Feedback control method for laser welding

Abstract

The invention discloses a feedback control method for laser welding, and relates to the technical field of laser welding, and the method comprises the following steps: obtaining target design information of a target assembly part and a target sub-design information set of a sub-component set; carrying out clamping positioning analysis to obtain a clamping feature point set and a workpiece alignment vector set; welding control analysis is conducted, and welding path control is obtained; pre-constructing a welding repair analysis module; clamping the sub-components of the sub-component set to obtain a to-be-welded entity; in the automatic welding process of a to-be-welded entity, the laser interferometer collects and feeds back a real-time interference fringe sequence to the welding repair analysis module; and welding repair analysis is conducted according to the real-time interference fringe sequence to obtain repair path control, and the target welding assembly executes feedback control optimization on the to-be-welded entity with the repair path control as a constraint. The technical problems that in the prior art, laser welding control precision is insufficient, and stability is poor are solved, and the technical effect of improving welding quality and efficiency is achieved.

Description

Technical Field

[0001] The invention relates to the technical field of laser welding, and in particular to a feedback control method for laser welding. Background Art

[0002] In the field of laser welding, the stability and accuracy of welding quality have a crucial impact on the final performance and reliability of the product. With the rapid development of the manufacturing industry and the intensification of market competition, laser welding technology has been widely used in industries such as aviation, automobiles, and electronics. However, the laser welding process is a highly complex and variable physical process involving the interaction of multiple aspects such as light, heat, and force, which places extremely high demands on the precise control of the welding process.

[0003] Although the traditional laser welding feedback control method can achieve stable control of the welding process to a certain extent, it often has problems such as insufficient control accuracy and poor stability. These problems may lead to welding defects such as uneven welds, pores, cracks, etc., thus affecting the quality and service life of the product.

[0004] As modern manufacturing continues to improve production efficiency and product quality, higher requirements are placed on the intelligence and automation level of the laser welding process. Traditional feedback control methods can no longer meet the high efficiency, precision and safety requirements of modern laser welding processes. Summary of the invention

[0005] The present application provides a feedback control method for laser welding, which is used to solve the technical problems of insufficient laser welding control accuracy and poor stability in the prior art.

[0006] In view of the above problems, the present application provides a feedback control method for laser welding.

[0007] In a first aspect of the present application, a feedback control method for laser welding is provided, the method comprising: Interactively obtain target design information of a target assembly and a target sub-design information set of a sub-component set, wherein the target assembly is obtained by laser welding the sub-component set; perform clamping positioning analysis according to the target design information and the target sub-design information to obtain a clamping feature point set and a workpiece alignment vector set; perform welding control analysis according to the target design information and the target sub-design information to obtain welding path control, wherein the welding path control includes welding trajectory parameters and welding control parameters; pre-build a welding repair analysis module, wherein the welding repair analysis module is connected to the laser interferometer of the target welding component in communication; target clamping assembly The target welding component and the target clamping component constitute an automated laser welding device; during the automated welding process of the target welding component on the entity to be welded with the welding path control as a constraint, the laser interferometer collects and feeds back a real-time interference fringe sequence to the welding repair analysis module; the welding repair analysis module performs welding repair analysis according to the real-time interference fringe sequence to obtain a repair path control, and the target welding component performs feedback control optimization on the entity to be welded with the repair path control as a constraint.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages: The present application interactively obtains target design information of a target assembly and a target sub-design information set of a sub-component set, wherein the target assembly is obtained by laser welding the sub-component set; performs clamping positioning analysis according to the target design information and the target sub-design information to obtain a clamping feature point set and a workpiece alignment vector set; performs welding control analysis according to the target design information and the target sub-design information to obtain welding path control, wherein the welding path control includes welding trajectory parameters and welding control parameters; pre-constructs a welding repair analysis module, wherein the welding repair analysis module is communicatively connected with a laser interferometer of a target welding component; the target clamping component clamps the sub-components of the sub-component set with the clamping feature point set and the workpiece alignment vector set as constraints to obtain a to-be-welded entity, wherein the target welding component and the target clamping component constitute an automated laser welding device; during the process of the target welding component performing automated welding on the to-be-welded entity with the welding path control as a constraint, the laser interferometer collects and feeds back a real-time interference fringe sequence to the welding repair analysis module; the welding repair analysis module performs welding repair analysis according to the real-time interference fringe sequence to obtain a repair path control, and the target welding component performs feedback control optimization on the to-be-welded entity with the repair path control as a constraint. The present invention solves the technical problems of insufficient laser welding control accuracy and poor stability in the prior art. It realizes precise control of the welding process by real-time monitoring of key parameters in the welding process and real-time adjustment of welding parameters according to the monitoring results, thereby achieving the technical effect of improving welding quality and efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0010] Figure 1 A schematic flow chart of a feedback control method for laser welding provided in an embodiment of the present application; Figure 2 A schematic diagram of a flow chart for obtaining a clamping feature point set and a workpiece alignment vector set in a feedback control method for laser welding provided in an embodiment of the present application. DETAILED DESCRIPTION

[0011] The present application provides a feedback control method for laser welding to solve the technical problems of insufficient laser welding control accuracy and poor stability in the prior art. By real-time monitoring of key parameters in the welding process and real-time adjustment of welding parameters based on the monitoring results, precise control of the welding process can be achieved, thereby achieving the technical effect of improving welding quality and efficiency.

[0012] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0013] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products or devices.

[0014] Embodiment 1 like Figure 1 As shown, the present application provides a feedback control method for laser welding, the method comprising: Step S100: interactively obtaining target design information of a target assembly and a target sub-design information set of a sub-component set, wherein the target assembly is obtained by laser welding the sub-component set; In the embodiment of the present application, the target design information of the target assembly and the target sub-design information set of the sub-assembly set are obtained through interaction. The interaction is achieved in a variety of ways, including but not limited to manual input by the user, file import, and the like.

[0015] In this application, the design information of the target assembly and subassembly is manually input by the user. The target design information and target sub-design information set include geometric dimensions, material information, assembly requirements, etc.

[0016] The geometric dimensions include the length, width, height, angle and other geometric parameters of the assembly and sub-assembly. The material information includes the type, thickness, thermal conductivity and other materials used in the assembly and sub-assembly. The material information includes the type, thickness, thermal conductivity and other materials used in the assembly and sub-assembly.

