Subframe automatic processing method, system and electronic device

Abstract

The application provides a sub-frame automatic processing method, system and electronic equipment, comprising generating a bending program of a bending device based on attribute data; and driving the bending device to perform bending operation by using the bending program to obtain a plastic deformation plate semi-finished product; and extracting an actual bending angle and a curved surface twist degree based on point cloud data of the plate semi-finished product to correct preset cutting parameters of a cutting device; and driving the cutting device to perform cutting operation on the plate semi-finished product with an adaptive cutting path to obtain sub-frame parts in response to the end of the cutting operation; and based on the actual pose and the preset assembly model, splicing each sub-frame part to obtain a sub-frame assembly sketch; generating initial welding parameters based on the joint state of the sub-frame assembly sketch, and controlling the welding device to weld each joint based on the initial welding parameters. The application significantly improves the manufacturing precision of the sub-frame through step-by-step compensation of each process error.

Description

Technical Field

[0001] This invention relates to the field of intelligent manufacturing technology, and more specifically, to an automated processing method, system, and electronic equipment for subframes. Background Technology

[0002] With the continuous improvement of industrial automation and intelligence, machine vision technology is being used more and more widely in the manufacturing industry, especially in precision machining fields such as automotive subframes.

[0003] In the subframe process of "bending before cutting", the bent sheet metal is prone to angular deviation and surface distortion due to material springback, causing the reference surface for subsequent cutting to deviate from the theoretical model. If the cutting is still carried out according to the preset procedure, the bending error will be solidified in the workpiece and further amplified in subsequent assembly, forming a cumulative error that is difficult to correct. Summary of the Invention

[0004] In view of this, the purpose of the present invention is to provide an automated processing method, system and electronic equipment for subframes, so as to reduce errors in the subframe processing and improve the processing accuracy of the subframes.

[0005] In a first aspect, an automated subframe processing method is provided, applied to a subframe processing system. The subframe processing system includes a central processing unit, bending equipment, cutting equipment, assembly platform, welding equipment, and visual recognition devices installed at each processing step. The method includes: Obtain the property data of the sheet metal to be processed for manufacturing the subframe; Based on attribute data, a bending program for the bending equipment is generated; and the bending program is used to drive the bending equipment to perform bending operations to obtain a plastically deformed sheet metal semi-finished product. In response to the end of the bending operation, the visual recognition device at the bending process collects the point cloud data of the semi-finished sheet material, and extracts the actual bending angle and surface distortion based on the point cloud data of the semi-finished sheet material to obtain the first processing data package. Based on the first processing data package, the preset cutting parameters of the cutting equipment are corrected to generate an adaptive cutting path that adapts to the actual posture of the current sheet metal semi-finished product; and the cutting equipment is driven to perform cutting operation on the sheet metal semi-finished product using the adaptive cutting path to obtain the subframe parts. In response to the end of the cutting operation, the actual pose of each subframe component is collected by the vision recognition device above the assembly platform; and based on the actual pose and the preset assembly model, the subframe components are assembled to obtain the prototype of the subframe assembly. Based on the joint status of the subframe assembly prototype, initial welding parameters are generated, and based on the initial welding parameters, the welding equipment is controlled to weld each joint to complete the final forming of the subframe.

[0006] Optionally, based on attribute data, generating a bending program for the bending equipment includes: Input the attribute data of the sheet material to be processed and the preset bending parameters into the pre-trained springback prediction model, and output the predicted springback deformation for the current sheet material to be processed; the attribute data includes at least the material and thickness of the sheet material; The rebound compensation angle is determined based on the difference between the predicted rebound deformation and the theoretical rebound deformation. The preset bending parameters are corrected based on the springback compensation angle, and a bending program is generated based on the corrected preset bending parameters.

[0007] Optionally, the preset bending parameters are corrected based on the springback compensation angle, and a bending program is generated based on the corrected preset bending parameters, including: The visual recognition device at the feed point collects the three-dimensional point cloud data of the board to be processed; and based on the three-dimensional point cloud data of the board to be processed, several bending lines are determined and the spatial coordinates of each bending line are extracted. Based on the distribution of the spatial coordinates of the bend lines, the execution order of each bend line is determined; For each bend line in the execution sequence, the springback compensation angle and the preset bend angle are algebraically superimposed to obtain the target bend angle; Based on the springback compensation angle and the effective length of the current bending line, the corresponding bending pressure compensation coefficient is matched in the preset process expert database, and the preset bending pressure parameters are corrected using the bending pressure compensation coefficient. According to the execution order, the spatial coordinates, preset bending direction, target bending angle and corrected bending pressure parameters corresponding to each bending line are sequentially arranged and encapsulated to generate an automated bending operation program containing multi-step sequential execution instructions.

