Welding method

Abstract

The application provides a welding method applied to a welding robot, and the method comprises the following steps: determining a workpiece to be welded; determining an image contour of a target workpiece included in the workpiece to be welded; determining a workpiece parameter according to the image contour, the workpiece parameter being a parameter related to the size of the target workpiece; determining a welding strategy and welding parameters according to the workpiece parameter; and welding the workpiece to be welded according to the welding strategy and the welding parameters. Thus, the whole welding process does not need human participation, and problems caused by insufficient experience of welders or mistakes of welders are avoided. Moreover, the determination of the workpiece parameter, the welding parameter and the welding strategy in the whole welding process is all performed by the welding robot, automation of the welding process is realized, production cost is effectively reduced, and user experience is improved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of welding, and particularly relates to a welding method. BACKGROUND

[0002] Welding robots are widely used in engineering fields such as petrochemical industry, shipbuilding, pressure vessels, heavy machinery, etc., which greatly improves production efficiency and reduces the working intensity of workers.

[0003] In the process of welding by using a welding robot, it is necessary to ensure the rationality and consistency of parameters such as a welding gap amount, a root face amount, a groove angle, and a stand plate inclination angle. However, in the pre-processing and assembly links of medium-thickness plate welding, various errors inevitably occur, it is difficult to ensure the rationality and consistency of these parameters, and the welding working condition state is complex and changeable. Therefore, the existing welding robots can generally only be used in the automobile industry and the welding scene of simple parts where the cost of special fixtures can be diluted, for the scene where the welding working condition state changes complex, a welder still welds according to experience, the welding efficiency is low, and if the welder is insufficient in experience, material waste will be caused, the generation cost is increased, and the forming quality of the weld cannot be guaranteed. SUMMARY

[0004] The purpose of the present application is to solve the problem in the prior art that welding robots can generally only be used in the automobile industry and the welding scene of simple parts where the cost of special fixtures can be diluted, for the scene where the welding working condition state changes complex, a welder still welds according to experience, the welding efficiency is low, and if the welder is insufficient in experience, material waste will be caused, the generation cost is increased, and the forming quality of the weld cannot be guaranteed.

[0005] To solve the above problems, an embodiment of the present application discloses a welding method applied to a welding robot, the method comprising: determining a workpiece to be welded; determining an image contour of a target workpiece included in the workpiece to be welded; determining a workpiece parameter according to the image contour; determining a welding strategy and a welding parameter according to the workpiece parameter; and welding the workpiece to be welded according to the welding strategy and the welding parameter.

[0006] The workpiece parameter is a parameter related to the size of the target workpiece, such as a welding gap amount, a groove angle, a root face amount, and an inclination angle of a stand plate, etc. Of course, it can also be other parameters related to the size of the target workpiece.

[0007] According to the technical scheme, in the welding process, first, the workpiece to be welded is determined, and the image contour of the target workpiece included in the workpiece to be welded is determined, the workpiece parameter corresponding to the target workpiece is determined according to the image contour, the welding strategy and the welding parameter are determined according to the workpiece parameter, and finally, the workpiece to be welded is welded according to the welding strategy and the welding parameter. Therefore, the various parameters required for welding and the welding strategy in the welding process can be accurately determined according to the image contour of the target workpiece included in the workpiece to be welded, and the welding is performed according to the obtained welding strategy and welding parameter, thereby improving the welding precision. Moreover, the entire welding process does not require human intervention, thereby avoiding problems caused by insufficient experience of welders or mistakes of welders. In addition, the determination of the workpiece parameter, the welding parameter and the welding strategy in the entire welding process is performed by the welding robot, thereby realizing the automation of the welding process, improving the welding efficiency, effectively reducing the production cost and increasing the user experience.

[0008] According to another specific embodiment of the present application, the welding method disclosed by the embodiment of the present application comprises the following steps: determining the image contour of the target workpiece included in the workpiece to be welded, comprising: performing laser scanning processing on the target workpiece included in the workpiece to be welded to obtain an initial image; inputting the initial image into a first model to perform semantic segmentation processing and obtain a second image; performing gray processing on the second image to obtain a third image; performing binary processing on the third image to obtain a fourth image; and performing edge extraction processing on the fourth image to obtain the image contour.

[0009] According to the technical scheme, in the determination of the image contour of the target workpiece included in the workpiece to be welded, the initial image contour is subjected to semantic segmentation processing, gray processing and binary processing, so that the noise in the initial image can be accurately and quickly eliminated, and the initial image can be reduced in dimension to reduce the calculation amount of subsequent feature parameter extraction. The semantic segmentation processing of the initial image can eliminate all the noise in the initial image and only retain the selected region of interest. The binary processing of the third image can achieve the effect of reducing the dimension to reduce the data processing amount and highlight the target contour of the image, thereby reducing the difficulty of contour edge detection and obtaining a more accurate image contour, so as to ensure the comprehensive and accurate acquisition of the workpiece parameter, the welding strategy and the welding parameter, thereby improving the welding precision.

[0010] According to another specific embodiment of the present application, the welding method disclosed by the embodiment of the present application comprises the following steps: determining the image contour of the target workpiece included in the workpiece to be welded, comprising: according to the image contour, determining a plurality of feature points, the plurality of feature points being points on the image contour related to the to-be-welded position of the target workpiece; and determining the workpiece parameter according to the plurality of feature points.

[0011] According to the technical scheme, the workpiece parameters can be determined only according to the plurality of feature points on the image contour and related to the to-be-welded position of the target workpiece, the determination process of the workpiece parameters is simple, the determined workpiece parameters are more accurate and reasonable, and thus the welding precision is improved.

[0012] According to another specific embodiment of the present application, the welding method disclosed by the embodiment of the present application, the workpiece parameters include a welding gap amount, a plurality of feature points are determined according to the image contour, and the workpiece parameters are determined according to the plurality of feature points, including: a plurality of first feature points are determined according to the image contour, the plurality of first feature points are points on the image contour and related to the inflection point at the to-be-welded position of the target workpiece; and the welding gap amount is determined according to the plurality of first feature points.

[0013] According to the technical scheme, the workpiece parameters can be determined only according to the plurality of feature points on the image contour and related to the to-be-welded position of the target workpiece, the determination process of the workpiece parameters is simple, the determined workpiece parameters are more accurate and reasonable, and thus the welding precision is improved.

[0014] According to another specific embodiment of the present application, the welding method disclosed by the embodiment of the present application, the workpiece parameters include a welding gap amount, a plurality of feature points are determined according to the image contour, and the workpiece parameters are determined according to the plurality of feature points, including: a plurality of first feature points are determined according to the image contour, the plurality of first feature points are points on the image contour and related to the inflection point at the to-be-welded position of the target workpiece; and the welding gap amount is determined according to the plurality of first feature points.

[0015] According to the technical scheme, the workpiece parameters can be determined only according to the plurality of feature points on the image contour and related to the to-be-welded position of the target workpiece, the determination process of the workpiece parameters is simple, the determined workpiece parameters are more accurate and reasonable, and thus the welding precision is improved.

[0016] According to another specific embodiment of the present application, the welding method disclosed by the embodiment of the present application, the workpiece parameters include a welding gap amount, a plurality of feature points are determined according to the image contour, and the workpiece parameters are determined according to the plurality of feature points, including: a plurality of first feature points are determined according to the image contour, the plurality of first feature points are points on the image contour and related to the inflection point at the to-be-welded position of the target workpiece; and the welding gap amount is determined according to the plurality of first feature points.

[0017] According to the technical scheme, the third feature points on the image contour and related to the horizontal plane and the vertical plane at the to-be-welded position of the target workpiece are used to determine the upright plate inclination angle, the determination process of the upright plate inclination angle is simple, the determined upright plate inclination angle is more reasonable and accurate, and thus the welding precision is improved.

[0018] According to another specific embodiment of the present application, the welding method disclosed by the embodiment of the present application includes a welding strategy, the welding strategy includes a welding gun path and a welding gun posture, and the welding strategy is determined according to workpiece parameters, including: determining the welding gun path according to a welding gap amount; and determining the welding gun posture according to the welding gap amount, a groove angle and an upright plate inclination angle.

[0019] According to the technical scheme, the welding gun path determined according to the welding gap amount is more reasonable and accurate, and the welding gun posture determined according to the welding gap amount, the groove angle and the upright plate inclination angle is more reasonable and accurate, and thus the welding precision is improved.