[0017] Step S200: performing clamping positioning analysis according to the target design information and the target sub-design information to obtain a clamping feature point set and a workpiece alignment vector set; In an embodiment of the present application, based on the acquired target design information and target sub-design information, an appropriate clamping strategy is determined according to factors such as the structural characteristics of the assembly, the shape and size of the sub-components, and the location, number and clamping method of the clamping points are selected.

[0018] The clamping feature point refers to the key position point used to fix the subcomponent. The specific position of the clamping feature point is determined according to the geometric dimensions and assembly requirements in the design information. After the position of the clamping feature point is determined, a clamping feature point set is generated. The clamping feature point set contains the coordinate information of all necessary clamping points.

[0019] Workpiece alignment vectors are vectors used to determine the correct position and orientation of a subcomponent relative to an assembly. Workpiece alignment vectors are calculated based on assembly requirements and geometric relationships in the design information. Workpiece alignment vectors ensure that subcomponents can be accurately aligned with the assembly before welding. Based on the analysis of workpiece alignment vectors, a workpiece alignment vector set is generated. The workpiece alignment vector set contains all necessary alignment vector information.

[0020] Step S300: performing welding control analysis according to the target design information and the target sub-design information to obtain welding path control, wherein the welding path control includes welding trajectory parameters and welding control parameters; In the embodiment of the present application, the target design information and the target sub-design information are first analyzed, but the overall structure of the assembly, the layout and relative position relationship of the sub-components, and the type and position of the welded joints are underestimated. Then the welded joints are analyzed. The type of joint, the size and shape of the joint, and the material properties of the joint are determined. The types of joints include butt joints, corner joints, T-joints, etc.

[0021] When planning the welding trajectory, the welding trajectory is generated according to the assembly requirements in the design information and the results of the welding joint analysis. The welding trajectory is the path followed by the welding head during the welding process. The welding trajectory is generated according to the position, direction and shape of the welding joint and the overall structure of the assembly to ensure that the welding trajectory can cover all areas that need to be welded and avoid interference with other parts.

[0022] Welding trajectory parameters include welding speed, welding angle, welding depth, etc. These parameters directly affect the quality and efficiency of welding. According to the material properties, joint type and expected welding quality in the design information, the welding trajectory parameters are determined by empirical formulas. For example, for welding different materials, the welding speed and welding current are adjusted; for different types of joints, the welding angle and depth are adjusted.

[0023] In addition to welding trajectory parameters, welding control parameters are also the key to ensure the smooth progress of the welding process. Welding control parameters include welding current, welding voltage, welding time, etc. Welding control parameters are set according to the material thickness in the design information, the size and shape of the welded joint, and the expected welding quality.

[0024] Finally, the welding trajectory parameters and welding control parameters are integrated to generate the welding path control.

[0025] Step S400: pre-constructing a welding repair analysis module, wherein the welding repair analysis module is communicatively connected to a laser interferometer of a target welding component; In the embodiment of the present application, the welding repair analysis module is a comprehensive module that integrates functions such as data analysis, defect identification, and repair strategy generation. The welding repair analysis module communicates with the laser interferometer to obtain data during the welding process in real time, and uses built-in algorithms and models to perform data analysis, discover welding defects, and provide corresponding repair solutions.

[0026] The communication interface between the welding repair analysis module and the laser interferometer adopts a standard data transmission protocol to ensure the accuracy and real-time nature of the data. The laser interferometer transmits the real-time collected welding data, such as welding speed and temperature distribution, to the welding repair analysis module through the communication interface.

[0027] The welding repair analysis module performs pre-processing operations such as denoising and filtering on the received welding data to improve the data quality. Then, the machine learning algorithm is used to extract characteristic parameters related to welding quality from the pre-processed data. Based on the extracted characteristic parameters, the preset defect recognition model is used to determine whether there are defects in the welding process, such as pores, cracks, and lack of fusion. For the identified welding defects, the module can accurately locate the location and size of the defects.

[0028] Finally, according to the type and severity of the defect, combined with the material properties and design requirements of the target welded component, a corresponding repair plan is generated. The repair plan includes adjusting welding parameters, replacing welding materials, etc.

[0029] Step S500: the target clamping component clamps the subcomponents of the subcomponent set with the clamping feature point set and the workpiece alignment vector set as constraints to obtain a to-be-welded entity, wherein the target welding component and the target clamping component constitute an automated laser welding device; In the embodiment of the present application, the target clamping assembly includes mechanical clamps, a servo drive system, sensors, and a control system. The mechanical clamps are used to directly contact and fix the subcomponents, and the servo drive system is responsible for controlling the opening and closing and movement of the clamps. The sensors are used to monitor various parameters in the clamping process in real time, such as clamping force, position, etc., to ensure the accuracy and stability of the clamping. The control system receives instructions from the welding control system.

[0030] The clamping feature point set is obtained from the previous design information analysis and clamping positioning analysis, and it contains the coordinate information of the key points used for clamping on the subcomponent. The target clamping component accurately adjusts the position and angle of the mechanical gripper based on these feature point sets to ensure that the gripper can accurately grasp the predetermined position of the subcomponent. At the same time, the clamping component will also calculate the best clamping force and method based on the information of the feature point set to avoid damage to the subcomponent.

[0031] The workpiece alignment vector set describes the vector information of the correct position and direction of the subcomponent relative to the assembly. When the target clamping component clamps the subcomponent, it strictly follows the workpiece alignment vector set and accurately controls the movement trajectory of the clamping jaws through the servo drive system to accurately place the subcomponent in the predetermined welding position.

[0032] The target welding assembly and the target clamping assembly constitute the automated laser welding equipment. During the clamping process, the welding control system sends corresponding instructions to the clamping assembly according to the requirements of the welding path control. After receiving the instructions, the clamping assembly clamps and positions the sub-component according to the preset clamping feature point set and the workpiece alignment vector set. Once the sub-component is accurately clamped and positioned at the predetermined position, the welding assembly starts the welding operation. Through the precise clamping and positioning of the target clamping assembly, the sub-component is stably fixed on the assembly to form a solid to be welded.