[0008] Optionally, the actual bending angle and surface distortion are extracted from the point cloud data of the semi-finished sheet material to obtain the first processing data package, which includes: The point cloud data of the semi-finished board is spatially registered with the theoretical three-dimensional model of the semi-finished board to eliminate rigid body displacement errors. The normal distance and direction vector from each data point in the registered point cloud to the surface of the theoretical 3D model are calculated to obtain the deviation mapping field characterizing the local deformation distribution of the plate. Based on the deviation mapping field, the feature plane normal vectors of the regions on both sides of the bending line are extracted, and the included angle is calculated to obtain the actual bending angle. The statistical bias map field shows the fluctuation characteristics along the bending direction, and quantifies the fluctuation characteristics as the surface distortion. The actual bending angle and surface distortion are encapsulated into the first processing data packet.

[0009] Optionally, based on the first processing data package, the preset cutting parameters of the cutting equipment are corrected to generate an adaptive cutting path that matches the actual posture of the current semi-finished board, including: Based on the actual bending angle and surface distortion in the first processing data package, a nonlinear spatial transformation matrix is ​​constructed from the theoretical coordinate system to the actual deformed coordinate system; By using a nonlinear spatial transformation matrix, the coordinate points of the original cutting trajectory in the preset cutting parameters are mapped and corrected point by point to obtain the corrected spatial trajectory. Based on the corrected spatial trajectory, the feed speed of the cutting head and the laser focus position are replanned to generate an adaptive cutting path.

[0010] Optionally, based on the actual position and a preset assembly model, the various subframe components are assembled to obtain a preliminary subframe assembly, including: In the virtual assembly space, with the goal of minimizing the gap between adjacent parts, the optimal pose transformation matrix of each subframe part relative to the preset assembly model is calculated using the iterative nearest point matching algorithm. Based on the optimal pose transformation matrix, the pose of each subframe component on the assembly platform is adjusted to assemble the subframe components and obtain the prototype of the subframe assembly.

[0011] Optionally, based on the joint condition of the subframe assembly prototype, initial welding parameters are generated, and based on these initial welding parameters, the welding equipment is controlled to weld each joint, including: Images of the seams between subframe components are captured using a visual recognition device at the welding equipment. Based on the seam image, initial welding parameters are generated using a large model; the initial welding parameters include at least welding current, welding speed, and welding direction. The welding equipment is controlled to weld along the joint trajectory according to the initial welding parameters; and the positional deviation between the welding torch and the joint in the welding equipment is monitored in real time during the welding process so as to adjust the direction of the welding torch in real time.

[0012] Optionally, the method also includes: In response to the end of the cutting operation, the actual contours of each subframe component are collected based on the visual recognition device at the cutting equipment. The actual contours of each subframe component are compared with the expected contours after the cutting parameters are corrected, and a second processing data package containing information on cutting edge deviation, cut perpendicularity, and local thermal deformation is generated and stored in a pre-built process database.

[0013] In a second aspect, an automated subframe processing system is provided, including a central processing unit, bending equipment, cutting equipment, assembly platform, welding equipment, and visual recognition devices installed at each processing step. The central processing unit is communicatively connected to bending equipment, cutting equipment, hoisting and positioning equipment of assembly platform, welding equipment, and visual recognition devices installed at each processing step; and executes any of the methods in the first aspect.

[0014] Thirdly, an electronic device is provided, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; A processor, when executing a program stored in memory, implements any of the methods of the first aspect.

[0015] This invention provides an automated processing method, system, and electronic device for subframes. The method involves acquiring attribute data of the sheet metal to be processed for manufacturing the subframe; generating a bending program for a bending machine based on the attribute data; driving the bending machine to perform bending operations using the bending program to obtain a plastically deformed semi-finished sheet metal; responding to the end of the bending operation, acquiring point cloud data of the semi-finished sheet metal using a visual recognition device at the bending stage, and extracting the actual bending angle and surface distortion from the point cloud data to obtain a first processing data package; and adjusting the preset cutting parameters of the cutting machine according to the first processing data package. The invention first predicts springback deformation and compensates for bending parameters in advance to reduce bending errors. It then uses measured deformation data after bending to correct the cutting path in real time, solving the problem of reduced cutting accuracy caused by material springback and torsion, and eliminating the accumulation of bending errors. Simultaneously, through vision-guided flexible assembly and intelligent welding, it further eliminates the impact of accumulated errors on the final product. By compensating for errors at each stage of the process, the manufacturing accuracy of the subframe is significantly improved.