[0020] According to another specific embodiment of the present application, the welding method disclosed by the embodiment of the present application includes determining the welding gun path according to a welding gap amount, including: determining first information according to the welding gap amount, the first information including information about whether the welding gun needs to swing and whether multiple-point spot welding is needed; and determining the welding gun path according to the first information.

[0021] According to the technical scheme, the welding gap amount determines the welding gun path during the welding process, such as whether the welding gun needs to swing and whether multiple-point spot welding is needed, and the welding gun path determined according to the welding gap amount can meet the actual welding requirements, and thus the welding precision is improved.

[0022] According to another specific embodiment of the present application, the welding method disclosed by the embodiment of the present application includes determining whether the welding gun needs to swing and whether multiple-point spot welding is needed according to the following formula:

[0023]

[0024] In the formula, S is the first information, b is the welding gap amount, zero means that the welding gun does not need to swing and multiple-point spot welding is not needed, one means that the welding gun needs to swing but multiple-point spot welding is not needed, and two means that the welding gun needs to swing and multiple-point spot welding is needed.

[0025] According to another specific embodiment of the present application, the welding method disclosed by the embodiment of the present application includes a welding strategy, and the welding method is determined by the following way: determining a weld type according to an image contour; and determining the welding method according to the weld type.

[0026] The technical scheme is adopted, the inclination of the welding gun has obvious influence on the forming of the weld, the process of determining the type of the weld is simple and accurate according to the image contour, the welding method is determined according to the type of the weld, the forming quality of the obtained weld is high, and the welding precision and efficiency are improved.

[0027] According to another specific embodiment of the present application, the welding method disclosed by the embodiment of the present application, the type of the weld includes flat corner weld and vertical corner weld, and the welding method is determined according to the type of the weld, including: if the type of the weld is flat corner weld, the welding method is determined as backward inclined welding method; and if the type of the weld is vertical corner weld, the welding method is determined as vertical upward welding method.

[0028] The technical scheme is adopted, the surface forming of the weld by the backward inclined method is more uniform than the surface forming of the weld by the forward inclined method, the fish scale is more delicate and regular, and the size of the weld leg is slightly larger. Therefore, if the type of the weld is flat corner weld, the welding method is determined as backward inclined welding method; and if the type of the weld is vertical corner weld, the welding method is determined as vertical upward welding method, so that the obtained weld meets the requirements, and the welding precision is improved.

[0029] According to another specific embodiment of the present application, the welding method disclosed by the embodiment of the present application, the welding parameter is determined according to the workpiece parameter, including: inputting the workpiece parameter into a preset prediction model for processing to determine the welding parameter.

[0030] The technical scheme is adopted, the prediction model has the abilities of robustness and anti-interference, the workpiece parameter is input into the preset prediction model for processing, the optimal welding parameter required by the current welding condition can be determined, the determination process is accurate and fast, and the welding efficiency is improved.

[0031] According to another specific embodiment of the present application, the welding method disclosed by the embodiment of the present application, the initial image is input into the first model for semantic segmentation processing to obtain a second image, including: inputting the initial image into a preset classification model for processing to obtain a processing result; and if it is determined that the welding condition state corresponding to the workpiece to be welded is reasonable according to the processing result, the initial image is input into the first model for semantic segmentation processing to obtain the second image.

[0032] The technical scheme is adopted, the initial image is input into the preset classification model for processing to obtain a processing result; and when it is determined that the welding condition state corresponding to the workpiece to be welded is reasonable according to the processing result, the subsequent semantic segmentation processing is continued. The situation that the welding condition state is unreasonable is excluded early, and the welding efficiency is improved.

[0033] According to another specific embodiment of the present application, the welding method disclosed by the embodiment of the present application further comprises: generating first reminding information if it is determined that the welding condition state corresponding to the workpiece to be welded is unreasonable according to the processing result; and sending the first reminding information to a user device, so that the user reassembles the workpiece to be welded according to the first reminding information.

[0034] By using the above technical solution, the first reminding information is generated when the welding condition state is unreasonable, so that the user can be reminded to reassemble the workpiece to be welded in time, and subsequent steps can be performed, thereby improving the welding efficiency.

[0035] According to another specific embodiment of the present application, the welding method disclosed by the embodiment of the present application comprises welding current, welding voltage, welding speed and wire feeding speed.

[0036] The specific embodiment of the present application further provides an electronic device, which comprises a memory for storing a computer program, the computer program comprising program instructions; and a processor for executing the program instructions to enable the electronic device to perform the welding method provided by any one of the possible embodiments.

[0037] The specific embodiment of the present application further provides a computer readable storage medium, which stores a computer program, the computer program comprising program instructions, and the program instructions are run by an electronic device to enable the electronic device to perform the welding method provided by any one of the possible embodiments.

[0038] The specific embodiment of the present application further provides a computer program product, which comprises computer program / instructions, and the computer program / instructions are executed by a processor to implement the welding method provided by any one of the possible embodiments.

[0039] The present application has the following beneficial effects:

[0040] The welding method provided by the present application, in the welding process, first determines the workpiece to be welded, determines the image contour of the target workpiece included in the workpiece to be welded, according to the image contour, the workpiece parameters corresponding to the target workpiece can be determined, and according to the workpiece parameters, the welding strategy and the welding parameters are determined, and finally the workpiece to be welded is welded according to the welding strategy and the welding parameters. Therefore, according to the image contour of the target workpiece included in the workpiece to be welded, the various parameters required for welding and the welding strategy in the welding process can be accurately determined, and the welding precision is improved. And the whole welding process does not need human intervention, avoiding the problems caused by the lack of experience of welders or the mistakes of welders. And the determination of the workpiece parameters, the welding parameters and the welding strategy in the whole welding process is carried out by the welding robot, realizing the automation of the welding process, improving the welding efficiency, effectively reducing the production cost and increasing the user experience. BRIEF DESCRIPTION OF DRAWINGS

[0041] Figure 1 It is a flowchart of a welding method according to a specific embodiment of the present application;

[0042] Figure 2 It is a flowchart of determining the image contour of the target workpiece included in the workpiece to be welded according to a specific embodiment of the present application;

[0043] Figure 3 It is a flowchart of inputting the initial image into the first model and performing semantic segmentation processing to obtain the second image according to a specific embodiment of the present application;

[0044] Figure 4 It is a flowchart of another welding method according to a specific embodiment of the present application;

[0045] Figure 5 It is a flowchart of determining the workpiece parameters according to the image contour according to a specific embodiment of the present application;

[0046] Figure 6 It is a flowchart of determining the welding gap according to the image contour according to a specific embodiment of the present application;

[0047] Figure 7 It is a flowchart of determining the bevel angle and the root face according to the image contour according to a specific embodiment of the present application;

[0048] Figure 8 It is a flowchart of determining the stand plate inclination angle according to the image contour according to a specific embodiment of the present application;

[0049] Figure 9 is a structural schematic diagram of a BP neural network according to an embodiment of the present application;

[0050] Figure 10 is a schematic diagram of an initial image, a second image, a third image and a fourth image according to an embodiment of the present application;

[0051] Figures 11a-11c is a schematic diagram of image contours of a first vertical plate, a second vertical plate and a bottom plate in a first direction, a second direction and a third direction according to an embodiment of the present application;

[0052] Figures 12a-12b is a schematic diagram of calculating a welding gap according to an embodiment of the present application;

[0053] Figure 13 is a schematic diagram of an extraction algorithm of a bevel angle and a land amount according to an embodiment of the present application;

[0054] Figure 14 is a schematic diagram of a bevel angle and a land amount calculation principle according to an embodiment of the present application;

[0055] FIG. 15 is a schematic diagram of calculating a bevel angle and a land amount according to an embodiment of the present application;

[0056] FIG. 16 is another schematic diagram of calculating a bevel angle and a land amount according to an embodiment of the present application;

[0057] FIG. 17 is another schematic diagram of calculating a bevel angle and a land amount according to an embodiment of the present application;

[0058] FIG. 18 is another schematic diagram of calculating a vertical plate inclination angle according to an embodiment of the present application;

[0059] Figure 19 is a schematic diagram of a welding path without welding gun swinging according to an embodiment of the present application;

[0060] Figures 20a-20b is a schematic diagram of a triangular welding gun swinging mode and parameters according to an embodiment of the present application;

[0061] Figures 21a-21b is a schematic diagram of a welding gun posture according to an embodiment of the present application;

[0062] Figure 22is a welding gun posture β1 decision schematic provided according to an embodiment of the present application;

[0063] Figure 23 is a structure block diagram of an electronic device provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0064] In order to make the objects, technical solutions and advantages of the present application clearer, the embodiments of the present application will be further described in detail below with reference to the drawings.