[0033] Step S600: the target welding assembly performs automated welding on the to-be-welded entity with the welding path control as a constraint, and the laser interferometer collects and feeds back a real-time interference fringe sequence to the welding repair analysis module; In an embodiment of the present application, after the target welding assembly receives the welding path control instruction sent by the welding control system, it starts to perform automated welding according to the preset welding trajectory parameters and welding control parameters. In this process, the laser interferometer works by emitting a laser beam and receiving its reflected light. When the laser beam irradiates the welding area of ​​the entity to be welded, due to factors such as thermal deformation and dynamic changes in the molten pool generated during the welding process, the reflected light of the laser beam will change, forming interference fringes. These interference fringe sequences contain real-time status information of the welding area, such as temperature distribution, molten pool morphology, welding defects, etc. The laser interferometer continuously collects these real-time interference fringe sequences and converts them into digital signals. Then, these digital signals are transmitted to the welding repair analysis module in real time through the communication interface.

[0034] Step S700: the welding repair analysis module performs welding repair analysis according to the real-time interference fringe sequence to obtain a repair path control, and the target welding assembly performs feedback control optimization on the entity to be welded with the repair path control as a constraint.

[0035] In the embodiment of the present application, after receiving the real-time interference fringe sequence, the welding repair analysis module first performs preprocessing, including denoising, filtering, etc., to improve the accuracy and reliability of the data. Then, the preprocessed interference fringe sequence is deeply analyzed using built-in algorithms and models, and the morphology, density, change rate and other characteristics of the interference fringes are extracted and quantified to obtain the real-time status information of the welding area.

[0036] Based on the analysis results of the interference fringe sequence, the welding repair analysis module identifies various defects that occur during the welding process, such as pores, cracks, and lack of fusion. The module determines the type and location of the defect by comparing the difference between the actual interference fringes and the preset standard interference fringes, combined with the welding process parameters and material properties. Once the welding defect is identified, the welding repair analysis module immediately generates repair path control instructions. These instructions are designed to guide the target welding component to accurately repair the defect. The repair path control includes parameters such as the location of the repair point, the repair speed, and the repair energy.

[0037] After receiving the repair path control instruction, the target welding component performs the repair operation on the welding entity according to the instruction requirements. During this process, the welding component maintains real-time communication with the welding repair analysis module and feeds back the real-time data of the repair process to the module. The welding repair analysis module monitors and evaluates the repair process in real time based on these feedback data. If the repair effect is not ideal or new defects exist, the module immediately adjusts the repair path control parameters and sends them to the welding component again to achieve feedback control optimization.

[0038] In this way, the Weld Repair Analysis module is able to continuously optimize the repair process and ensure improved weld quality.

[0039] Further, such as Figure 2 As shown, step S200 in the method provided in the application embodiment also includes: Perform digital modeling and restoration of the target assembly according to the target design information to generate a target assembly model; Performing digital modeling restoration of the subcomponent set according to the target sub-design information to generate a target subcomponent model set, wherein the subcomponent set includes M subcomponents, the target sub-design information includes M subcomponent design information and M assembly association information of the M subcomponents, and the target subcomponent model set includes M subcomponent models of the M subcomponents; Performing welding apparent center point recognition on the M sub-component models to obtain M clamping feature points, wherein the M clamping feature points constitute the clamping feature point set; Performing component edge feature point recognition on the M sub-component models to obtain M edge feature points; According to the M assembly association information, the M sub-component models are assembled and fitted to the target assembly model, and the M clamping feature points are used as workpiece alignment constraints, and the M edge feature points are used as alignment vector constraints to generate M workpiece alignment vectors, which constitute the workpiece alignment vector set.

[0040] In the embodiment of the present application, according to the target design information, computer-aided design software, such as CAD or digital modeling tools, is used to digitally model and restore the target assembly. According to the parameters such as size, shape, material, etc. in the design information, corresponding digital models are constructed one by one. Finally, a complete target assembly model that is highly consistent with the actual assembly is obtained.

[0041] Next, the subcomponent set is digitally modeled and restored according to the target sub-design information. The subcomponent set includes M subcomponents, each of which has corresponding subcomponent design information and assembly-related information. Similarly, CAD software or digital modeling tools are used to build corresponding subcomponent models one by one according to the design information of each subcomponent, and a target subcomponent model set containing M subcomponent models is obtained.

[0042] After the sub-component model is generated, the clamping feature points are identified. The clamping feature points are key points that guide the clamping assembly to clamp the sub-component. The welding apparent center points of the M sub-component models are identified using digital model processing software or algorithms, such as Geomagic Design X. The welding apparent center point is a position on the sub-component that is suitable for clamping, and the welding apparent center points constitute a clamping feature point set.

[0043] In addition to the clamping feature points, the edge feature points of the subcomponent are identified. The edge feature points of the subcomponent are located at the edge of the subcomponent and are used for subsequent alignment vector generation. Similarly, the digital model processing software or algorithm is used to identify the edge feature points of the M subcomponent models to obtain M edge feature points.

[0044] According to the M assembly association information in the target sub-design information, the M sub-component models are assembled and fitted to the target assembly model according to the preset assembly order and association relationship. In this process, the relative position and angle between the sub-components are ensured to be consistent with the design information. After the assembly fitting is completed, the workpiece alignment vector is generated with the M clamping feature points as the workpiece alignment constraints and the M edge feature points as the alignment vector constraints. The alignment vector is an important parameter to guide the precise alignment of the welding assembly, which describes the position and direction information of the sub-component relative to the target assembly model during the assembly process. By calculating the relative position and angle relationship between each sub-component model and the target assembly model, M workpiece alignment vectors are obtained, which together constitute the workpiece alignment vector set.

[0045] Furthermore, the method further comprises: According to the M assembly association information, the M subcomponent models are assembled and fitted to the target assembly model, and the target assembly model is used as a subtracted set and the M subcomponent models are used as a subtracted set to perform a difference set solution to obtain a weld frame model; Performing connection point identification on the weld frame model, and decomposing the weld frame model into multiple welding tracks based on the connection point identification results; Based on the continuity of welding trajectories, the connection schemes of the multiple welding trajectories are enumerated in the weld frame model to obtain multiple groups of welding trajectory connection schemes and multiple welding space jump distances; The plurality of welding space jump distances are serialized and minimum value calls are performed to obtain the welding trajectory parameters.