[0016] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This diagram illustrates the structure of an automated subframe processing system provided in an embodiment of the present invention. Figure 2 A flowchart of an automated subframe manufacturing method provided by an embodiment of the present invention is shown; Figure 3 A flowchart of another automated subframe manufacturing method provided by an embodiment of the present invention is shown; Figure 4 A schematic diagram of the structure of an electronic device provided in an embodiment of the present invention is shown. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0020] This invention provides an automated subframe machining method, applied to a subframe machining system, such as... Figure 1 As shown, the subframe processing system includes a central processing unit 101, a bending device 102, a cutting device 103, an assembly platform 104, a welding device 105, and a visual recognition device 106 installed at each processing step.

[0021] In this embodiment, the bending equipment is a multi-axis CNC bending machine, equipped with independently controlled left and right hydraulic cylinders and an automatically replaceable mold library, used to bend flat sheet metal into a three-dimensional structure with a specific angle and shape.

[0022] The preferred cutting equipment is a five-axis linkage laser cutting machine or a high-pressure water jet cutting machine, equipped with a high-precision motion control system, which can perform precise cutting of complex contours according to dynamic paths.

[0023] The assembly platform consists of a flexible fixture array or a multi-axis industrial robot, which has six degrees of freedom adjustment capability and can change the spatial position and posture of parts in real time according to instructions to achieve non-rigid assembly.

[0024] The welding equipment uses an adaptive welding robot that integrates a wire feeding mechanism, welding torch, and real-time tracking sensors to perform welding based on welding parameters.

[0025] The visual recognition device is configured with different types according to the needs of different processes. Specifically, a wide field-of-view 3D structured light camera is installed at the feeding point to scan the flatness and surface defects of the entire board.

[0026] A high-precision structured light 3D scanner is installed above and behind the bending machine during the bending process. After bending, the workpiece is scanned in a non-contact manner to generate high-density point cloud data (with an accuracy of ±0.05mm).

[0027] A high-speed area array CCD camera is coaxially mounted next to the cutting head at the cutting process site to monitor the quality of the cutting edge and the accuracy of the contour in real time.

[0028] A global binocular stereo vision system is suspended above the assembly platform, covering the entire assembly area, and is used to simultaneously locate the three-dimensional pose of multiple parts.

[0029] The welding equipment has an industrial camera with a narrow-band filter integrated at the front end of the welding torch, which is used to clearly capture images of the molten pool and joint gaps in a strong arc light environment.

[0030] The central processing unit (CPU), acting as the execution entity, is connected to bending equipment, cutting equipment, assembly platforms, welding equipment, and visual recognition devices installed at each processing stage. For example... Figure 2 As shown, the automated manufacturing method for the subframe includes the following steps: Step S201: Obtain the property data of the sheet metal to be processed for manufacturing the subframe.

[0031] In this embodiment of the invention, attribute data is a set of basic information describing the physicochemical properties of raw materials. Specifically, this data covers the material grade of the sheet (such as low-carbon steel, high-strength steel, aluminum alloy, etc.), thickness, tensile strength, and surface coating type.

[0032] Specifically, the central processing unit can directly read electronic work order data from the upstream ERP or MES system via the industrial Ethernet interface, or read the QR code information on the sheet material label via a barcode scanner; it can also be linked with the weighing and thickness measuring device at the feeding point for physical verification.

[0033] Step S202: Based on the attribute data, generate a bending program for the bending equipment; and use the bending program to drive the bending equipment to perform bending operations to obtain a plastically deformed sheet metal semi-finished product.

[0034] In this embodiment of the invention, the bending program is a sequence of instructions that guides the bending machine to complete the processing of a specific shape.

[0035] Specifically, the program content includes mold selection instructions, bending speed, holding pressure time, and the sequential logic of each step.

[0036] After the driving device executes the process, the flat sheet undergoes the expected plastic deformation, becoming a semi-finished product with a preliminary geometric shape. This step achieves the first morphological transformation from material data to a physical entity.

[0037] Step S203: In response to the end of the bending operation, the visual recognition device at the bending process collects the point cloud data of the semi-finished sheet material, and extracts the actual bending angle and surface distortion based on the point cloud data of the semi-finished sheet material to obtain the first processing data package.

[0038] In this embodiment of the invention, point cloud data is a dense set of three-dimensional coordinates that characterize the microscopic geometric morphology of the surface of a semi-finished product.

[0039] Specifically, the actual bending angle reflects the true angle between the two planes on both sides of the bending line, while the surface distortion quantifies the non-planar warping or spiral deformation that occurs along the bending line direction. This is a key indicator that traditional two-dimensional measurements cannot obtain.