[0065] Welding robots are widely used in petrochemical industry, shipbuilding, pressure vessels, heavy machinery and other engineering fields, greatly improving production efficiency and reducing the working intensity of workers. However, in the pre-processing and assembly links of medium plate welding, various errors inevitably occur, it is difficult to ensure the rationality and consistency of the welding gap amount, the root face amount, the groove angle and the upright plate inclination angle and other parameters, resulting in complex and changeable welding working conditions. At present, the working modes of welding robots mainly include "teaching-reproduction" and "offline programming" two working modes. "Teaching-reproduction" and "offline programming" have many limitations: ① manual teaching is required before welding, which greatly reduces the welding production efficiency; ② high requirements are required for the processing accuracy of the welding parts and the accuracy of the pre-welding assembly; ③ the welding fixture has high requirements, which is far from the flexible treatment of the welder according to different welding working conditions. The above factors result in that welding robots can generally only be used in the automobile industry and the welding scene of simple parts which can dilute the cost of special fixture manufacturing. For the scene with complex changes of welding working conditions, the welder still welds according to experience, the welding efficiency is low, and if the welder is inexperienced, it will cause material waste, increase production cost, and also cannot guarantee the forming quality of the weld.

[0066] The welder can adjust the welding process parameters and determine the welding strategy, welding gun swing mode, welding path and welding gun posture by observing the changes of the welding working conditions, especially the changes of the welding gap amount, the groove angle and the root face amount, and combining the welding experience accumulated through the "cognition-feedback-learning" mode for a long time to obtain a weld with good appearance and qualified quality. Although the experience and skills of the welder are particularly important for welding high-quality welds, the welder will be tired, distracted and other situations after a long time of welding work, which will cause the welding quality to decrease or even become unqualified. In addition, the experience and skills of the welder need to be trained for several years to form.

[0067] To solve the problem that the welding robot in the prior art can only be used in the automobile industry and the welding scene of simple parts, and the welding robot cannot be used in the scene with complex welding conditions, and the welding is performed according to experience, the welding efficiency is low, the material is wasted if the experience of the welder is insufficient, the production cost is increased, and the forming quality of the weld cannot be guaranteed, the welding method provided by the present application can accurately determine various parameters required for welding and a welding strategy in a welding process according to an image contour of a target workpiece included in a workpiece to be welded, and the workpiece is welded according to the welding strategy and the welding parameters, so that the automation of the welding process is realized, the production cost is effectively reduced, and the user experience is improved.

[0068] Next, reference is made to the accompanying drawings Figures 1-22 The specific implementation process and advantages of the welding method provided by the present application are described in detail.

[0069] In an embodiment of the present application, the welding method provided by the present application is applied to a welding robot, as shown in the figure, and the welding method comprises the following steps: Figure 1

[0070] S100: Determine a workpiece to be welded.

[0071] S200: Determine an image contour of a target workpiece included in the workpiece to be welded.

[0072] S300: Determine a workpiece parameter according to the image contour.

[0073] S400: Determine a welding strategy and a welding parameter according to the workpiece parameter.

[0074] S500: Weld the workpiece to be welded according to the welding strategy and the welding parameter.

[0075] The workpiece to be welded refers to a workpiece that needs to be welded together, and can include two sub-workpieces, three sub-workpieces, and more sub-workpieces. Further, the target workpiece can be one, two, or all of the sub-workpieces of the workpiece to be welded, and can be set as required. The image contour of the target workpiece can also include the image contours of multiple target workpieces at multiple angles.

[0076] The workpiece parameter is a parameter related to the size of the target workpiece, and specifically can be a parameter related to the size of the welding position of the target workpiece participating in welding, such as at least one of a welding gap amount, a bevel angle, a root face amount, and a stand plate inclination angle.

[0077] ​The welding parameters, i.e., welding process parameters output by the welding robot during the welding process, can include at least one of a welding current, a welding voltage, a welding speed, and a wire feeding speed. In an embodiment of the present application, the welding parameters include the welding current, the welding voltage, the welding speed, and the wire feeding speed.

[0078] With the above technical solution, in the welding process, the workpiece to be welded is first determined, the image contour of the target workpiece included in the workpiece to be welded is determined, the workpiece parameters corresponding to the target workpiece are determined according to the image contour, the welding strategy and the welding parameters are determined according to the workpiece parameters, and finally the workpiece to be welded is welded according to the welding strategy and the welding parameters. Thus, the various parameters required for welding and the welding strategy in the welding process can be accurately determined according to the image contour of the target workpiece included in the workpiece to be welded, and the welding piece is welded according to the obtained welding strategy and welding parameters, thereby improving the welding precision. Moreover, the entire welding process does not require human intervention, thereby avoiding problems caused by insufficient experience of welders or mistakes of welders. Furthermore, the determination of the workpiece parameters, the welding parameters, and the welding strategy in the entire welding process is performed by the welding robot, thereby realizing automation of the welding process, improving the welding efficiency, effectively reducing the production cost, and increasing the user experience.

[0079] In an embodiment of the present application, as shown in Figure 2 the image contour of the target workpiece included in the workpiece to be welded is determined, including the following steps:

[0080] S201: performing laser scanning processing on the target workpiece included in the workpiece to be welded to obtain an initial image of the target workpiece.

[0081] S202: inputting the initial image into a first model to perform semantic segmentation processing to obtain a second image.

[0082] S203: performing gray scale processing on the second image to obtain a third image.

[0083] S204: performing binaryzation processing on the third image to obtain a fourth image.

[0084] S205: performing edge extraction on the fourth image to obtain an image contour.

[0085] The initial image of the target workpiece can include original images of the top, bottom and interior of the target workpiece to be welded. The first model can specifically perform semantic segmentation processing on the selected attention region of the initial image by using DeepLab V3+, so as to eliminate all noises in the initial image and only retain the selected attention region, thereby obtaining a second image. When performing grayscale processing on the second image, the cv2.COLOR_BGR2GRAY function can be used to convert the second image (RGB image) segmented by the semantic segmentation into a grayscale image, thereby obtaining a third image. Then, the third image is binarized, and the cv2.threshold() function can be used to binarize the third image, so as to obtain a binarized image, that is, a fourth image. It should be noted that the first model can also be other semantic segmentation processing models, and is not limited to the DeepLab V3+ model. The grayscale processing on the second image and the binarization of the third image can also use other existing processing models or processing functions, which can be selected by a person skilled in the art as needed.

[0086] Finally, the fourth image is edge extracted to obtain an image contour. Specifically, the edge detection algorithm for edge extraction can be any one of Sobel, Scharr, Laplacian, Canny and the like, and in an embodiment of the present application, the Canny edge detection algorithm is selected to extract the fourth image contour to obtain the image contour.

[0087] By using the above technical solutions, when the image contour of the target workpiece included in the workpiece to be welded is determined, the initial image contour is subjected to semantic segmentation processing, grayscale processing and binarization processing, so that the noises in the initial image can be accurately and quickly eliminated, and the initial image can be reduced in dimension to reduce the calculation amount of subsequent feature parameter extraction. The semantic segmentation processing on the initial image can eliminate all noises in the initial image and only retain the selected attention region. The binarization processing on the third image can achieve the effect of reducing dimension to reduce the data processing amount and highlight the target contour of the image, thereby reducing the difficulty of contour edge detection and obtaining a more accurate image contour, so as to ensure that the workpiece parameters, the welding strategy and the welding parameters are comprehensively and accurately acquired, thereby improving the welding precision.

[0088] In an embodiment of the present application, as shown in FIG. 2A, the initial image is input into the first model to perform semantic segmentation processing, thereby obtaining a second image, including the following steps: Figure 3

[0089] S2021: The initial image is input into a preset classification model for processing, thereby obtaining a processing result.

[0090] ​S2022: If it is determined according to the processing result that the welding working condition state corresponding to the workpiece to be welded is reasonable, the initial image is input into the first model for semantic segmentation processing to obtain a second image.

[0091] First, the rationality of the welding working condition state corresponding to the workpiece to be welded is qualitatively judged by the classification model. If it meets the welding requirements, the welding working condition state feature parameters (i.e., image contours) are extracted by taking the semantic segmentation network as the core. If it does not meet the welding requirements, the assembly is re-performed. Specifically, the initial image is input into a preset classification model for processing. The classification model can judge the assembly condition of the output initial image. For example, it is judged from the initial image that the gap between the two workpieces to be welded is too large to be welded, or the angle of the two workpieces to be welded is unreasonable, which also leads to the failure of welding, and then an unreasonable processing result is output. If it is preliminarily determined that the assembly of the workpiece to be welded meets the requirements, a reasonable processing result is output. The classification model can be a MobileNetV2 classification network. After the initial image is input into the preset classification model for processing to obtain a processing result, when it is determined according to the processing result that the welding working condition state corresponding to the workpiece to be welded is reasonable, the subsequent semantic segmentation processing is continued. The unreasonable welding working condition state is excluded early, and the welding efficiency is improved. The classification model can also be other types of classification networks, which can be selected by those skilled in the art as needed.