[0046] In an embodiment of the present application, the exact position and direction of each subcomponent in the target assembly model is first determined based on M assembly association information. The association information includes the relative position, angle, alignment, etc. between the subcomponents. Using this information, the M subcomponent models are accurately placed at the corresponding positions of the target assembly model according to the preset assembly order and association relationship. The target assembly model is used as a subtracted set, and the M subcomponent models are used as subtracted sets for difference set solving. Difference set solving is a geometric operation used to subtract one geometric shape from another to obtain the difference between them. In this scenario, the weld frame model is obtained by difference set solving. The weld frame model represents the area where welding is required to complete the final assembly after the subcomponent models are assembled. These areas are the gaps or contact surfaces between the subcomponent models, and they are firmly connected together by welding.

[0047] After obtaining the weld frame model, the connection points are identified. The connection points refer to the specific locations in the weld frame model where welding is required. These connection points are located at the boundaries or intersections of the weld frame model and are the starting and ending points of the welding trajectory. By identifying these connection points, the weld frame model is disassembled into multiple welding trajectories. Each welding trajectory represents a welding path from one connection point to another. After being disassembled into multiple welding trajectories, the connection problem between these trajectories is considered. Based on the continuity of the welding trajectory, the connection schemes of multiple welding trajectories are enumerated on the weld frame model. That is, all possible trajectory connection methods are tried to find the optimal welding sequence and path. In the enumeration process, multiple sets of welding trajectory connection schemes and multiple welding space jump distances are obtained. The welding space jump distance refers to the spatial distance that the welding head needs to move from one welding trajectory to another welding trajectory.

[0048] Finally, these welding space jump distances are serialized and minimum value calls are made. Through serialization, these distances are arranged in a certain order for subsequent processing. Through the minimum value call, the minimum value is found from these serialized distances. This minimum value represents the optimal welding trajectory connection solution, that is, the shortest distance that the welding head needs to move during the welding process. Through the above process, the welding trajectory parameters are obtained.

[0049] Furthermore, the method further comprises: Interactively obtain material parameters and thickness parameters of the target assembly, and obtain a sample welding control parameter set by calling the material parameters and the thickness parameters as constraints, wherein the sample welding control parameter set includes a plurality of sample weld parameters, a plurality of sample welding strengths, and a plurality of sample welding control parameters; Building a laser control analysis model based on a CNN neural network, and using the sample welding control parameter set to perform training optimization on the laser control analysis model; Preset a multi-level weld width constraint, and traverse the multiple welding tracks based on the multi-level weld width constraint to divide the multiple welding tracks into multiple groups of connecting weld widths at two levels; Interactively obtaining multi-region welding strength constraints of the target assembly, and dividing the multi-region welding strength constraints according to the multiple welding trajectories to obtain multiple groups of connection welding strengths; Synchronize the multiple groups of connection weld widths and the multiple groups of connection weld strengths to the laser control analysis model to obtain multiple groups of local welding control parameters, and sequence the multiple groups of local welding control parameters according to the welding trajectory parameters to generate the welding control parameters; The welding trajectory parameters and the welding control parameters are fitted to generate the welding path control.

[0050] In the embodiment of the present application, the material parameters and thickness parameters of the target assembly are obtained through interaction. The material parameters include the type, melting point, thermal conductivity, etc. of the material, and the thickness parameters describe the thickness of different parts of the assembly.

[0051] A pre-prepared sample welding control parameter set is called with material parameters and thickness parameters as constraints. The set contains multiple sample weld parameters, sample welding strengths, and sample welding control parameters.

[0052] Next, a CNN neural network was used to build a laser control analysis model. The model was trained and optimized using a sample welding control parameter set so that it could learn and identify the optimal welding control parameters for different materials and thicknesses.

[0053] Then, analyze the required weld width based on the material, structure, and usage conditions of the target assembly. According to the analysis results, determine the weld width range for each welding area based on industry standards, experimental data, or empirical formulas. Set multi-level weld width constraints based on the characteristics and importance of different areas. For example, critical stress areas require stricter weld width requirements, while non-critical areas have a certain width floating range.

[0054] Then, according to the multi-level weld width constraint, traverse multiple welding trajectories, and for each welding trajectory, calculate or simulate the actual weld width generated by the welding trajectory. Check whether the corresponding weld width meets the preset multi-level weld width constraint. If the predicted weld width of a certain welding trajectory does not meet the preset multi-level weld width constraint, split it. Determine the location of the split point based on the specific situation of not meeting the constraint, the change of weld width, the requirements of the welding process, and the structural characteristics of the assembly. Finally, split the multi-segment welding trajectory into multiple groups of connecting weld widths at the determined split point.

[0055] Through interaction with engineers, technicians or relevant experts, clarify the strength constraints of each welding area. The strength constraints of the welding area include the minimum and maximum welding strengths required for each area, as well as the possible strength variation range. Analyze the correspondence between each welding trajectory and the welding area. Determine which areas each trajectory is responsible for welding, as well as the relative position and length of these areas in the trajectory. According to the correspondence between the welding trajectory and the welding area, divide the determined welding strength constraints into each welding trajectory by calculating the average strength requirement and maximum strength requirement of the area covered by each trajectory. After completing the division of the welding strength constraints, calculate the connection welding strength of each trajectory according to the specific parameters of each welding trajectory, such as welding speed, temperature, pressure, etc. and the characteristics of the welding material, and obtain multiple groups of connection welding strengths.

[0056] Synchronize multiple sets of data on the width and strength of the weld seams to the laser control analysis model. The laser control analysis model performs internal calculations and analysis based on the synchronized weld seam width and welding strength data. Based on the model analysis results, optimize the various control parameters in the welding process to ensure that the welding quality is maximized while meeting the requirements of the weld seam width and welding strength. Extract multiple sets of local welding control parameters from the optimized control parameters. These parameters include laser power, welding speed, focal length, gas flow, etc. Serialize multiple sets of local welding control parameters in the order of the welding trajectory, match and correspond the local parameters with the trajectory parameters, and ensure that each trajectory point has corresponding control parameters. Integrate the serialized welding control parameters with the welding trajectory parameters to form a complete set of welding control parameters.

