A welding path planning method based on machine vision

Abstract

The present application discloses a welding path planning method based on machine vision, which relates to the field of welding robot technology. The method comprises: collecting and classifying and storing sample data of various groove types, establishing a groove type database, and defining an extraction condition template; using machine vision technology to obtain groove cross-section data, constructing a three-dimensional model after grouping, analyzing the characteristics of welds, materials and heat-affected zones, determining the welding sequence principle, and generating a comprehensive welding path; through real-time monitoring of key parameters in the welding process, setting an early warning mechanism, and dynamically adjusting the optimal path; updating the database according to the path planning efficiency value, and comprehensively analyzing the candidate paths to recommend the optimal welding path; dynamically adjusting the extraction condition template to improve the accuracy, adaptability and efficiency of welding path planning, and ensure welding quality and stability; analyzing multiple different cross-section data to determine a comprehensive welding path that can simultaneously meet the characteristics of multiple different cross-section data.

Description

Technical Field

[0001] The present invention relates to the technical field of welding robots, and in particular to a welding path planning method based on machine vision. Background Art

[0002] In the welding industry, the development and application of welding robots have been realized in standardized welding workshops and are being applied in more and more scenarios, achieving more efficient, safer and more standardized welding. The core tube in the construction industry is in the form of box-shaped steel, and the crossbeam is in the form of H-shaped steel. The steel structure is thick, the groove is deep, and the degree of groove standardization is low. For the deeper grooves in building steel structures, multi-layer and multi-pass welding is usually used. The application of welding robots in multi-layer and multi-pass welding requires the completion of groove weld bead planning and welding path planning before welding begins. The introduction of machine vision technology provides strong support for its application in multi-layer and multi-pass welding. Through high-resolution cameras and advanced image processing algorithms, the shape, size and surface condition of the groove are captured and analyzed in real time, ensuring the accuracy and consistency of the welding process. Existing welding path planning methods cannot effectively guarantee welding quality when faced with non-standard grooves.

[0003] The Chinese invention patent with application number 202210985990.3 is a groove welding path planning method, device, electronic device and storage medium, which collects multiple sets of groove data and extracts key information from the multiple sets of groove data based on preset extraction conditions to form multiple sets of groove section data; for each groove section data, the number of welding layers and the number of welding passes of each welding layer of the groove section are calculated using preset weld parameters; the position coordinates of each weld pass are marked in each groove section, and the groove welding path is planned based on the weld passes with the same position coordinates in each groove section.

[0004] In the existing patented technology, cross-sectional data analysis is performed on irregular grooves to plan welding paths. However, the cross-sectional data obtained for irregular grooves may also have obvious differences. When faced with cross-sectional data with different characteristics, it is impossible to accurately plan the path based on the position coordinates. Therefore, the existing patented technology ignores the differentiated characteristics of different cross-sectional data, resulting in the inability to extract key information based on preset extraction conditions, reducing the accuracy of cross-sectional data analysis and affecting the accuracy of welding path planning. Summary of the Invention

[0005] This application solves the problem in the prior art of ignoring the differentiated features of data of different cross-sections and reducing the accuracy of welding path planning by providing a welding path planning method based on machine vision. It achieves the technical effect of accurately identifying the differentiated features of data of different cross-sections and improving the accuracy of welding path planning.

[0006] The present application provides a welding path planning method based on machine vision, the method comprising:

[0007] S100: Collect and classify sample data of various groove types, establish a groove type database, and define an extraction condition template for each groove type;

[0008] S200: Identify the current groove type, select the corresponding extraction condition template, obtain groove section data, group the groove section data according to different characteristics, establish a three-dimensional model based on multiple groups of groove section data, generate a comprehensive welding path, and generate a welding analysis report based on the actual welding effect;

[0009] S300: According to the welding analysis report, the path planning status of each groove type is obtained, the path planning efficiency value is generated, and the groove type database is updated;

[0010] Wherein, step S200 includes:

[0011] S210: Acquire groove cross-section data based on machine vision technology, including a two-dimensional image and relative position information, and pre-process the acquired groove cross-section data to obtain standard cross-section data;

[0012] S220: Setting a grouping strategy to group the standard cross-section data, selecting a reconstruction strategy for each group of cross-section data according to its characteristics, and constructing a three-dimensional model using a surface reconstruction algorithm based on the reconstruction strategy for each group of cross-section data;

[0013] S230: Analyze weld characteristics, material characteristics, and heat-affected zone sensitivity based on a 3D model to determine welding sequence principles;

[0014] S240: generating a preliminary welding path corresponding to each cross-section group according to each group of cross-section data and a welding sequence principle, and merging the preliminary welding paths of each cross-section group to form a comprehensive welding path.