[0092] In an embodiment of the present application, as shown in Figure 4 the method further comprises the following steps:

[0093] S2023: If it is determined according to the processing result that the welding working condition state corresponding to the workpiece to be welded is unreasonable, a first reminder information is generated.

[0094] S2024: The first reminder information is sent to a user device, so that the user re-assembles the workpiece to be welded according to the first reminder information.

[0095] When the welding working condition state is unreasonable, the first reminder information is generated and sent to the user device, which can timely remind the user to re-assemble the workpiece to be welded, so that the subsequent steps are performed, and the welding efficiency is improved. The user device can be a mobile phone, a computer, etc. in communication connection with the welding robot, and the first reminder information can be voice broadcast information or warning light flashing information, etc.

[0096] In an embodiment of the present application, as shown in Figure 5 the workpiece parameters are determined according to the image contours, comprising the following steps:

[0097] S301: A plurality of feature points are determined according to the image contours, and the plurality of feature points are points on the image contours related to the welding position of the target workpiece.

[0098] S302: determining the workpiece parameter according to the plurality of feature points.

[0099] The plurality of feature points are points on the image contour related to the to-be-welded position of the target workpiece, which can be inflection points at the to-be-welded position, points on a groove plane and a root face plane at the to-be-welded position, or points on a horizontal plane and a vertical plane at the to-be-welded position. The workpiece parameter can be determined only according to the plurality of feature points on the image contour, the determination process of the workpiece parameter is simple, and the determined workpiece parameter is more accurate and reasonable, thereby improving the welding precision.

[0100] In an embodiment of the present application, the workpiece parameter includes a welding gap amount, as shown in FIG. 2. Figure 6 According to the image contour, the plurality of feature points are determined, and the workpiece parameter is determined according to the plurality of feature points, including the following steps:

[0101] S3011: determining a plurality of first feature points on the image contour, the plurality of first feature points being points on the image contour related to inflection points at the to-be-welded position of the target workpiece.

[0102] S3012: determining the welding gap amount according to the plurality of first feature points.

[0103] The welding gap amount can be determined only according to the plurality of first feature points on the image contour related to the inflection points at the to-be-welded position of the target workpiece, the determination process of the welding gap amount is simple, and the determined welding gap amount is more reasonable and accurate, thereby improving the welding precision.

[0104] In an embodiment of the present application, the workpiece parameter includes a groove angle and a root face amount, as shown in FIG. 3. Figure 7 According to the image contour, the plurality of feature points are determined, and the workpiece parameter is determined according to the plurality of feature points, including the following steps:

[0105] S3011': determining a plurality of second feature points on the image contour, the plurality of second feature points being points on the image contour related to a groove plane and a root face plane at the to-be-welded position of the target workpiece.

[0106] S3012': determining the groove angle and the root face amount according to the plurality of second feature points.

[0107] The groove angle and the root face amount can be determined only according to the plurality of second feature points on the image contour related to the groove plane and the root face plane at the to-be-welded position of the target workpiece, the determination process of the groove angle and the root face amount is simple, and the determined groove angle and root face amount are more reasonable and accurate, thereby improving the welding precision.

[0108] In an embodiment of the present application, the workpiece parameter comprises a stand plate inclination angle, such as Figure 8 As shown, according to the image contour, a plurality of feature points are determined, and according to the plurality of feature points, the workpiece parameter is determined, comprising the following steps:

[0109] S3011'': According to the image contour, a plurality of third feature points are determined, the plurality of third feature points being points on the image contour related to the horizontal plane and the vertical plane at the position to be welded of the target workpiece.

[0110] S3012'': According to the plurality of third feature points, the stand plate inclination angle is determined.

[0111] Only according to the plurality of third feature points on the image contour related to the horizontal plane and the vertical plane at the position to be welded of the target workpiece, the stand plate inclination angle can be determined, the determination process of the stand plate inclination angle is simple, and the determined stand plate inclination angle is more reasonable and accurate, thereby improving the welding precision.

[0112] In an embodiment of the present application, the welding strategy comprises a welding gun path and a welding gun posture, and according to the workpiece parameter, the welding strategy is determined, comprising: determining the welding gun path according to the welding gap amount; and determining the welding gun posture according to the welding gap amount, the bevel angle and the stand plate inclination angle.

[0113] The welding gap amount represents the distance between two workpieces, when the value of the welding gap amount is relatively small, the welding gun does not need to swing to realize welding. When the value of the welding gap amount is relatively large, the welding gun needs to swing and multi-point spot welding to completely weld the workpiece to be welded. The welding gun path determined according to the welding gap amount is more reasonable and accurate.

[0114] Further, the posture of the welding gun has a great influence on the weld forming quality when welding the weld, and unreasonable welding gun posture is easy to produce undercut and incomplete penetration and other defects and lead to small weld leg size and thus affect the weld strength. According to the spatial position of the welding gun, the welding posture of the welding gun in the welding process can be determined. The welding gun posture determined according to the welding gap amount, the bevel angle and the stand plate inclination angle is more reasonable and accurate, thereby improving the welding precision.

[0115] In an embodiment of the present application, the welding gun path is determined according to the welding gap amount, comprising: determining first information according to the welding gap amount, the first information comprising information whether the welding gun needs to swing and whether multi-point spot welding is needed; and determining the welding gun path according to the first information.

[0116] The size of the welding gap amount determines the welding gun path in the welding process of the welding gun, such as whether the welding gun needs to swing and whether multi-point spot welding is needed, and the welding gun path determined according to the welding gap amount can be more in line with the actual welding requirements, thereby improving the welding precision.

[0117] In an embodiment of the present application, whether the welding gun needs to swing and whether the multi-point spot welding is needed are determined according to the following formula:

[0118]

[0119] Wherein, S is the first information, b is the welding gap, zero is that the welding gun does not need to swing and the multi-point spot welding is not needed, one is that the welding gun needs to swing but the multi-point spot welding is not needed, and two is that the welding gun needs to swing and the multi-point spot welding is needed.

[0120] That is, when the welding gap satisfies 1 < b < 2 mm, the welding gun does not need to swing and the multi-point spot welding is not needed in the welding process; when the welding gap satisfies 2 < b < 3 mm, the welding gun needs to swing but the multi-point spot welding is not needed in the welding process; and when the welding gap satisfies 3 < b < 4 mm, the welding gun needs to swing and the multi-point spot welding is needed in the welding process.

[0121] In an embodiment of the present application, the welding strategy further includes a welding method, and the welding method is determined by the following manner: determining the weld type according to the image profile; and determining the welding method according to the weld type.

[0122] The inclination of the welding gun has a significant influence on the formation of the weld, the process of determining the weld type according to the image profile is simple and accurate, the welding method is determined according to the weld type, the forming quality of the obtained weld is high, and thus the welding precision and efficiency are improved.

[0123] In an embodiment of the present application, the weld type includes a flat corner weld and a vertical corner weld, and the welding method is determined according to the weld type, including: if the weld type is the flat corner weld, the welding method is determined as the backward inclined welding method; and if the weld type is the vertical corner weld, the welding method is determined as the vertical upward welding method.

[0124] In an embodiment of the present application, the welding parameter is determined according to the workpiece parameter, including: inputting the workpiece parameter into a preset prediction model for processing to determine the welding parameter.

[0125] The welding process parameters (such as welding current, welding voltage, welding speed, wire feeding speed, etc.) have a huge influence on the weld formation quality, and thus the best welding process parameters under different working conditions need to be intelligently determined for the intelligent welding. At present, there are two methods for predicting the welding process parameters, one is the polynomial response surface method, and the other is the machine learning method. The response surface method (RSM) is easy to estimate and apply, but the response surface method is not conducive to processing complex problems. Therefore, the prediction model is used to predict the welding process parameters in the present application.

[0126] like Figure 9 The diagram shows the principle of the prediction model (backpropagation neural network, BP) used in this invention, which includes two stages: forward propagation and backward propagation. In the forward propagation stage, the input data undergoes weighted processing and activation functions at each layer to obtain the final output. In the backward propagation stage, the error between the output and the true value is calculated, and the weights of each neuron are adjusted layer by layer to continuously reduce the error until convergence or a predetermined number of iterations is reached. At this point, for newly input similar samples, the network automatically outputs the result information with the minimum error.