[0057] When fitting the welding trajectory parameters and welding control parameters, the appropriate fitting method is selected according to the specific characteristics of the welding trajectory parameters and welding control parameters. These characteristics include data distribution, noise level, nonlinearity, etc. Fitting methods include linear regression, polynomial regression, spline interpolation, neural network, etc. After selecting the fitting method, the corresponding mathematical model or algorithm is established. The model or algorithm describes the relationship between the welding trajectory parameters and the welding control parameters. For example, when using a neural network for fitting, the parameters such as the number of layers, the number of neurons, and the activation function of the network are determined, and the network structure is constructed. After the model or algorithm is established, the model parameters are adjusted by gradient descent, least squares method, etc. to minimize the difference between the predicted value and the actual value, so that the model better fits the actual data and improves the accuracy of the prediction. The fitting process is iterated and optimized multiple times. In each iteration, the model parameters are adjusted according to the current fitting results, and the difference between the predicted value and the actual value is recalculated. Through multiple iterations, the optimal model parameters are gradually found to achieve the best fitting effect. After the fitting is completed, the fitting effect is evaluated by calculating the goodness of fit, residual analysis, etc. Finally, the fitted welding control parameters are mapped to the welding trajectory to generate the welding path control.

[0058] Furthermore, the welding repair analysis module performs welding repair analysis according to the real-time interference fringe sequence to obtain repair path control, and the method further includes: The welding repair analysis module includes a welding quality identification submodule and a repair path analysis submodule; The welding quality identification submodule performs dynamic identification of welding deviations according to the real-time interference fringe sequence to obtain a welding deviation node set, wherein the welding deviation node set includes K deviation node position information and K deviation node image features; The repair path analysis submodule performs welding repair analysis according to the welding deviation node set to obtain a welding repair control set; The welding repair control set is fitted to the welding trajectory parameters to obtain the repair path control.

[0059] In the embodiment of the present application, the welding repair analysis module identifies welding quality problems and performs corresponding repair path analysis. The welding repair analysis module is composed of two submodules: a welding quality identification submodule and a repair path analysis submodule.

[0060] The main responsibility of the welding quality identification submodule is to dynamically identify welding deviations based on the real-time interference fringe sequence. The welding quality identification submodule uses image processing to analyze the real-time interference fringe sequence and dynamically identify the deviations that occur during the welding process. This includes identifying defects such as unevenness, lack of fusion, and slag inclusions in welded joints. Once the welding deviation is identified, the welding quality identification submodule generates a welding deviation node set. The welding deviation node set contains K deviation node position information and K deviation node image features. The deviation node position information refers to the specific position of the deviation on the welding trajectory, and the deviation node image feature is a visual description of the deviation.

[0061] The repair path analysis submodule performs welding repair analysis based on the welding deviation node set. According to the image features in the welding deviation node set, the submodule analyzes the type and degree of deviation. The deviation types include lack of fusion, slag inclusion, etc., and the deviation degree includes the size of the deviation, the scope of influence, etc.

[0062] Based on the type and degree of deviation, the repair path analysis submodule selects an appropriate repair strategy. The repair strategy includes adjusting welding parameters, changing welding trajectories, etc., where adjusting welding parameters includes adjusting welding speed, laser power, etc. Then, according to the selected repair strategy, the repair path analysis submodule generates a welding repair control set. The welding repair control set contains all the control parameters and instructions required in the repair process.

[0063] After obtaining the welding repair control set, it is fitted to the welding trajectory parameters to obtain the repair path control. In the fitting process, the control parameters in the welding repair control set are mapped to the corresponding welding trajectory parameters, and the repair instructions are matched with the specific position, speed and other parameters of the welding trajectory. By adjusting the welding trajectory parameters, the repair path control is generated.

[0064] Further, the welding quality identification submodule performs dynamic identification of welding deviation according to the real-time interference fringe sequence to obtain a welding deviation node set, wherein the welding deviation node set includes K deviation node position information and K deviation node image features, and the method further includes: Classifying a plurality of real-time interference fringe images of the real-time interference fringe sequence based on the overlap of interference fringe to obtain a plurality of groups of real-time interference fringe images, wherein the plurality of real-time interference fringe images have a plurality of image acquisition position identifiers; Preset a frequency threshold, and screen and obtain H groups of abnormal interference fringe images from the multiple groups of real-time interference fringe images based on the preset frequency threshold; splicing a plurality of real-time interference fringe images of the real-time interference fringe sequence based on the overlap of interference fringe to obtain an interference fringe splicing image, and calculating the spacing between adjacent interference fringe based on the interference fringe splicing image to obtain a plurality of interference fringe spacings; Clustering the plurality of interference fringe spacings to obtain a plurality of groups of interference fringe spacings, and screening the plurality of groups of interference fringe spacings to obtain N groups of interference fringes with abnormal spacings according to the preset frequency threshold; The H groups of abnormal interference fringes and the N groups of abnormal spacing interference fringes are divided into groups to obtain the K deviation node image features, and image acquisition site identification calls are performed on the multiple real-time interference fringe images based on the K deviation node image features to obtain the K deviation node position information.

[0065] In the embodiment of the present application, the acquired real-time interference fringe image is preprocessed to eliminate noise, enhance image contrast, and highlight the characteristics of the interference fringe. By using image processing technology, the information of the interference fringe is extracted from each preprocessed real-time interference fringe image through edge detection, threshold segmentation and other steps, and the interference fringe is separated from the background or other irrelevant information.

[0066] For any two real-time interference fringe images, the overlap is calculated by comparing the overlap of the interference fringes in terms of their shape, position, density and other characteristics. The overlap is calculated using a similarity measurement algorithm, such as Euclidean distance, correlation coefficient, etc. Based on the calculated overlap of the interference fringes, multiple images in the real-time interference fringe sequence are divided into different groups. When dividing, a overlap threshold is set by comparison. When the overlap of two images exceeds the threshold, they are grouped together. The overlap threshold is set based on historical experience.

[0067] Through the above steps, multiple groups of real-time interference fringe images with similar interference fringe patterns are obtained. Since each real-time interference fringe image carries an image acquisition position identifier, these identifiers are associated with the classification results to determine which position or area on the welding track each group of interference fringe images corresponds to.

[0068] The preset frequency threshold is a key parameter set in the process of screening abnormal interference fringe images, which is used to distinguish normal interference fringe images from abnormal interference fringe images. The preset frequency threshold is determined based on historical data, empirical knowledge or expert judgment. By statistically analyzing the interference fringe images of a large number of welding processes, the frequency of occurrence of various interference fringe images under normal welding conditions is determined. Based on these statistical data, a reasonable threshold is set, and images with a frequency lower than the threshold are considered abnormal.