[0015] Furthermore, the grouping strategy is to group different sections according to the shape similarity and the characteristic nodes between different sections. For the two sections with the highest shape similarity, the trajectory trends of their adjacent nodes are analyzed, and the adjacent nodes with the same trajectory are selected as characteristic nodes to group the groove section data.

[0016] Furthermore, the trajectory trend refers to the welding angle passing through the node when welding between different cross sections. If the welding angles are consistent, the trajectories are determined to be the same; if the welding angles change, the trajectories are determined to be different.

[0017] Furthermore, step S240 includes:

[0018] S241: Obtain each set of cross-sectional data, extract key feature points respectively, and obtain the influence value of each set of cross-sectional data according to the number of key feature points. Compare the influence value with the influence threshold. If the influence value is greater than the difference threshold, mark it, and generate preliminary welding paths according to the marking.

[0019] S242: generating an optional path group by fusing the preliminary welding paths and annotations, setting evaluation indicators to score each optional path, and selecting the path with the highest comprehensive score as the comprehensive welding path.

[0020] Furthermore, the influence value is used to evaluate the influence of the key feature point on the welding path, and the influence threshold is set according to all current influence values. All influence values ​​are arranged in descending order, and the average value of the top 20% of the influence values ​​is selected as the influence threshold.

[0021] Furthermore, step S240 also includes:

[0022] S243: Real-time monitoring of key parameters and conditions during welding, processing and analysis of real-time monitoring data, setting warning thresholds, and triggering warnings when monitoring values ​​exceed the warning thresholds;

[0023] S244: Based on the warning information, determine whether the welding path needs to be changed. If it needs to be changed, select a new optimal path for switching based on the optional path group and the comprehensive score.

[0024] Furthermore, the warning threshold is pre-set based on historical experimental data, and different monitoring values ​​are set for changes in the monitoring data. Different changes have different impacts. The greater the impact, the greater the monitoring value. The monitoring value that changes in real time is compared with the warning threshold.

[0025] Furthermore, step S240 also includes:

[0026] S245: Setting a maximum number of adjustments and monitoring the number of path adjustments in real time. If the maximum number of adjustments is reached, further determining the subsequent unwelded condition and generating a demand value. If the demand value is greater than a demand threshold, path adjustment is performed; if the demand value is not greater than the demand threshold, no path adjustment is performed.

[0027] The demand value is set according to the impact of the current welding path on the subsequent unwelded part. The greater the impact, the greater the demand.

[0028] Furthermore, step S300 also includes:

[0029] S310: Selecting actual path planning efficiency values ​​under multiple paths, matching the new groove shape and cross-section information with the data in the database to generate a matching value;

[0030] S320: Setting a matching threshold, selecting welding paths corresponding to groove types with matching values ​​greater than the matching threshold as candidate paths, and performing a comprehensive analysis on the candidate paths;

[0031] S330: Based on the comprehensive analysis results of the candidate paths, the final recommended welding path is selected and the groove type database is updated.

[0032] Furthermore, step S320 also includes:

[0033] S321: Setting analysis and evaluation indicators to generate a comprehensive evaluation value of one for candidate paths with switching nodes; the evaluation indicators include switching time, operation efficiency, and actual operation difficulty; the switching time includes robot repositioning time and welding parameter adjustment time;

[0034] S322: Evaluate the overall time consumption, welding efficiency, and welding quality of the common candidate path to generate a second comprehensive evaluation value;

[0035] S323: Select the highest candidate path as the recommended welding path based on the comprehensive evaluation value 1 and the comprehensive evaluation value 2 respectively.

[0036] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0037] By defining basic extraction condition templates for different types of grooves and dynamically adjusting them according to actual conditions to adapt to specific changes in groove characteristics, the accuracy and efficiency of welding path planning are improved; by analyzing multiple different cross-sectional data, determining a comprehensive welding path, and generating a comprehensive welding path plan that meets multiple different cross-sectional characteristics, it is possible to simultaneously meet the characteristics of multiple different cross-sectional data and improve the overall welding quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 Schematic diagram of the overall process of a welding path planning method based on machine vision in an embodiment of the present invention;

[0039] Figure 2 1 is a flow chart of step S200 in a welding path planning method based on machine vision in an embodiment of the present invention;

[0040] Figure 3 1 is a flow chart of step S240 in a welding path planning method based on machine vision in an embodiment of the present invention;

[0041] Figure 4 3 is a flow chart of step S300 in a welding path planning method based on machine vision in an embodiment of the present invention. DETAILED DESCRIPTION

[0042] To facilitate understanding of the present invention, the present application will be described more comprehensively below with reference to the relevant drawings; the drawings show preferred embodiments of the present invention, but the present invention can be implemented in many different forms and is not limited to the embodiments described herein; on the contrary, the purpose of providing these embodiments is to enable a more thorough and comprehensive understanding of the disclosed content of the present invention.