[0127] To ensure weld formation quality under different welding conditions, workpiece parameters that express the welding condition, such as welding gap, bevel angle, blunt edge, and tilt angle of the vertical plate (obtained from image contours), are used as input vectors to a BP neural network. Process parameters (i.e., welding parameters) such as welding current, welding voltage, welding speed, and wire feed speed are used as the network's predicted output vectors. The BP neural network possesses robustness and anti-interference capabilities. By inputting the workpiece parameters into a pre-defined prediction model, it can determine the optimal welding parameters required for the current welding condition accurately and quickly.

[0128] In one specific embodiment of the present invention, the workpiece to be welded is a hydraulic support box structure, and the target workpiece of the hydraulic support box structure includes a base plate, a first upright plate and a second upright plate. The specific application scenario of using a welding robot to weld the first upright plate, the second upright plate and the base plate is described in detail. The implementation process of the welding method provided by the present invention and the process of determining the rationality of the welding method are explained in detail.

[0129] First, a laser scan of the hydraulic support box structure is performed using a vision sensor to acquire images of the workpiece from multiple directions, such as... Figure 10 The images shown (a), (b), and (c) are examples of initial images. (a) is an image in the first direction, specifically an image of the top of the hydraulic support box structure; (b) is an image in the second direction, specifically an image of the bottom of the hydraulic support box structure; and (c) is an image in the third direction, specifically an image of the interior of the hydraulic support box structure. Further, S1 represents the first upright plate, S2 represents the second upright plate, and B represents the base plate.

[0130] Next, the reasonableness of the welding condition is judged using the MobileNet V2 classification network (as an example of a classification model). Specifically, the reasonableness of the welding condition is judged using the MobileNet V2 classification network (as an example of a classification model).Figure 10 The images in (a), (b), and (c) are input into the MobileNet V2 classification network to obtain classification results, which include reasonable and unreasonable classification results.

[0131] If the classification result is reasonable, the welding condition feature information is extracted; if it does not meet the welding requirements, the assembly is repeated.

[0132] Before obtaining the contour of the selected attention region, the image must first be segmented and preprocessed to accurately and quickly eliminate noise in the original image (i.e., the initial image) and the image is reduced in dimensionality to reduce the computational cost of subsequent feature parameter extraction.

[0133] The extraction of welding condition feature information includes:

[0134] First of all Figure 10 The original images in the first, second, and third directions shown in (a), (b), and (c) can be segmented using the optimized DeepLab V3+ (as an example of the first model) to obtain the selection attention region. Figure 10 The images shown in (a1), (b1), and (c1) (as an example of the second image) remove all noise from the original image, retaining only the selected attention region. At this point, the image obtained from semantic segmentation needs to be binarized to achieve dimensionality reduction, thereby reducing data processing volume and highlighting the target contour of the image, thus reducing the difficulty of contour edge detection. Before binarizing the segmented image, the RGB image obtained from semantic segmentation is first converted to a grayscale image using the cv2.COLOR_BGR2GRAY function, as shown below. Figure 10 (a2)(b2)(c2) (as an example of a third image) are shown. Then, the grayscale image can be processed using the cv2.threshold() function to obtain its binarized image, as shown. Figure 10 (a3)(b3)(c3) (as an example of the fourth image) are shown.

[0135] By comparing the binarized images after the feature information of the welding working condition is extracted by Sobel, Scharr, Laplacian, and Canny edge detection algorithms, it can be shown from the edge detection results that the Laplacian and Canny edge detection algorithms are more excellent than the Sobel and Scharr edge detection algorithms. However, when the welding operation environment is relatively complex, the collected images have more noise, the Laplacian edge detection algorithm is more sensitive to image noise, and the Canny edge detection algorithm has stronger robustness and accuracy in the face of complex environments. Therefore, in the embodiment, the Canny edge detection algorithm is selected to extract the contour of the binarized image after the feature information of the welding working condition is extracted, and the image contour is obtained, as shown in Figures 11a-11c .

[0136] Next, according to the image contour, the workpiece parameters are determined.

[0137] First, according to the image contour, the welding gap b is determined.

[0138] The welding gap has a great influence on the welding strategy, welding path, welding gun posture, and welding process parameters, but it is difficult to guarantee the rationality and consistency of the welding gap when positioning welding is performed, thereby affecting the weld forming quality. Therefore, in order to guarantee the weld forming quality, it is necessary to be able to measure the welding gap on line, thereby being able to provide a basis for the decision of the welding strategy, the welding path, the welding gun posture, and the welding process parameters. The extraction process of the welding gap in the embodiment is shown in Figure 12a and Figure 12b , wherein S1 contour represents the image contour of the first vertical plate, S2 contour represents the image contour of the second vertical plate, and B contour represents the projection contour of the bottom plate.

[0139] As shown in Figure 12a , the coordinates of the first vertical plate inflection point t g on the top image contour are extracted, and the intersection t i between the second vertical plate and the vertical line passing through the point t g is obtained, and then the distance between the point t g and t i (as an example of the first feature point) is calculated, so as to obtain the welding gap of the top of the vertical weld (as an example of the welding gap). As shown in Figure 12b , the coordinates of the second vertical plate inflection points b g1 and b g2 on the bottom image contour are extracted, and the intersections b i1 and b i2 between the bottom plate and the vertical lines passing through the points b g1 and b g2 are obtained, respectively.Then calculate point b g1 With b i1 b g2 With b i2 The distance between (as another example of the first feature point) and the average value can be used to obtain the welding gap on one side of the fillet weld (as another example of the welding gap). When the vision sensor moves to the other side of the weld and acquires an image, this method can be used to solve for the gap on the other side (as another example of the welding gap). Extract the inflection point b of the second vertical plate on the bottom image contour. g3 The coordinates and passing through point b g3 Draw a horizontal line to obtain the intersection point b between the horizontal line and the first vertical board. i3 Then calculate point b g3 With b i3 The distance between (as another example of the first feature point) can be used to obtain the welding gap at the bottom of the fillet weld (as another example of the welding gap); extract the inflection point b of the first vertical plate on the bottom image contour. g4 The coordinates and passing through point b g4 Draw a vertical line to obtain the intersection point b between the line and the base plate. i4 Then calculate point b g4 With b i4 The distance between (as another example of the first feature point) can be used to obtain the welding gap on one side of the fillet weld (as another example of the welding gap). When the vision sensor moves to the other side of the weld and acquires the image contour, the gap on the other side can be obtained using this method.

[0140] Then, based on the image contour, determine the bevel angle α and the blunt edge amount c.

[0141] The bevel angle and the amount of blunt edge have a significant impact on the welding path, welding torch posture, and welding process parameters. Therefore, this embodiment... Figure 13 The extraction algorithm shown enables online calculation (i.e., measurement) of bevel angle and blunt edge amount.

[0142] Depend on Figure 13 As can be seen, the algorithm established in this embodiment determines the bevel angle and the blunt edge amount by utilizing the intersection lines and intersection points between the bevel plane, the blunt edge plane, and the cross-section. For example... Figure 14 As shown, L1 and L3 are the intersection lines formed by the cross section and the bevel plane, and L2 is the intersection line between the cross section and the blunt edge plane. By calculating the angle between L2 and L1 and L3, the bevel angles α1 and α2 inside and outside the K-type bevel can be determined. By calculating the distance between the intersection points c1 and c2 formed by L2 and L1 and L3, the magnitude of the blunt edge can be obtained.

[0143] Next, we will introduce the calculation process for the bevel angle and blunt edge amount corresponding to the vertical fillet weld.