[0069] When screening abnormal interference fringe images based on a preset frequency threshold, multiple groups of real-time interference fringe images are counted, and the frequency of each group of images in the entire interference fringe sequence is calculated by traversing the interference fringe sequence and counting the number of occurrences of each group of images. Next, the calculated frequency of each group of interference fringe images is compared with the preset frequency threshold. If the frequency of a group of images is lower than the threshold, the group of images is marked as an abnormal interference fringe image. Through comparison and screening, H groups of abnormal interference fringe images are finally obtained.

[0070] When stitching multiple real-time interference fringe images of a real-time interference fringe sequence, select an image from the real-time interference fringe sequence as a reference image, which has clear and stable interference fringes and can represent a typical state of the welding process. For each image in the sequence, compare it with the reference image, and find the overlapping area between them by comparing the similarity of feature points, edge information or interference fringes in the image. According to the determined overlapping area, translate, rotate or scale each image to ensure that they can be seamlessly connected when stitched. The aligned and transformed images are stitched according to their time sequence or spatial position in the welding process to form an interference fringe stitching image.

[0071] The interference fringe stitching image is subjected to image processing techniques such as edge detection and binarization to extract clear interference fringes. Among the extracted interference fringes, the center line of each interference fringe is further identified and extracted. For the adjacent interference fringe center lines, the distance between the center lines is measured by measuring the pixel distance, and the distance between them is calculated as the interference fringe spacing in combination with the calibration information of the image acquisition system. Repeat the above steps to calculate the spacing of all adjacent interference fringes in the stitching image, and obtain the values ​​of multiple interference fringe spacings.

[0072] When clustering multiple interference fringe spacings, first preprocess the data of multiple interference fringe spacings to remove noise data, process outliers or fill missing values ​​to ensure the accuracy and completeness of the data. According to the characteristics of the interference fringe spacing data and the analysis requirements, select a suitable clustering algorithm, such as K-means clustering, hierarchical clustering, density clustering, etc. to cluster multiple interference fringe spacings. The algorithm divides them into different groups according to the similarity of the data. For each group of interference fringe spacings obtained after clustering, the frequency of its occurrence in the entire data set is counted by traversing the data and counting the number of spacings in each group. The calculated frequency of each group of spacings is compared with the preset frequency threshold. The group with a frequency lower than the preset threshold is marked as abnormal spacing interference fringes. Through the screening process, N groups of abnormal spacing interference fringes are finally obtained.

[0073] When grouping H groups of abnormal interference fringes and N groups of abnormal spacing interference fringes, first, match the H groups of abnormal interference fringes with the N groups of abnormal spacing interference fringes based on their corresponding relationship in time series or spatial position. Through matching, determine which abnormal interference fringes are associated with which abnormal spacing. After the matching is completed, the groups are split according to the similarity of abnormal interference fringes and abnormal spacing. The similarity is evaluated based on factors such as the morphology of the interference fringes and the stability of the spacing. Through splitting, similar abnormal interference fringes and abnormal spacing are combined into new groups. For each split group, select image features that can represent its deviation characteristics. Image features include the morphology, direction, density, contrast, etc. of the interference fringes. Use image processing techniques such as edge detection, filtering, morphological analysis, etc. to extract the selected features from the interference fringe images of each group. Finally, the extracted features are integrated to form the deviation node image features of each group.

[0074] When calling the image acquisition site identification in multiple real-time interference fringe images based on the image features of K deviation nodes, the mapping relationship between the image acquisition site identification and the real-time interference fringe image is established by recording the position information of each image at the time of acquisition, such as coordinates, timestamps, etc. Then, in multiple real-time interference fringe images, the extracted K deviation node image features are used for matching through template matching, feature point matching and other methods to find areas similar to the deviation node features. After finding the matching areas, the location information of these areas is obtained by calling the corresponding image acquisition site identification. These location information represent the specific locations of the K deviation nodes. Finally, the obtained K deviation node location information is integrated to form a deviation node location information set.

[0075] Furthermore, the method further comprises: Obtaining a sample repair control parameter set by taking the material parameter and the thickness parameter as constraints, wherein the sample repair control parameter set includes a density repair control parameter set and a shape repair control parameter set; and The density repair control parameter set includes a plurality of sample abnormal stripe densities, a plurality of sample standard stripe densities and a plurality of sample repair control parameters; and The shape restoration control parameter set includes a plurality of sample abnormal stripe shapes, a plurality of sample standard stripe shapes, and a plurality of sample restoration control parameters; A first welding repair branch is constructed based on the density repair control parameter set, a second welding repair branch is constructed based on the shape repair control parameter set, and the first welding repair branch and the second welding repair branch are connected in parallel to generate a welding repair analysis network.

[0076] In the embodiment of the present application, a sample repair control parameter set matching the current welding task is called from a pre-stored sample database with material parameters and thickness parameters as constraints. This parameter set contains repair control parameters for different welding deviations. The sample repair control parameter set includes a density repair control parameter set and a shape repair control parameter set.

[0077] The density repair control parameter set includes a plurality of sample abnormal stripe densities, a plurality of sample standard stripe densities, and a plurality of sample repair control parameters.

[0078] The sample abnormal fringe density is extracted from historical welding data, representing the abnormal interference fringe density under different welding deviation conditions. Multiple sample standard fringe densities are interference fringe densities obtained under normal welding conditions, for comparison and reference. Multiple sample repair control parameters are derived through analysis and optimization algorithms based on the above abnormal fringe density and standard fringe density, and are used to guide density adjustment during welding repair.

[0079] The shape restoration control parameter set includes a plurality of sample abnormal stripe shapes, a plurality of sample standard stripe shapes, and a plurality of sample restoration control parameters.

[0080] The sample abnormal fringe shape is extracted from historical data and represents the abnormal interference fringe shape under different welding deviation conditions. The sample standard fringe shape is the interference fringe shape obtained under normal welding conditions, which is used as a comparison and reference. The sample restoration control parameters are derived through analysis and optimization algorithms based on the abnormal fringe shape and the standard fringe shape, and are used to guide the shape adjustment during the welding repair process.