[0043] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this invention pertains; the terms used herein in the specification of the present invention are for the purpose of describing specific embodiments only and are not intended to limit the present invention; the term "and / or" used herein includes any and all combinations of one or more of the associated listed items.

[0044] Example 1: Figure 1 As shown, a welding path planning method based on machine vision, the method includes:

[0045] S100: Collect and classify sample data of various groove types, establish a groove type database, and define an extraction condition template for each groove type.

[0046] In some embodiments, the sample data includes basic parameters and key parameters, the basic parameters including the bottom width of the groove, which affects the formation of the molten pool and the width of the weld during welding; the thickness of the groove, which determines the heat input and number of welding layers required during welding, and directly affects the welding quality and efficiency; the angle between the right-angled side and the hypotenuse of the groove, which is used to describe the shape of the groove, affects the molten pool flow and weld shape during welding, and is used to design the welding process and select the welding angle; the depth of the groove is the vertical distance from the material surface to the bottom of the groove, which determines the required penetration depth and weld strength during welding; the length of the groove is the total length of the groove along the welding direction, which affects the heat distribution and the continuity of the weld during welding, and is used to design the welding sequence and select the welding speed. The key parameters include the groove contour line, a continuous curve describing the shape and size of the groove. Machine vision technology is used to scan the groove to obtain the actual contour, accurately matching and identifying the contour line of irregular grooves; the groove area, an auxiliary parameter used to verify the shape and size of the groove; the irregularity index, a quantitative indicator used to describe the degree of deviation between the groove shape and the standard shape (such as X-shaped, V-shaped, U-shaped, etc.); and local feature points, representative points of the groove, such as inflection points and extreme points, which serve as auxiliary parameters for determining the exact location of groove irregularities. Based on sample data, statistical analysis is performed on different types of grooves, classified and stored, forming a groove type database, and extraction condition templates are defined for different types of grooves.

[0047] In some embodiments, a set of basic extraction condition templates are defined for each groove type to guide the extraction of key information. When extracting cross-sectional data for groove types with different regular changes, the extraction conditions need to be adjusted to ensure efficiency and accuracy. The extraction conditions are dynamically adjusted to adapt to specific changes in groove features. The extraction condition template includes edge detection thresholds, filtering parameters, and feature extraction algorithms. Different thresholds are set according to the type of groove to accurately identify the edge of the groove; appropriate filtering algorithms and parameters are selected to smooth the image and remove noise interference; according to the shape characteristics of the groove, a suitable algorithm (such as Hough transform, contour extraction, etc.) is selected to extract the coordinate information of key points. For example, the X-shaped groove: the shape of the groove is similar to the letter "X", with two symmetrical bevels and a narrow bottom width, which is suitable for butt welding of thick plates; its key parameters are the bottom width of the groove, the thickness of the groove, and the angle between the right-angled side of the groove and the bevel (usually two equal acute angles); a higher threshold should be selected for edge detection to highlight the straight line features of the bevel. At the same time, considering the possible shadows or reflections at the bottom of the groove, the threshold range can be appropriately adjusted to ensure accurate identification of the bottom edge. The Hough transform algorithm detects the straight line segments of the bevel, and the contour extraction algorithm obtains the overall contour. V-groove: The groove shape is similar to the letter "V", with a sharp bottom and two beveled edges, and is suitable for welding medium and thick plates. Its key parameters are the groove bottom width, groove thickness, and the angle between the right-angled edge and the beveled edge (an acute angle). A medium threshold is selected to balance the recognition effect of the beveled edge and the bottom. Morphological operations are used to refine the edges and improve recognition accuracy. Mean filtering is selected with a small parameter to retain detailed information about the beveled edge and the bottom. The edge detection algorithm obtains the preliminary outline, the corner detection algorithm locates the intersection position, and the Hough transform algorithm detects the straight line segments of the beveled edge. U-shaped groove: The shape of the groove is similar to the letter "U", with a wide and smooth bottom, which is suitable for thin plates or welding that requires a large penetration depth; its key parameters are the groove bottom width, groove thickness, and groove arc radius. A lower threshold is selected to capture the curve characteristics of the bottom. At the same time, in order to avoid interference from bottom shadows or reflections, a dynamic threshold or adaptive threshold algorithm is used to optimize the recognition effect. Median filtering is selected with moderate parameters to smooth the image and remove noise interference. The edge detection algorithm obtains the preliminary contour, the contour fitting algorithm fits the bottom arc curve, and the contour extraction algorithm obtains the overall contour. Three typical groove types are selected above for illustration. The feature extraction algorithms used in the above are all algorithm models well known to people in this field. Use actual groove data for training and adjust specific parameters to obtain an accurate model for extraction.

[0048] In some embodiments, the local feature points can serve as feature nodes to distinguish different sections. When a local feature point is identified, the section data of the local feature point is intercepted. When the next different local feature point is identified, the section data is intercepted again. Obtaining the section data in the middle range can accurately analyze the data differences between different sections.