[0144] As shown in Figures 15a-15d , wherein, Figure 15d the images in are respectively from left to right Figure 15a region I, region II in Figure 15b region III, region IV, region V and region VI in Figure 15c region VII in corresponds to the local enlarged view. According to the position and direction of the vision sensor, it can be determined that the points m4, m5, o1, o2 are on the inner side bevel plane G i upper of the K-type groove of the fillet weld; m5, m6, n 10 , n 11 are on the outer side bevel plane G o of the K-type groove of the fillet weld; and m4, n9, n 10 , o1 are on the root face plane T of the K-type groove of the fillet weld. Therefore, by obtaining the pixel coordinates of the feature points and bringing them into the conversion matrix completed by calibration, the coordinates (X w , Y w , Z w ) of the feature points in the welding robot coordinate system can be obtained, and then the plane equations of the planes G i , G o , T in the welding robot coordinate system can be calculated by formula 1. (Wherein, m4, m5, m6, o1, o2, n9, n 10 , n 11 are examples of the second feature points)

[0145]

[0146] By calculation, the plane equation of the inner side bevel plane G i of the K-type groove of the fillet weld in the welding robot coordinate system is: A1X i +B1Y i +C1Z i +D1=0; the plane equation of the outer side bevel plane G o of the K-type groove is: A2X o +B2Y o +C2Z o +D2=0; and the plane equation of the root face plane T of the K-type groove is: A3X T +B3Y T +C3Z T +D3=0. Therefore, in the welding robot coordinate system, when the height is Z, then the intersection lines L1, L3, L2 of the planes Z w =Z and the inner side bevel plane G i , the outer side bevel plane G o , and the root face plane T of the K-type groove of the fillet weld are respectively shown in formulas (2), (3), (4).

[0147] A1X i +B1Y i +C1Z+D1=0 (2)

[0148] A2X o +B2Y o +C2Z+D2=0 (3)

[0149] A3X T +B3Y T +C3Z+D3=0 (4)

[0150] According to the straight line equations of L1, L2, L3 in the welding robot coordinate system, the inside groove angle a1 and the outside groove angle a2 (as an example of the groove angle) of the K-type groove of the fillet weld can be further calculated, and the calculation methods are shown in formulas (5) and (6) respectively. The distance between the intersection points c1 and c2 of the straight lines L2 and L1, L3 can be used to further solve the size of the root face c. If c1 (x1, y1) and c2 (x2, y2), formulas (7) and (8) can be obtained, and further, the size of the root face can be calculated by formula (9).

[0151]

[0152]

[0153]

[0154]

[0155]

[0156] Next, the calculation process of the groove angle and the root face of the fillet weld is introduced.

[0157] The image obtained by positioning and shooting the fillet weld at the bottom cannot extract the parameters required to calculate the groove angle and the root face, so two images shot by related positioning points are needed for parameter extraction and calculation of the groove angle and the root face of the fillet weld. When obtaining the groove angle and the root face of the fillet weld, there are two types of visual images of the fillet weld formed by the first vertical plate, the second vertical plate and the bottom plate. Therefore, the acquisition algorithm of the groove angle and the root face needs to be designed for these two types of visual images to more accurately and efficiently extract the groove angle and the root face.

[0158] The calculation method of the groove angle and the root face of the fillet weld formed by the first vertical plate and the bottom plate is basically the same as that of the fillet weld, and both need to calculate the plane equation of the groove plane and the root plane in the welding robot coordinate system to further calculate the groove angle and the root face. Specifically, as shown in formulas (5) and (6), the plane equation of the groove plane and the root plane in the welding robot coordinate system is calculated, and then the groove angle a1 and the root face c are calculated by formulas (7) and (8).Figures 16a-16e As shown in the figure, Figure 16e The images in are from left to right respectively Figure 16a Region I, region II in, Figure 16b Region III in, Figure 16c Region IV in, Figure 16d Region V in corresponds to the local enlarged view. According to the position of the visual sensor at the shooting point, it can be determined that the points o3, o4, o3', o4' are on the inside beveling plane F i1 of the first vertical plate; n 14 , n 15 , n 14 ', n 15 ' are on the outside beveling plane F o1 formed by the first vertical plate and the bottom plate; n 15 , o4, n 15 ', o4' are on the root face T1 of the weld seam. (Wherein, n 14 , n 15 , n 14 ', n 15 ', o3, o4, o3', o4' are another example of the second feature point)

[0159] The plane equation of the inside beveling plane F i1 formed by the first vertical plate and the bottom plate to form a square weld K type groove is obtained by calculation: A i1 X i1 +B i1 Y i1 +C i1 Z i1 +D i1 =0; the plane equation of the outside beveling plane F o1 of the K type groove is: A o1 X o1 +B o1 Y o1 +C o1 Z o1 +D o1 =0; the plane equation of the root face T1 of the K type groove is: A T1 X T1 +B T1 Y T1 +C T1 Z T1 +D T1 =0. Therefore, when the coordinate of the Y w axis direction in the welding robot coordinate system is Y, then the plane Y w =Y respectively with the inside beveling plane F i1 , the outside beveling plane F o1 of the K type groove formed by the first vertical plate and the bottom plate to form a square weldThe intersection lines L1, L3, L2 of the blunt edge plane T1 are shown in formulas (10), (11), (12) respectively.

[0160] A i1 X i1 +B i1 Y+C i1 Z i1 +D i1 = 0 (10)

[0161] A o1 X o1 +B o1 Y+C o1 Z o1 +D o1 = 0 (11)

[0162] A T1 X T1 +B T1 Y+C T1 Z T1 +D T1 = 0 (12)

[0163] According to the straight line equations of L1, L2, L3 in the welding robot coordinate system, the inner groove angle a1 and the outer groove angle a2 of the first vertical plate and the bottom plate forming the flat angle weld K groove can be further calculated, and the calculation method is shown in formulas (13), (14) respectively. The blunt edge amount c can be further solved according to the distance between the intersection points c1, c2 of the straight lines L2 and L1, L3. If c1(x1, y1), c2(x2, y2), then formulas (15), (16) can be obtained, and the size of the blunt edge amount can be calculated by using formula (9).

[0164]

[0165]

[0166]

[0167]

[0168] The groove angle and the blunt edge amount of the second vertical plate and the bottom plate forming the flat angle weld are obtained by the method of obtaining the groove angle and the blunt edge amount of the flat angle weld. As shown in 17a-17c, the images from left to right are Figure 17c the region I in 17a, Figure 17a the region II in 17b, Figure 17bThe region II in the figure corresponds to the local enlarged view. According to the angle between the straight line where the points n5, n6 are located and the straight line where the points n6, n7 are located, the second vertical plate and the bottom plate form the inside groove angle a1 of the K-type groove on one side of the flat angle weld; the angle between the straight line where the points n4, n5 are located and the straight line where the points n5, n6 are located can calculate the inside groove angle a2 of the K-type groove on one side of the flat angle weld formed by the second vertical plate and the bottom plate; the distance between the points n5, n6 is the size of the c blunt edge on one side. (Where n4, n5, n6, n7 are another example of the second feature point)

[0169] Through calculation, the straight line equation of n4, n5 in the welding robot coordinate system is obtained: A o2 X o2 +B o2 Y o2 +C o2 =0, the straight line equation of n5, n6 is: A T2 X T2 +B T2 Y T2 +C T2 =0, the straight line equation of n6, n7 is: A i2 X i2 +B i2 Y i2 +C i2 =0, and the calculation methods of a1, a2 and c are shown in formulas (17), (18) and (9) respectively. As Figure 17b The feature point image extracted on the other side of the second vertical plate is shown in the figure. Using points n4', n5', n6', n7' (another example of the second feature point), the groove angle and blunt edge amount a1', a2', c' on the other side of the second vertical plate can be obtained according to the above calculation method. Then, according to the coordinates X of the X w axis direction of the points n5, n5' in the welding robot coordinate system and the groove angle and blunt edge amount a1, a2, c on both sides of the second vertical plate, the groove angle and blunt edge amount of the whole weld seam varying with the coordinate X can be calculated according to formulas (19), (20), (21).

[0170]

[0171]

[0172]

[0173]

[0174]

[0175] Finally, the inclination angle θ of the vertical plate is obtained and calculated according to the image contour.

[0176] After the hydraulic support box structure is assembled, the tilt of its upright plate can be divided into two directions: vertical and horizontal. Therefore, it is necessary to solve the tilt angles in both the vertical and horizontal directions to fully express the tilt angle of the upright plate.