[0040] The structured light 3D scanner at the bending process scans the semi-finished product line by line to reconstruct a complete 3D surface model.

[0041] This step not only detects conventional angular errors, but also accurately captures complex spatial distortions, providing detailed geometric basis for adaptive compensation in subsequent processes.

[0042] Step S204: Based on the first processing data package, the preset cutting parameters of the cutting equipment are corrected to generate an adaptive cutting path that adapts to the actual posture of the current sheet metal semi-finished product; and the cutting equipment is driven to perform a cutting operation on the sheet metal semi-finished product using the adaptive cutting path to obtain the subframe parts.

[0043] In this embodiment of the invention, the adaptive cutting path is a spatial trajectory that dynamically conforms to the actual deformed surface of the sheet material, rather than a rigid theoretical curve. The correction process involves a spatial transformation of the original cutting coordinates to ensure that the cutting head is always perpendicular to the actual sheet material surface and maintains a constant focal length.

[0044] For example, when the board material is significantly twisted, the originally straight cutting path is spatially corrected into a slightly curved line, and the cutting head adjusts its height in real time according to the undulation of the curved surface, thereby ensuring that the cut width is consistent and there is no slag.

[0045] This step effectively eliminates the transmission effect of previous bending deformation on cutting accuracy, avoids the accumulation of errors, and significantly improves the dimensional qualification rate of parts.

[0046] Step S205: In response to the end of the cutting operation, the actual pose of each subframe component is collected by the vision recognition device above the assembly platform; and based on the actual pose and the preset assembly model, the subframe components are assembled to obtain the prototype of the subframe assembly.

[0047] In this embodiment of the invention, the actual pose refers to the true three-dimensional coordinates (X,Y,Z) and Euler angles (Rx,Ry,Rz) of the component on the assembly platform, reflecting the random errors generated during the loading and transmission process.

[0048] Specifically, the assembly process involves using flexible actuators to move the scattered components to their optimal matching positions.

[0049] For example, even if a crossbeam component warps slightly due to thermal deformation during cutting, the system can calculate the optimal fit and fit it tightly with other longitudinal beams to form a preliminary assembly without obvious misalignment.

[0050] This step overcomes accumulated errors through vision-guided flexible assembly, achieving high-precision tooling-free assembly.

[0051] Step S206: Based on the joint status of the subframe assembly prototype, generate initial welding parameters, and based on the initial welding parameters, control the welding equipment to weld each joint to complete the final forming of the subframe.

[0052] In this embodiment of the invention, the joint condition includes the weld gap width, misalignment, bevel angle, and surface cleanliness.

[0053] Specifically, the initial welding parameters include welding current, voltage, wire feed speed, oscillation amplitude, and welding direction.

[0054] The vision sensor at the welding equipment takes high-definition pictures of the joint before the arc is started. The central processing unit matches the best welding specifications based on the image features and controls the robot to perform the welding operation.

[0055] For example, when the system detects that the gap between the welds has increased from the standard 2mm to 3mm, it automatically increases the welding current and the swing amplitude to fill the molten pool and prevent incomplete fusion defects.

[0056] This step enables intelligent closed-loop control of the welding process, ensuring the structural strength and consistency of the subframe under complex working conditions.

[0057] This embodiment constructs an adaptive subframe processing mode. First, the springback deformation is predicted, and bending parameters are compensated in advance to reduce bending errors. Furthermore, the cutting path is corrected in real time using measured deformation data after bending, solving the problem of reduced cutting accuracy caused by material springback and torsion, and eliminating the accumulation of bending errors. At the same time, through vision-guided flexible assembly and intelligent welding, the impact of accumulated errors on the final product is further eliminated. By compensating for errors in each process step by step, the manufacturing accuracy of the subframe is significantly improved.

[0058] Based on the above embodiments, the bending program for the bending equipment, generated based on attribute data, includes: Step S202A: Input the attribute data of the sheet material to be processed and the preset bending parameters into the pre-trained springback prediction model, and output the predicted springback deformation for the current sheet material to be processed.

[0059] The attribute data should include at least the material and thickness of the board.

[0060] In this embodiment of the invention, the springback prediction model is an intelligent algorithm model based on a deep learning architecture (such as a convolutional neural network or a long short-term memory network), specifically designed to simulate the elastic recovery of materials after bending.

[0061] Specifically, the input features include the material, thickness, texture direction, preset bending angle, die radius, and bending speed of the sheet metal; the output is the predicted springback deformation.

[0062] In one feasible implementation, the initial training of the springback prediction model is based on a large-scale historical process dataset. Specifically, the training process includes data cleaning, feature engineering, model building, and hyperparameter tuning.