[0049] S200: Identify the current groove type, select the corresponding extraction condition template, obtain the groove section data, group the groove section data according to different characteristics, establish a three-dimensional model based on multiple groups of groove section data, generate a comprehensive welding path, and generate a welding analysis report based on the actual welding effect.

[0050] S300: According to the welding analysis report, the path planning status of each groove type is obtained, a path planning efficiency value is generated, and the groove type database is updated.

[0051] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages:

[0052] This application defines basic extraction condition templates for different types of grooves, and dynamically adjusts them according to actual conditions to adapt to specific changes in groove characteristics, thereby improving the accuracy and efficiency of welding path planning.

[0053] Embodiment 2: In embodiment 1, basic extraction condition templates are defined for different types of grooves, and a groove type database is set up. This embodiment makes further improvements on the above basis.

[0054] like Figure 2 As shown, step S200 includes:

[0055] S210: Acquire groove cross-section data based on machine vision technology, including a two-dimensional image and relative position information, and pre-process the acquired groove cross-section data to obtain standard cross-section data;

[0056] S220: Setting a grouping strategy to group the standard cross-section data, selecting a reconstruction strategy for each group of cross-section data according to its characteristics, and constructing a three-dimensional model using a surface reconstruction algorithm based on the reconstruction strategy for each group of cross-section data;

[0057] S230: Analyze weld characteristics, material characteristics, and heat-affected zone sensitivity based on a 3D model to determine welding sequence principles;

[0058] In some embodiments, groove cross-sectional data is acquired based on machine vision technology. A high-resolution industrial camera, a light source system, and an image processing and computing unit are used to calibrate the camera and determine its intrinsic and extrinsic parameters. The intrinsic parameters include focal length, optical center, and distortion coefficient, while the extrinsic parameters are the camera's position and posture in the world coordinate system. The camera's position and posture information is recorded for each acquisition, and relative position information is calculated to describe the spatial relationship between different cross-sectional data. A two-dimensional image of the groove cross-sectional area is acquired, including the shape, size, and state information of the groove. The acquired groove cross-sectional data is preprocessed by using a median filter to remove noise from the image, and histogram equalization to enhance image contrast and make the groove contour clearer. An edge detection algorithm is used to extract the groove edge contour. Based on the camera calibration results, the acquired relative position information is calibrated to ensure the accuracy of the position information, and the position information in the camera coordinate system is converted to the world coordinate system.

[0059] In some embodiments, feature information such as shape, size, and surface state is extracted from the preprocessed cross-sectional data, and the extracted feature information is analyzed to find cross-sectional data with similar features. A grouping strategy is formulated based on the analysis results, and the grouping strategy should ensure that the data within the group have similar features.

[0060] The grouping strategy is to group different sections according to the shape similarity and the characteristic nodes between different sections. For the two sections with the highest shape similarity, the trajectory trends of their adjacent nodes are analyzed, and the adjacent nodes with the same trajectory are selected as characteristic nodes to group the groove section data. For each group of section data, a reconstruction strategy is selected according to its characteristics. Based on the reconstruction strategy of each group of sections, a surface reconstruction algorithm is used to construct a three-dimensional model.

[0061] The trajectory trend refers to the welding angle passing through the node when welding between different cross sections. If the welding angles are consistent, the trajectories are judged to be the same; if the welding angles change, the trajectories are judged to be different.

[0062] In some embodiments, a shape matching algorithm is used to calculate the similarity between each section, and the sections are divided into several groups according to the shape similarity. In the section group with the highest similarity, the trajectory trends of adjacent nodes are analyzed, and adjacent nodes with the same trajectory are selected as feature nodes according to the trajectory trend. For sections with low shape similarity but possibly similar trajectory trends, cross-group analysis is performed to confirm whether regrouping is needed. The effectiveness of grouping and feature node selection is verified through actual welding tests. Based on the welding results and feedback, the grouping strategy and feature node selection are optimized, and the shape similarity calculation method and trajectory trend analysis method are continuously optimized to improve the accuracy of grouping and the rationality of feature node selection.

[0063] In some embodiments, the weld characteristics include the number of weld passes and the number of weld layers. When the groove is large, the weld may require several weld layers to form, and each layer is composed of several welds. The number of welds here refers to the number of weld passes, and the number of weld layers refers to the number of layers repeatedly welded on the same weld, usually referring to the number of consecutive welds performed within a single weld. The number of weld layers is mainly determined by factors such as the thickness of the weld structure, the diameter of the welding rod, and the groove size. The weld characteristics also include the location, type (such as butt weld, fillet weld), number, and distribution density of the welds. The material characteristics are based on the different requirements for welding parameters (such as current, voltage, welding speed) and welding sequence of materials of different thicknesses. Thick materials require preheating or layered welding, while thin materials need to avoid overheating and deformation. The heat-affected zone sensitivity is used to analyze the material's sensitivity to the weld heat-affected zone to determine the welding sequence and cooling measures to reduce welding deformation. The welding sequence principle is used to determine the welding sequence, reduce welding deformation, and balance welding stress and deformation. Optimizing the welding sequence can improve welding efficiency while ensuring welding quality.

