutting record data, wherein the first punching cutting record data includes multiple sets of punching cutting paths and multiple sets of punching cutting images; Based on the geometric feature information and texture feature information of the punching cutting sample feature information, the plurality of groups of punching cutting images are screened to obtain a punching cutting image screening result; A punching cutting path corresponding to the punching cutting image screening result is extracted as the initial punching cutting path.

5. The method for generating a punching cutting path combined with graphic optimization according to claim 4, wherein: The method of screening the plurality of groups of punching cut images based on the geometric feature information and the texture feature information of the punching cut sample feature information to obtain the punching cut image screening results includes: Extracting multiple sets of comparison geometric feature information and multiple sets of comparison texture feature information of the multiple sets of punching and cutting images; Calculating the geometric feature deviations between the plurality of sets of control geometric feature information and the plurality of sets of geometric feature information; Calculating the deviations of the multiple sets of comparison texture feature information and the multiple sets of texture feature information; Based on the multiple groups of geometric feature deviations and the multiple groups of texture feature deviations, image screening of the multiple groups of punching and cutting images is performed to obtain the punching and cutting image screening results.

6. The method for generating a punching cutting path combined with graphic optimization according to claim 5, wherein: The method of performing image screening on the multiple sets of punching and cutting images based on the multiple sets of geometric feature deviations and the multiple sets of texture feature deviations to obtain the punching and cutting image screening results includes: Obtaining geometric feature deviation thresholds and texture feature deviation thresholds; Eliminating punching and cutting images that do not meet the geometric feature deviation threshold from the multiple groups of geometric feature deviations to obtain a first image screening result; Eliminating the punching cut images that do not meet the texture feature deviation threshold from the multiple groups of texture feature deviations to obtain a second image screening result; The first image screening result and the second image screening result are integrated to obtain the punching cutting image screening result.

7. The method for generating a punching cutting path combined with graphic optimization according to claim 1, wherein: The method of performing path feedback adjustment based on the pre-punching cutting feedback data to obtain a target punching cutting path includes: Extracting processing motion pause data, path intersection and repetition data based on the pre-punching cutting feedback data; Based on the processing motion pause data and the path intersection and repetition data, processing path defect marking is performed, and path feedback adjustment is performed according to the processing path defect marking result.

8. The punching cutting path generation system combined with graphic optimization is characterized by: A system for implementing the punching cutting path generation method combined with graphic optimization according to any one of claims 1 to 7, comprising: A basic information acquisition module is used to obtain basic information of the punching and cutting parts, wherein the basic information of the punching and cutting parts includes the punching and cutting part model information and the punching and cutting part positioning information; A graphic semantic segmentation module is used to match the punching cut sample image according to the punching cut part model information, and perform graphic semantic segmentation on the punching cut sample image to obtain punching cut sample feature information, wherein the punching cut sample feature information includes geometric feature information and texture feature information; A path planning module is used to plan a punching cutting path based on the punching cutting piece positioning information and the punching cutting sample feature information to generate an initial punching cutting path; A graphics optimization module is used to perform graphics optimization on the initial punching cutting path by minimizing the total path length, minimizing the repeated path length, and minimizing the number of path intersections to generate an optimized punching cutting path; a path feedback adjustment module, configured to perform pre-punching cutting using the optimized punching cutting path, collect pre-punching cutting feedback data, perform path feedback adjustment based on the pre-punching cutting feedback data, and obtain a target punching cutting path; The path guidance module is used to synchronize the target punching and cutting path to the punching and cutting machine for punching and cutting path guidance.

Images