[0177] like Figures 18a-18d As shown, where, Figure 18d The images in the image, from left to right, are: Figure 18a Region I and Region II in Figure 18b Regions III, IV, V, and VI in the text. Figure 18c The enlarged views of regions VII and VIII are shown. The horizontal tilt angle of the first vertical plate can be obtained using the slopes of the lines containing m3, m4, m6, and m7 (as an example of feature points). If the equations of the lines containing m3, m4, m6, and m7 are A... l1 X l1 +B l1 Y l1 +C l1 =0 and A l2 X l2 +B l2 Y l2 +C l2 =0, then the tilt angle of the first vertical plate in the horizontal direction can be expressed by formula (22). For example Figure 18b , 18c As shown, the horizontal tilt angle of the second vertical plate needs to be obtained by using the slopes of the lines containing n3, n4, n7, and n8, and the slopes of the lines containing n3′, n4′, n7′, and n8′ to determine the horizontal deflection angle. If the equations of the lines containing n3, n4, n7, n8, n3′, n4′, and n7′, n8′ (as another example of the third feature point) are respectively A... l3 X l3 +B l3 Y l3 +C l3 =0, A l4 X l4 +B l4 Y l4 +C l4 =0, A l3′ X l3′ +B l3′ Y l3′ +C l3′ =0, A l4′ X l4′ +B l4′ Y l4′ +C l4′= 0, the tilt angle of the first vertical plate in the horizontal direction can be calculated using formula (23). Since the calculation method of the tilt angle in the horizontal direction is consistent, the first vertical plate is taken as an example to calculate the tilt angle in the horizontal direction. As shown in FIG. 7, the tilt angle of the first vertical plate in the vertical direction is obtained by taking the angle between the plane of the vertical plate and the vertical plane. As can be seen from the figure, points n Figure 18b 11 , n 12 , n 13 , n 14 are all on the outer plane of the first vertical plate, and the plane equation of the outer plane of the first vertical plate in the welding robot coordinate system can be obtained using formula (1) and the conversion matrix.

[0178] Through calculation, the outer plane equation of the first vertical plate is A y X y +B y Y y +C y Z+D=0, and the equation of the vertical plane corresponding thereto is X w = 0, so the tilt angle of the first vertical plate in the vertical direction can be calculated using formula (24). As shown in FIG. 8, the tilt angle of the second vertical plate in the vertical direction is obtained in the same way as the tilt angle of the second vertical plate in the horizontal direction, that is, by using the slopes of the straight lines on which n3, n4 and n7, n8 are located to obtain the tilt angle of the second vertical plate in the vertical direction. Figure 18b

[0179]

[0180]

[0181]

[0182] Then, according to the workpiece parameters, the welding strategy and welding parameters in the welding process are determined. The welding strategy includes the welding path, the welding gun posture and the welding method, etc.

[0183] ​​Welders' experience in determining whether welding torch oscillation and multi-point welding are necessary during the welding process is a summary of the relationship between welding gap and welding interval quantity, accumulated over long-term operations. Therefore, to transform welders' experience in formulating welding strategies into an intelligent welding decision-making program executable by welding robots, it is necessary to quantitatively describe the relationship between welding strategy (S) and welding gap quantity (b). When welders make welding decisions based on welding gap quantity, they do not formulate a corresponding welding strategy for each welding gap quantity, but rather use the same welding strategy when the welding gap quantity is within a certain range. By observing and consulting welders on the process of formulating welding decisions, it was finally determined that the same welding strategy should be used when the welding gap quantity fluctuation range is within 1 mm. Therefore, this embodiment provides a piecewise function between welding strategy (S) and welding gap quantity (b) as shown in formula (25), where zero indicates that welding torch oscillation and multi-point welding are not required, one indicates that welding torch oscillation is required but multi-point welding is not required, and two indicates that welding torch oscillation and multi-point welding are required.

[0184]

[0185] Therefore, before welding, the welding robot can input the welding gap amount obtained by the vision image processing module into the decision function of the welding strategy to obtain the welding strategy under the current welding gap amount.

[0186] Furthermore, intelligent planning and decision-making for welding paths are carried out.

[0187] (1) No welding torch oscillation is required during the welding process.

[0188] When the welding conditions do not require welding torch oscillation to meet the weld leg size requirements, such as Figure 19 As shown, the tilt of the vertical plate causes inconsistencies in the working conditions on the inner and outer sides, resulting in a change in the welding path. During welding, the welding torch must be pointed towards the center of the weld joint. By drawing perpendicular lines from the weld joint feature points a and a1 to the upper surface of the base plate, the feet of the perpendiculars o and o1 can be obtained. Therefore, the welding torch must be pointed towards the midpoint of line segments ao and a1o1. Thus, the coordinates of the weld joint center can be calculated simply by obtaining the coordinates of the weld joint feature points obtained from the image processing module. Taking the inner side of the vertical plate as an example, if the coordinates of point a are (x... a ,y a The coordinates of point O are (x o ,y o Then the center P of the weld joint mid The coordinates are Therefore, before welding, the welding torch path under the current welding condition can be obtained simply by using the coordinates of points a, o, a1, and o1 obtained by the image processing system.

[0189] (2) The welding torch needs to be swung during the welding process.

[0190] Based on the characteristics of the weld seams in the hydraulic support box structure, a triangular oscillation method was selected. To determine the welding path, the parameters of the welding torch oscillation method must first be analyzed and determined. For example... Figures 20a-20b The diagram shows the triangular welding torch oscillation method and parameters. Therefore, before welding, it is necessary to determine the weld width B, the welding torch oscillation amplitude A, γ, γ1, γ2, γ3, and the oscillation distance d. b Dwell time on the left side T l Dwell time on the right side T r These parameters serve as determinants of the welding path.

[0191] The relationship between the oscillation amplitude of the welding torch and the weld width is very close. The oscillation amplitude of the welding torch is selected to be half the weld width. In order to obtain the oscillation amplitude of the welding torch under different welding conditions, the weld width needs to be calculated first. The weld width B can be calculated by formula (26) using the geometric relationship between the weld width B, the welding gap b, the bevel angle α, the bevel depth h and the weld leg size K, and thus the size of the welding torch oscillation amplitude A can be obtained.

[0192]

[0193] like Figure 20a As shown, when welding vertical fillet welds, an equilateral triangle oscillation method is used. Before welding, it is only necessary to determine the oscillation amplitude A and the oscillation distance d of the welding torch. b Dwell time on the left side T l Dwell time on the right side T r The welding path L for one oscillation cycle can then be obtained. Tp It can be expressed by formula (27). The welding path L of the entire fillet weld is... p This involves continuously repeating the welding path of this cycle until the entire weld seam is completed.

[0194] L Tp =h1(A,d b ,T l ,T r (27)

[0195] like Figure 20b As shown, when welding fillet welds, a slanted triangle oscillation method is used. Compared to the equilateral triangle oscillation method, determining the path requires more parameters and is more complex. Before welding, it is necessary to determine the deflection angles γ1, γ2, and γ3 of the welding torch oscillation, as well as the oscillation distance d. b Dwell time on the left side T l Dwell time on the right side T r The welding path L for one oscillation cycle can then be obtained. Ts It can be expressed by formula (28). The welding path L of the entire vertical fillet weld is... s This involves continuously repeating the welding path of this cycle until the entire weld seam is completed.

[0196] L Ts =h2(γ1,γ2,γ3,d b ,T l ,T r (28)

[0197] In summary, before welding, it is necessary to first determine whether the welding torch needs to be oscillated to meet the weld formation quality based on the welding gap (b), bevel angle (α), blunt edge (c), and tilt angle (θ) of the vertical plate obtained from the vision measurement system. Then, the welding path under the current welding conditions can be obtained according to their respective requirements.

[0198] Furthermore, intelligent decision-making is made regarding the welding torch posture.

[0199] During the welding of hydraulic support box structures, the welding torch posture needs to be adjusted according to the working conditions after assembly. In fillet welds, the welding torch posture significantly affects the weld formation quality. An improper welding torch posture can easily lead to defects such as undercut and incomplete penetration, resulting in a smaller weld bead size and thus affecting weld strength. Based on the spatial position of the welding torch, angles β1 and β2 are defined to characterize the spatial posture of the welding torch, such as... Figures 21a-21b As shown. To ensure weld formation quality, this embodiment designs an intelligent decision-making algorithm for appropriate welding torch postures corresponding to different welding conditions based on welder experience.

[0200] To ensure weld quality, welders first observe the weld type, the inclination angle of the vertical plate, the welding gap, and the bevel angle before welding. Based on these factors, they determine the angle β1, and the weld type and welding gap determine the angle β2. Therefore, a decision-making algorithm for β1 and β2 was established based on welder experience.

[0201] like Figure 22 As shown, the algorithm for the welding gun posture β1 on the inner and outer sides of the vertical plate is established by using the geometric relationship between the bevel angle and the tilt angle of the vertical plate as shown in formula (29).

[0202]

[0203] As the β2 angle increases, the weld width also increases. For vertical welds, the increase of the β2 angle not only increases the weld width but also increases the lifting effect of the arc on the molten pool. Therefore, the maximum tilt angle is used. Thus, the algorithm designed in this embodiment is as shown in formula (30).