[0064] S240: generating a preliminary welding path corresponding to each cross-section group according to each group of cross-section data and a welding sequence principle, and merging the preliminary welding paths of each cross-section group to form a comprehensive welding path.

[0065] In some embodiments, the preliminary welding path is generated based on the cross-sectional data of each group and the welding sequence principle, and the preliminary welding paths of each cross-sectional group are merged to form a comprehensive welding path. During the merging process, it is necessary to ensure the continuity between the paths, that is, the end path of the previous cross-sectional group and the start path of the next cross-sectional group can smoothly transition. The preliminary welding path of each cross-sectional group is segmented and identified, and the starting and ending points of each path are marked. The marked characteristic nodes are used as transition nodes to ensure that the angle remains unchanged during the welding process. According to the order and adjacent relationship of the cross-sectional groups, the end path of the previous cross-sectional group is docked with the start path of the next cross-sectional group. When docking, ensure that the spatial position and direction between the path segments are continuous to avoid breakage or dislocation. At the junction of the path segments, a smooth transition process is performed. If there is no transition node, a curve fitting method is used to generate a smooth transition path. All docked and processed path segments are combined to generate a complete comprehensive welding path.

[0066] In some embodiments, the integrated welding path is adjusted and improved to improve welding efficiency and quality. The kinematic constraints of the welding robot, including joint angle limits and end effector speed limits, are analyzed. Based on the constraints, the integrated welding path is adjusted to ensure that the robot can weld accurately and smoothly according to the planned path. Joint space planning and speed planning methods are used to meet the kinematic constraints. The integrated welding path is reviewed to identify and delete unnecessary turning points and pauses. Path clipping and merging methods are used to simplify the path. When simplifying the path, it should be ensured that the welding quality and the smoothness of the robot movement are not affected. The simplified welding path is smoothed to reduce vibration and impact during the welding process. The corresponding welding process parameters are matched according to the characteristics and requirements of the welding path. The optimized welding path is subjected to actual welding tests. The path merging and optimization process is verified based on the test results and feedback. If problems or deficiencies are found, return to the corresponding steps for correction and optimization.

[0067] In this example, machine vision technology was used to capture groove cross-sectional data, preprocess it, and group it to construct an accurate 3D model. A comprehensive welding path generated based on this 3D model ensured path accuracy and continuity. Analyzing the characteristics of the weld, material, and heat-affected zone determined a reasonable welding sequence.

[0068] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages:

[0069] This application analyzes multiple different cross-sectional data, determines a comprehensive welding path, and generates a comprehensive welding path plan that meets multiple different cross-sectional characteristics, thereby achieving the effect of simultaneously meeting multiple different cross-sectional data characteristics and improving the overall welding quality.

[0070] Embodiment 3: In the above embodiment, a plurality of different cross-sectional data are analyzed to determine a comprehensive welding path. This embodiment makes further improvements on the above basis.

[0071] like Figure 3 As shown, step S240 includes:

[0072] S241: Obtain each set of cross-sectional data, extract key feature points respectively, and obtain the influence value of each set of cross-sectional data according to the number of key feature points, compare the influence value with the influence threshold, and mark it if the influence value is greater than the difference threshold, and generate preliminary welding paths according to the marking.

[0073] S242: generating an optional path group by fusing the preliminary welding paths and annotations, setting evaluation indicators to score each optional path, and selecting the path with the highest comprehensive score as the comprehensive welding path.

[0074] In some embodiments, key feature points are extracted based on each set of cross-sectional data, and the key feature points include cross-sectional shape change points, size mutation points, and material thickness change points; the cross-sectional shape change points are located at positions where the cross-sectional geometry changes significantly, such as from a straight line to a curve, or from one curve to another, and are used to assist in determining parameters that need to be adjusted during welding, such as welding angle and speed; the size mutation points represent positions where cross-sectional dimensions (such as width and height) change significantly, affecting heat distribution and weld quality during welding; the material thickness change points are uneven thickness of cross-sectional materials that can affect the penetration depth and heat-affected zone during welding, affecting the strength of the weld. The above-mentioned key feature points can be identified and analyzed using image processing design software for cross-sectional data, or automatically detected and marked by detection algorithms for shape, size, and thickness change points, which will not be elaborated here.

[0075] The influence value is used to evaluate the influence of the key feature point on the welding path. The influence threshold is set according to all current influence values. All influence values ​​are arranged in descending order, and the average value of the top 20% influence values ​​is selected as the influence threshold.