[0081] When constructing the first welding repair branch based on the density repair control parameter set, key repair control parameters are extracted from the density repair control parameter set, and the repair control parameters include abnormal stripe density, standard stripe density and corresponding repair control parameters. These parameters are preprocessed, such as standardization and normalization, to ensure the consistency and comparability of the parameters. Then, a density repair model is constructed based on the extracted density repair control parameters using machine learning or deep learning algorithms, such as support vector machines. The model predicts the corresponding repair control parameters based on the input abnormal stripe density to guide the adjustment of density during the welding process. Through the above training process, the first welding repair branch is formed. This branch receives density data during the welding process in real time and calculates the repair control parameters through the density repair model.

[0082] When constructing the second welding repair branch based on the shape repair control parameter set, key repair control parameters are extracted from the shape repair control parameter set, and the repair control parameters include abnormal stripe shapes, standard stripe shapes, and corresponding restoration control parameters. These parameters are preprocessed to eliminate noise and outliers to ensure the validity and accuracy of the parameters. A shape repair model is constructed based on the extracted shape repair control parameters using machine learning or deep learning algorithms, such as support vector machines. The model can predict the corresponding restoration control parameters based on the input abnormal stripe shape to guide the adjustment of the shape during the welding process. Through the above training process, a second welding repair branch is formed. This branch receives the shape data during the welding process in real time, calculates the restoration control parameters through the shape repair model, and outputs these parameters to the welding control system to adjust the welding parameters and achieve shape repair.

[0083] The first welding repair branch and the second welding repair branch are connected in parallel to form a welding repair analysis network. During the welding repair process, the welding repair analysis network respectively calls the first welding repair branch and the second welding repair branch for calculation and repair according to the welding data received in real time.

[0084] Further, the repair path analysis submodule performs welding repair analysis according to the welding deviation node set to obtain a welding repair control set, and the method further includes: Based on the preset frequency threshold, the plurality of groups of real-time interference fringe images are screened to obtain real-time standard interference fringes, and based on the preset frequency threshold, the plurality of groups of interference fringe spacings are screened to obtain real-time standard fringe density; The K deviation node image features, the real-time standard interference fringes and the real-time standard fringes density are input into the welding repair analysis network to obtain K welding repair controls, and the K welding repair controls constitute the welding repair control set.

[0085] In the embodiment of the present application, when real-time standard interference fringes are screened based on a preset frequency threshold, the frequency of each real-time interference fringe image appearing in multiple sets of data is first counted by comparing the unique identifier or feature of each image. Then the frequency of each image is compared with the preset frequency threshold. When the frequency of a real-time interference fringe image exceeds the preset threshold, it is identified as a real-time standard interference fringe. For multiple sets of interference fringe spacing data, the fringe density of each group is calculated by measuring the distance between adjacent fringes and calculating the number of fringes per unit length.

[0086] Similar to screening real-time standard interference fringes, the fringe density of each group is compared with the preset frequency threshold. When the density value of the spacing of a group of interference fringes is within the preset frequency threshold, it is identified as the real-time standard fringe density.

[0087] The previously extracted K deviation node image features, the screened real-time standard interference fringes, and the real-time standard fringe density are integrated to form a complete data set. The integrated data set is input into the pre-built welding repair analysis network. The welding repair analysis network calculates K welding repair control parameters based on the input deviation node image features, real-time standard interference fringes, and real-time standard fringe density.

[0088] Finally, the obtained K welding repair control parameters are integrated into a welding repair control set.

[0089] In the embodiments of the present application, in summary, the embodiments of the present application have at least the following technical effects: The present application interactively obtains target design information of a target assembly and a target sub-design information set of a sub-component set, wherein the target assembly is obtained by laser welding the sub-component set; performs clamping positioning analysis according to the target design information and the target sub-design information to obtain a clamping feature point set and a workpiece alignment vector set; performs welding control analysis according to the target design information and the target sub-design information to obtain welding path control, wherein the welding path control includes welding trajectory parameters and welding control parameters; pre-constructs a welding repair analysis module, wherein the welding repair analysis module is communicatively connected with a laser interferometer of a target welding component; the target clamping component clamps the sub-components of the sub-component set with the clamping feature point set and the workpiece alignment vector set as constraints to obtain a to-be-welded entity, wherein the target welding component and the target clamping component constitute an automated laser welding device; during the process of the target welding component performing automated welding on the to-be-welded entity with the welding path control as a constraint, the laser interferometer collects and feeds back a real-time interference fringe sequence to the welding repair analysis module; the welding repair analysis module performs welding repair analysis according to the real-time interference fringe sequence to obtain a repair path control, and the target welding component performs feedback control optimization on the to-be-welded entity with the repair path control as a constraint. The present invention solves the technical problems of insufficient laser welding control accuracy and poor stability in the prior art. It realizes precise control of the welding process by real-time monitoring of key parameters in the welding process and real-time adjustment of welding parameters according to the monitoring results, thereby achieving the technical effect of improving welding quality and efficiency.

[0090] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above-mentioned specific embodiments of this specification are described. Other embodiments are within the scope of the attached claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0091] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application. This specification and the drawings are merely exemplary illustrations of the present application and are deemed to have covered any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is intended to include these modifications and variations.

Claims

1. A feedback control method for laser welding, characterized in that: The method comprises: Interactively obtaining target design information of a target assembly and a target sub-design information set of a sub-component set, wherein the target assembly is obtained by laser welding the sub-component set; Perform clamping positioning analysis according to the target design information and the target sub-design information to obtain a clamping feature point set and a workpiece alignment vector set; Performing welding control analysis according to the target design information and the target sub-design information to obtain welding path control, wherein the welding path control includes welding trajectory parameters and welding control parameters; Pre-building a welding repair analysis module, wherein the welding repair analysis module is communicatively connected to a laser interferometer of a target welding assembly; The target clamping component clamps the subcomponents of the subcomponent set with the clamping feature point set and the workpiece alignment vector set as constraints to obtain a to-be-welded entity, wherein the target welding component and the target clamping component constitute an automated laser welding device; During the process in which the target welding assembly performs automated welding on the entity to be welded with the welding path control as a constraint, the laser interferometer collects and feeds back a real-time interference fringe sequence to the welding repair analysis module; The welding repair analysis module performs welding repair analysis according to the real-time interference fringe sequence to obtain a repair path control, and the target welding assembly performs feedback control optimization on the entity to be welded with the repair path control as a constraint.