[0204]

[0205] Therefore, this embodiment uses formulas (29) and (30) to make intelligent planning decisions on the welding gun posture based on the welding status, i.e., the welding gap b, the bevel angle a, and the tilt angle θ of the vertical plate obtained by the visual measurement system.

[0206] Furthermore, intelligent decision-making is carried out regarding welding methods.

[0207] The tilt of the welding torch has a significant impact on the weld formation. Based on the direction of the tilt, welding can be divided into forward tilting and backward tilting methods. Forward tilting refers to tilting the welding torch along the welding direction, while backward tilting refers to tilting it against the welding direction. Related literature indicates that backward tilting welds have a more uniform surface finish, finer and more regular fish-scale patterns, and slightly larger weld leg dimensions compared to forward tilting welds. For vertical welds, downward vertical welding is suitable for thin plates, while upward vertical welding meets the requirements for thick plates. Therefore, in welding hydraulic support box structures, backward tilting welding is used for fillet welds, and upward vertical welding is used for vertical fillet welds.

[0208] Finally, intelligent decisions are made regarding welding process parameters.

[0209] Welding process parameters (welding current, welding voltage, welding speed, wire feed speed, etc.) have a significant impact on weld formation quality. Therefore, to achieve intelligent welding, it is necessary to make intelligent decisions on the optimal welding process parameters under different working conditions. In this embodiment, a BP neural network is used to predict the welding process parameters.

[0210] To ensure weld formation quality under different welding conditions, parameters that express the welding condition, such as the welding gap (b), bevel angle (α), blunt edge (c), and plate tilt angle (θ) obtained from the vision measurement system, are used as... Figure 9 The input vector of the neural network shown, along with process parameters such as welding current, welding voltage, welding speed, and wire feed speed, serves as the network's predicted output vector, thus yielding the welding parameters corresponding to the current process parameters.

[0211] Finally, using the welding strategy and welding parameters obtained by the above method, a welding robot was used to weld the hydraulic support box structure.

[0212] Please see Figure 23 , Figure 23A structural block diagram of an electronic device provided by an embodiment of the present application is shown. The electronic device can include one or more processors 1002, system control logic 1008 connected to at least one of the processors 1002, system memory 1004 connected to the system control logic 1008, non-volatile memory (NVM) 1006 connected to the system control logic 1008, and a network interface 1010 connected to the system control logic 1008. In one specific embodiment of the present application, the electronic device is a welding robot.

[0213] The processor(s) 1002 can include one or more single-core or multi-core processors. The processor(s) 1002 can include any combination of general-purpose processors and dedicated processors (e.g., graphics processors, application processors, baseband processors, etc.). In embodiments herein, the processor(s) 1002 can be configured to perform the welding method described above.

[0214] In some embodiments, the system control logic 1008 can include any suitable interface controllers to provide for any suitable interface to at least one of the processor(s) 1002 and / or any suitable device or component in communication with the system control logic 1008.

[0215] In some embodiments, the system control logic 1008 can include one or more memory controllers to provide an interface to the system memory 1004. The system memory 1004 can be used to load and store data and / or instructions. In some embodiments, the system memory 1004 of the electronic device can include any suitable volatile memory, such as suitable Dynamic Random Access Memory (DRAM).

[0216] The NVM / memory 1006 can include one or more tangible, non-transitory computer-readable media used to store data and / or instructions. In some embodiments, the NVM / memory 1006 can include any suitable non-volatile memory, such as flash memory, and / or any suitable non-volatile storage device, such as at least one of a Hard Disk Drive (HDD), a Compact Disc (CD) drive, a Digital Versatile Disc (DVD) drive.

[0217] The NVM / memory 1006 can include a portion of the storage resources installed on the device of the electronic device, or it can be accessible by the device but not necessarily part of the device. For example, the NVM / memory 1006 can be accessed over a network via the network interface 1010.

[0218] The network interface 1010 can include a transceiver to provide a radio interface for the electronic device to communicate with any other suitable device (e.g., a front end module, an antenna, etc.) over one or more networks. In some embodiments, the network interface 1010 can be integrated with other components of the electronic device. For example, the network interface 1010 can be integrated with at least one of the processor 1002, the system memory 1004, the NVM / memory 1006, and a firmware device (not shown) having instructions that, when executed by at least one of the processors 1002, cause the electronic device to implement the welding method described above.

[0219] In one embodiment, at least one of the processors 1002 can be packaged with logic for one or more controllers of the system control logic 1008 to form a system in a package (SiP). In one embodiment, at least one of the processors 1002 can be integrated on the same die with logic for one or more controllers of the system control logic 1008 to form a system on chip (SoC).

[0220] It can be appreciated that the structure illustrated by the embodiments of the present application does not constitute a specific limitation to the electronic device. In other embodiments of the present application, the electronic device can include more or fewer components than illustrated, or combine certain components, or split certain components, or different arrangement of components. The illustrated components can be implemented in hardware, software, or a combination of software and hardware.

[0221] The embodiments of the present application also provide a computer readable storage medium. The computer readable storage medium stores a computer program. The computer program includes program instructions. The computer readable storage medium can be any available medium or data storage device that can be accessed by an electronic device or a data center and the like containing one or more available media. The available medium can be a magnetic medium (e.g., a floppy diskette, a hard disk, a magnetic tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid state disk) and the like. The program instructions are executed by the electronic device to implement the welding method described above.

[0222] The embodiments of the present application also provide a computer program product containing instructions. The computer program product can be a software or program product containing computer programs / instructions, which can be run on an electronic device or stored in any available medium. When the computer program product is run on at least one electronic device, the at least one electronic device is caused to execute the welding method described above.

Claims

1. A welding method, characterized in that, Applied to welding robots, the method includes: Identify the workpieces to be welded, which are the workpieces that need to be welded together during assembly; The target workpiece, including the workpiece to be welded, is subjected to laser scanning processing to obtain images of the target workpiece in multiple directions as initial images; The initial image is input into a preset classification model for processing, so as to make a qualitative judgment on the rationality of the welding condition state corresponding to the workpiece to be welded through the classification model, and obtain the processing result. Based on the processing results, if it is determined that the welding condition of the workpiece to be welded is reasonable, then the image contours of the target workpiece from multiple angles are determined. If it is determined that the welding condition of the workpiece to be welded is unreasonable, then a first reminder message is generated and sent to the user device so that the user can reassemble the workpiece to be welded according to the first reminder message. Based on the image contour, multiple feature points are determined. Based on the multiple feature points, workpiece parameters are determined. The multiple feature points are points on the image contour related to the welding position of the target workpiece. The workpiece parameters include welding gap, bevel angle, blunt edge amount, and vertical plate tilt angle. Wherein, when the workpiece parameter is the welding gap, the multiple feature points are multiple first feature points, which are points on the image contour related to the inflection point at the welding position of the target workpiece. When the workpiece parameters are the bevel angle and the blunt edge amount, the multiple feature points are multiple second feature points, which are points on the image contour related to the bevel plane and blunt edge plane at the welding position of the target workpiece. When the workpiece parameter is the vertical plate tilt angle, the multiple feature points are multiple third feature points, which are points on the image contour related to the horizontal plane and vertical plane at the welding position of the target workpiece. Based on the workpiece parameters, determine the welding strategy and welding parameters; The workpiece to be welded is welded according to the welding strategy and the welding parameters.

2. The welding method as described in claim 1, characterized in that, Determining the image contours of the target workpiece from multiple angles, including the workpiece to be welded, includes: The initial image is input into the first model for semantic segmentation processing to obtain the second image; The second image is processed into grayscale to obtain the third image; The third image is binarized to obtain the fourth image; Edge extraction processing is performed on the fourth image to obtain the image contour.

3. The welding method as described in claim 2, characterized in that, The welding strategy includes the welding torch path and welding torch posture. The welding strategy is determined based on the workpiece parameters, including: The welding torch path is determined based on the welding gap amount; The welding torch posture is determined based on the welding gap, the bevel angle, and the tilt angle of the vertical plate.

4. The welding method as described in claim 3, characterized in that, Determining the welding torch path based on the welding gap amount includes: The first information is determined based on the welding gap amount, and the first information includes whether welding torch oscillation is required and whether multi-point spot welding is required. Based on the first information, the welding torch path is determined.

5. The welding method as described in claim 4, characterized in that, The welding strategy also includes a welding method, which is determined by the following means: Based on the image outline, determine the weld type; The welding method is determined based on the weld type.

6. The welding method according to any one of claims 1-5, characterized in that, Based on the workpiece parameters, the welding parameters are determined, including: The workpiece parameters are input into a preset prediction model for processing to determine the welding parameters.

Images