[0076] In some embodiments, an influence value is generated for each group of cross-sectional data based on the number of key feature points, and compared with an influence threshold. If it is greater than the influence threshold, it is marked, indicating that it has a significant impact on the welding path. The marked feature points serve as an important reference factor for generating a preliminary welding path. Preliminary welding paths are generated according to the marked situation. If three are marked, a preliminary welding path will be generated according to the marked situation. The preliminary welding paths and the marks are fused to generate an optional path group. Each optional path is preferentially satisfied with one of the marks. The number of paths in the optional path group is consistent with the number of marks. For example, a groove contains three different cross-sectional groups, two of which are marked, and three preliminary welding paths are generated respectively. When fusion is performed, the welding requirements of one of the marks are first met to generate an optional path. Similarly, the welding requirements of the other mark are met to generate another optional path, that is, two optional paths are generated to form an optional path group. In the above optional path generation process, the quality and requirements of the overall welding still need to be considered. The difference between the two optional paths is that there is a deviation in the welding method and requirements selected when welding at the marked cross-sectional position.

[0077] In some embodiments, evaluation indicators are set to score each optional path, and the evaluation indicators include welding materials, speed, temperature, and specific equipment conditions. The score is estimated based on the specific implementation of each optional path. All evaluation indicators are integrated and standardized to generate a unified score value. The path with the highest comprehensive score is selected as the comprehensive welding path and is used as the preferred welding path.

[0078] S243: Real-time monitoring of key parameters and conditions in the welding process, processing and analysis of real-time monitoring data, setting warning thresholds, and triggering warnings when the monitoring values ​​exceed the warning thresholds.

[0079] S244: Based on the warning information, determine whether the welding path needs to be changed. If it needs to be changed, select a new optimal path for switching based on the optional path group and the comprehensive score.

[0080] In some embodiments, key parameters and conditions during the welding process are monitored in real time using sensors, cameras, and other detection equipment. These include welding current, voltage, welding speed, weld quality, and welding deformation. The real-time monitoring data is processed and analyzed to extract key information for evaluating the suitability of the current welding conditions and path. Based on the real-time monitoring results and early warning mechanisms, the suitability of the current welding path is determined.

[0081] The warning threshold is pre-set based on historical experimental data. Different monitoring values ​​are set for changes in the monitoring data. Different changes have different impacts. The greater the impact, the greater the monitoring value. The monitoring value that changes in real time is compared with the warning threshold.

[0082] In some embodiments, when welding conditions change significantly (e.g., excessive welding deformation, substandard weld quality, etc.), the monitoring value increases. If it exceeds the warning threshold, the welding path needs to be adjusted. Based on the selectable path group and the comprehensive score, a new optimal path is selected for switching.

[0083] S245: Setting a maximum number of adjustments and monitoring the number of path adjustments in real time. If the maximum number of adjustments is reached, further determining the subsequent unwelded condition and generating a demand value. If the demand value is greater than a demand threshold, path adjustment is performed; if the demand value is not greater than the demand threshold, no path adjustment is performed.

[0084] The demand value is set according to the impact of the current welding path on the subsequent unwelded part. The greater the impact, the greater the demand.

[0085] In some embodiments, a maximum number of adjustments is pre-set to ensure welding efficiency and quality. The number of path adjustments is monitored in real time. If the maximum number of adjustments is reached, the subsequent unwelded portion is further determined and a demand value is generated. The demand value is set based on the impact of the current welding path on the subsequent unwelded portion; the greater the impact, the greater the demand value. The potential adverse effects of the current welding path on the subsequent welded portion are assessed, and the demand threshold is pre-set based on historical experimental data. If the demand value exceeds the demand threshold, path adjustment is performed; if the demand value is not greater than the demand threshold, path adjustment is not performed.

[0086] In this embodiment, by extracting key feature points and generating a preliminary welding path, which is then integrated into a set of optional paths for evaluation, the welding path is ensured to more accurately adapt to different cross-sectional data and welding requirements. Real-time monitoring of key parameters and conditions during the welding process and the establishment of an early warning mechanism enable timely adjustment of the welding path when welding conditions change significantly, thus avoiding welding failures and quality issues. The optional paths are scored by setting evaluation indicators, and the path with the highest overall score is selected as the comprehensive welding path, ensuring optimal welding efficiency and quality. Furthermore, the maximum number of adjustments and the impact of subsequent unwelded conditions are taken into account during the path adjustment process, further improving the overall welding effect.

[0087] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages:

[0088] This application achieves the effect of improving the accuracy and adaptability of the welding path by extracting key feature points and generating a preliminary welding path, and then fusing them to generate an optional path group for evaluation; it monitors the key parameters and conditions of the welding process in real time, and sets an early warning mechanism to be able to adjust the welding path in time when the welding conditions change significantly, thereby achieving the effect of dynamically adjusting the optimal path.