2. The method according to claim 1, characterized in that Performing clamping positioning analysis according to the target design information and the target sub-design information to obtain a clamping feature point set and a workpiece alignment vector set, the method further comprising: Perform digital modeling and restoration of the target assembly according to the target design information to generate a target assembly model; Performing digital modeling restoration of the subcomponent set according to the target sub-design information to generate a target subcomponent model set, wherein the subcomponent set includes M subcomponents, the target sub-design information includes M subcomponent design information and M assembly association information of the M subcomponents, and the target subcomponent model set includes M subcomponent models of the M subcomponents; Performing welding apparent center point recognition on the M sub-component models to obtain M clamping feature points, wherein the M clamping feature points constitute the clamping feature point set; Performing component edge feature point recognition on the M sub-component models to obtain M edge feature points; According to the M assembly association information, the M sub-component models are assembled and fitted to the target assembly model, and the M clamping feature points are used as workpiece alignment constraints, and the M edge feature points are used as alignment vector constraints to generate M workpiece alignment vectors, which constitute the workpiece alignment vector set.

3. The method according to claim 2, characterized in that The method further comprises: According to the M assembly association information, the M subcomponent models are assembled and fitted to the target assembly model, and the target assembly model is used as a subtracted set and the M subcomponent models are used as a subtracted set to perform a difference set solution to obtain a weld frame model; Performing connection point identification on the weld frame model, and decomposing the weld frame model into multiple welding tracks based on the connection point identification results; Based on the continuity of welding trajectories, the connection schemes of the multiple welding trajectories are enumerated in the weld frame model to obtain multiple groups of welding trajectory connection schemes and multiple welding space jump distances; The plurality of welding space jump distances are serialized and minimum value calls are performed to obtain the welding trajectory parameters.

4. The method according to claim 3, characterized in that The method further comprises: Interactively obtain material parameters and thickness parameters of the target assembly, and obtain a sample welding control parameter set by calling the material parameters and the thickness parameters as constraints, wherein the sample welding control parameter set includes a plurality of sample weld parameters, a plurality of sample welding strengths, and a plurality of sample welding control parameters; Building a laser control analysis model based on a CNN neural network, and using the sample welding control parameter set to perform training optimization on the laser control analysis model; Preset a multi-level weld width constraint, and traverse the multiple welding tracks based on the multi-level weld width constraint to divide the multiple welding tracks into multiple groups of connecting weld widths at two levels; Interactively obtaining multi-region welding strength constraints of the target assembly, and dividing the multi-region welding strength constraints according to the multiple welding trajectories to obtain multiple groups of connection welding strengths; Synchronize the multiple groups of connection weld widths and the multiple groups of connection weld strengths to the laser control analysis model to obtain multiple groups of local welding control parameters, and sequence the multiple groups of local welding control parameters according to the welding trajectory parameters to generate the welding control parameters; The welding trajectory parameters and the welding control parameters are fitted to generate the welding path control.

5. The method according to claim 4, characterized in that The welding repair analysis module performs welding repair analysis according to the real-time interference fringe sequence to obtain repair path control, and the method further includes: The welding repair analysis module includes a welding quality identification submodule and a repair path analysis submodule; The welding quality identification submodule performs dynamic identification of welding deviations according to the real-time interference fringe sequence to obtain a welding deviation node set, wherein the welding deviation node set includes K deviation node position information and K deviation node image features; The repair path analysis submodule performs welding repair analysis according to the welding deviation node set to obtain a welding repair control set; The welding repair control set is fitted to the welding trajectory parameters to obtain the repair path control.

6. The method according to claim 5, characterized in that The welding quality identification submodule performs dynamic identification of welding deviation according to the real-time interference fringe sequence to obtain a welding deviation node set, wherein the welding deviation node set includes K deviation node position information and K deviation node image features, and the method further includes: Classifying a plurality of real-time interference fringe images of the real-time interference fringe sequence based on the overlap of interference fringe to obtain a plurality of groups of real-time interference fringe images, wherein the plurality of real-time interference fringe images have a plurality of image acquisition position identifiers; Preset a frequency threshold, and screen and obtain H groups of abnormal interference fringe images from the multiple groups of real-time interference fringe images based on the preset frequency threshold; splicing a plurality of real-time interference fringe images of the real-time interference fringe sequence based on the overlap of interference fringe to obtain an interference fringe splicing image, and calculating the spacing between adjacent interference fringe based on the interference fringe splicing image to obtain a plurality of interference fringe spacings; Clustering the plurality of interference fringe spacings to obtain a plurality of groups of interference fringe spacings, and screening the plurality of groups of interference fringe spacings to obtain N groups of interference fringes with abnormal spacings according to the preset frequency threshold; The H groups of abnormal interference fringes and the N groups of abnormal spacing interference fringes are divided into groups to obtain the K deviation node image features, and image acquisition site identification calls are performed on the multiple real-time interference fringe images based on the K deviation node image features to obtain the K deviation node position information.

7. The method according to claim 6, characterized in that The method further comprises: Obtaining a sample repair control parameter set by taking the material parameter and the thickness parameter as constraints, wherein the sample repair control parameter set includes a density repair control parameter set and a shape repair control parameter set; and The density repair control parameter set includes a plurality of sample abnormal stripe densities, a plurality of sample standard stripe densities and a plurality of sample repair control parameters; and The shape restoration control parameter set includes a plurality of sample abnormal stripe shapes, a plurality of sample standard stripe shapes, and a plurality of sample restoration control parameters; A first welding repair branch is constructed based on the density repair control parameter set, a second welding repair branch is constructed based on the shape repair control parameter set, and the first welding repair branch and the second welding repair branch are connected in parallel to generate a welding repair analysis network.

8. The method according to claim 7, characterized in that The repair path analysis submodule performs welding repair analysis according to the welding deviation node set to obtain a welding repair control set, and the method further includes: Based on the preset frequency threshold, the plurality of groups of real-time interference fringe images are screened to obtain real-time standard interference fringes, and based on the preset frequency threshold, the plurality of groups of interference fringe spacings are screened to obtain real-time standard fringe density; The K deviation node image features, the real-time standard interference fringes and the real-time standard fringes density are input into the welding repair analysis network to obtain K welding repair controls, and the K welding repair controls constitute the welding repair control set.

Images