[0089] Embodiment 4: In the above embodiment, the effect of dynamically adjusting the optimal path is achieved. This embodiment makes further improvements on the above basis.

[0090] like Figure 4 As shown, step S300 also includes:

[0091] S310: Selecting actual path planning efficiency values ​​under multiple paths, matching the new groove shape and cross-section information with the data in the database to generate a matching value;

[0092] S320: Setting a matching threshold, selecting welding paths corresponding to groove types with matching values ​​greater than the matching threshold as candidate paths, and performing a comprehensive analysis on the candidate paths;

[0093] S330: Based on the comprehensive analysis results of the candidate paths, the final recommended welding path is selected and the groove type database is updated.

[0094] In some embodiments, the actual path planning efficiency value is a comprehensive indicator used to quantitatively evaluate the pros and cons of welding paths under specific groove types. It takes into account multiple factors such as welding time, material consumption, welding quality, etc., and is obtained through a standardized calculation method. When matching the new groove shape with the data in the database, a value is calculated based on factors such as shape similarity, size difference, and material properties. The higher the matching value, the more similar the new groove is to a groove type in the database, and thus the path planning of this type is selected. A matching threshold is pre-set, for example, set to 95%. If the matching value is greater than 95%, the corresponding welding path under the groove type is selected as a candidate path, and then a comprehensive analysis is performed on the candidate path. Based on the comprehensive analysis results of the candidate path, the final recommended welding path is selected and the groove type database is updated.

[0095] In some embodiments, if there is a switching node (i.e., a point where switching between different paths is required) in the candidate path, further analysis is performed, and step S320 further includes:

[0096] S321: Setting analysis and evaluation indicators to generate a comprehensive evaluation value of one for candidate paths with switching nodes; the evaluation indicators include switching time, operation efficiency, and actual operation difficulty; the switching time includes robot repositioning time and welding parameter adjustment time;

[0097] S322: Evaluate the overall time consumption, welding efficiency, and welding quality of the common candidate path to generate a second comprehensive evaluation value;

[0098] S323: Select the highest candidate path as the recommended welding path based on the comprehensive evaluation value 1 and the comprehensive evaluation value 2 respectively.

[0099] In some embodiments, the robot repositioning time refers to the time required for the robot to reposition to the new welding position when switching from one welding node to another. The welding parameter adjustment time is because different welding nodes may require different welding parameters (such as current, voltage, welding speed, etc.), and this part of time is used to adjust the welding equipment to adapt to the new welding conditions. The operating efficiency is the overall speed of the robot when performing the welding task, including the movement speed and welding speed, as well as the proportion of idle time during the task execution. The actual operating difficulty value is used to assess the complexity of the welding path, such as the degree of tortuosity of the path, whether frequent posture changes are required, etc., taking into account the increased difficulty and possible risks of welding. The comprehensive evaluation value 1 is obtained by standardizing each evaluation indicator, converting the value of each indicator to the range of [0, 1], assigning a weight to each evaluation indicator according to the specific requirements of the welding task, setting the weight based on historical experimental data, multiplying the standardized indicator value by the corresponding weight, and then summing them to obtain the comprehensive evaluation value 1.

[0100] In some embodiments, the overall time refers to the total time from the start to the end of welding, including welding time and non-welding time (such as movement and waiting time). The welding efficiency refers to the amount of welding completed per unit time, which is calculated by dividing the welding length or area by the total time. The welding quality is evaluated based on the quality of the finished product after welding, such as the uniformity of the weld, whether there are defects, etc. For the overall time and welding efficiency, their values ​​are used directly; for welding quality, they are quantified by expert scoring or the scores given by quality inspection equipment. Similarly, a weight is assigned to each evaluation indicator according to the importance of the welding task. Each indicator value is multiplied by the corresponding weight and summed to obtain a comprehensive evaluation value of two.

[0101] For candidate paths with switch nodes, the path with the highest comprehensive evaluation value 1 is selected. For common candidate paths, the path with the highest comprehensive evaluation value 2 is selected. The number of recommended welding paths is one to two.

[0102] The technical solutions in the above embodiments of the present application have at least the following technical effects or advantages:

[0103] This application improves the efficiency and accuracy of welding path planning by generating path planning efficiency values ​​and updating the database; by conducting a comprehensive analysis of candidate paths, setting analysis and evaluation indicators and comprehensive evaluation values, and according to different welding task requirements and groove type characteristics, it achieves the effect of flexible selection of the optimal welding path; by updating the groove type database, it achieves the effect of improving welding efficiency and quality.

[0104] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Various modifications and variations are readily apparent to those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A welding path planning method based on machine vision, characterized in that: The method comprises: S100: Collect and classify sample data of various groove types, establish a groove type database, and define an extraction condition template for each groove type; S200: Identify the current groove type, select the corresponding extraction condition template, obtain groove section data, group the groove section data according to different characteristics, establish a three-dimensional model based on multiple groups of groove section data, generate a comprehensive welding path, and generate a welding analysis report based on the actual welding effect; S300: According to the welding analysis report, the path planning status of each groove type is obtained, the path planning efficiency value is generated, and the groove type database is updated; Wherein, step S200 includes: S210: Acquire groove cross-section data based on machine vision technology, including a two-dimensional image and relative position information, and pre-process the acquired groove cross-section data to obtain standard cross-section data; S220: Setting a grouping strategy to group the standard cross-section data, selecting a reconstruction strategy for each group of cross-section data according to its characteristics, and constructing a three-dimensional model using a surface reconstruction algorithm based on the reconstruction strategy for each group of cross-section data; S230: Analyze weld characteristics, material characteristics, and heat-affected zone sensitivity based on a 3D model to determine welding sequence principles; S240: generating a preliminary welding path corresponding to each cross-section group according to each group of cross-section data and a welding sequence principle, and merging the preliminary welding paths of each cross-section group to form a comprehensive welding path; S241: Obtain each set of cross-sectional data, extract key feature points respectively, and obtain the influence value of each set of cross-sectional data according to the number of key feature points. Compare the influence value with the influence threshold. If the influence value is greater than the influence threshold, mark it, and generate preliminary welding paths according to the marking. S242: generating an optional path group by fusing the preliminary welding paths and annotations, setting evaluation indicators to score each optional path, and selecting the path with the highest comprehensive score as the comprehensive welding path.

2. A welding path planning method based on machine vision according to claim 1, characterized in that: The grouping strategy is to group the sections according to the shape similarity and the characteristic nodes between the sections. For the two sections with the highest shape similarity, the trajectory trends of their adjacent nodes are analyzed, and the adjacent nodes with the same trajectory are selected as characteristic nodes to group the groove section data.

3. A welding path planning method based on machine vision as claimed in claim 2, characterized in that: The trajectory trend refers to the welding angle passing through the node when welding between different cross sections. If the welding angles are consistent, the trajectories are judged to be the same; if the welding angles change, the trajectories are judged to be different.

4. The welding path planning method based on machine vision according to claim 1, characterized in that: The influence value is used to evaluate the influence of the key feature point on the welding path. The influence threshold is set based on all current influence values. All influence values ​​are arranged in descending order, and the average value of the top 20% of the influence values ​​is selected as the influence threshold.

5. The welding path planning method based on machine vision according to claim 1, characterized in that: Step S240 also includes: S243: Real-time monitoring of key parameters and conditions during welding, processing and analysis of real-time monitoring data, setting warning thresholds, and triggering warnings when monitoring values ​​exceed the warning thresholds; S244: Based on the warning information, determine whether the welding path needs to be changed. If it needs to be changed, select a new optimal path for switching based on the optional path group and the comprehensive score.

6. A welding path planning method based on machine vision according to claim 5, characterized in that: The warning threshold is pre-set based on historical experimental data. Different monitoring values ​​are set for changes in the monitoring data. Different changes have different impacts. The greater the impact, the greater the monitoring value. The monitoring value that changes in real time is compared with the warning threshold.

7. The welding path planning method based on machine vision according to claim 5, characterized in that: Step S240 includes: S245: Setting a maximum number of adjustments and monitoring the number of path adjustments in real time. If the maximum number of adjustments is reached, further determining the subsequent unwelded condition and generating a demand value. If the demand value is greater than a demand threshold, path adjustment is performed; if the demand value is not greater than the demand threshold, no path adjustment is performed. The demand value is set according to the impact of the current welding path on the subsequent unwelded part. The greater the impact, the greater the demand value.

8. The welding path planning method based on machine vision according to claim 1, characterized in that: Step S300 includes: S310: Selecting actual path planning efficiency values ​​under multiple paths, matching the new groove shape and cross-section information with the data in the database to generate a matching value; S320: Setting a matching threshold, selecting welding paths corresponding to groove types with matching values ​​greater than the matching threshold as candidate paths, and performing a comprehensive analysis on the candidate paths; S330: Based on the comprehensive analysis results of the candidate paths, the final recommended welding path is selected and the groove type database is updated.

9. A welding path planning method based on machine vision according to claim 8, characterized in that: Step S320 includes: S321: Setting analysis and evaluation indicators to generate a comprehensive evaluation value of one for candidate paths with switching nodes; the evaluation indicators include switching time, operation efficiency, and actual operation difficulty; the switching time includes robot repositioning time and welding parameter adjustment time; S322: Evaluate the overall time consumption, welding efficiency, and welding quality of the common candidate path to generate a second comprehensive evaluation value; S323: Select the highest candidate path as the recommended welding path based on the comprehensive evaluation value 1 and the comprehensive evaluation value 2 respectively.

Images