[0063] Collect massive amounts of attribute data, process parameters, and measured rebound amount triplet data from historical production, and use the backpropagation algorithm to train a neural network to learn the complex nonlinear mapping relationship between material properties and rebound behavior until the model's prediction error on the validation set converges to within the threshold.

[0064] In addition, the model has the capability for online incremental updates. Specifically, in subsequent production processes, the system uses the actual bending angle and surface distortion (i.e., the measured springback result) collected in step S203 of each round as new labeled data, which, together with the input attribute data at that time, constitutes a new sample pair.

[0065] Specifically, the central processing unit periodically (e.g., per shift or per hundred pieces) or when it detects that the prediction deviation exceeds the warning line, triggers the incremental learning algorithm to fine-tune the pre-trained model using this newly generated real-time data, updating the model weights without retraining from scratch.

[0066] For example, when it is discovered that the springback characteristics of a batch of steel have slightly shifted due to seasonal temperature and humidity changes, the model is updated in real time using the latest production data to quickly adapt to this change.

[0067] Step S202B: Determine the rebound compensation angle based on the difference between the predicted rebound deformation and the theoretical rebound deformation.

[0068] In this embodiment of the invention, the springback compensation angle is used to compensate for the increase in the bending angle between the predicted elastic recovery of the sheet material and the theoretical springback deformation.

[0069] Specifically, the central processing unit calculates the difference between the two. If the predicted value is greater than the theoretical value, a positive compensation angle is generated to increase the deflection; otherwise, it is reduced.

[0070] For example, if the theory suggests a rebound of 1.5°, while the model predicts 2.8°, then an additional compensation angle of 1.3° is determined to set the target bending angle to 91.3° (assuming the target is 90°).

[0071] This step effectively solves the problem of angular instability caused by batch-to-batch material performance fluctuations through a dynamic correction and compensation strategy.

[0072] Step S202C: Correct the preset bending parameters based on the springback compensation angle, and generate a bending program based on the corrected preset bending parameters.

[0073] This step ensures that the generated bending program is forward-looking and adaptive, guaranteeing the accuracy of the forming angle from the source.

[0074] This embodiment achieves intelligent feedforward control of the bending process by introducing a pre-trained springback prediction model. It can accurately predict the springback behavior of different materials and batches of sheet metal. By dynamically determining the springback compensation angle and correcting bending parameters, the system can automatically adapt to fluctuations in material properties, ensuring the consistency of bending angles and significantly improving the first-pass yield of the bending process. This lays a solid geometric foundation for subsequent high-precision cutting and assembly.

[0075] Based on the above embodiments, the preset bending parameters are corrected based on the springback compensation angle, and a bending program is generated based on the corrected preset bending parameters, including: Step S202C1: Collect the three-dimensional point cloud data of the board to be processed based on the visual recognition device at the feeding point; and determine several bending lines based on the three-dimensional point cloud data of the board to be processed, and extract the spatial coordinates of each bending line.

[0076] In this embodiment of the invention, the three-dimensional point cloud data completely restores the flatness, edge contour and pre-stored hole position information of the board in the feeding state.

[0077] Specifically, the bending line is the theoretical axis on the sheet metal where bending deformation needs to occur, while the spatial coordinates define its absolute position and direction on the machine tool table.

[0078] The wide-field-of-view 3D structured light camera at the feeding point performs a panoramic scan of the sheet material. Through edge detection and feature matching algorithms, it automatically identifies all bending feature lines to be processed and calculates the precise coordinates of their endpoints in the machine tool coordinate system.

[0079] For example, for an irregularly shaped plate with pre-punched holes, it can accurately identify the three bending lines that need to bypass the holes and their respective start and end point coordinates, and can accurately extract them even if the plate is slightly rotated.

[0080] Step S202C2: Determine the execution order of each bend line based on the distribution of the spatial coordinates of the bend lines.

[0081] In this embodiment of the invention, the execution order refers to the sequential logic of processing multiple bending lines, which directly determines the processing efficiency and whether interference occurs.

[0082] Specifically, the central processing unit constructs a bending process topology map, uses an interference check algorithm to simulate the processing under different sequences, and selects the optimal sequence that is collision-free, reachable by the fixture, and has the fewest flips.

[0083] For example, if bending the outer edge first would cause the clamp to fail to hold or cause a collision when bending the middle edge, the order will be automatically adjusted to bend the middle edge first and then the outer edge.

[0084] This step, through intelligent sorting, maximizes the stability and efficiency of the processing and reduces human intervention.

[0085] Step S202C3: For each bending line in the execution sequence, the springback compensation angle and the preset bending angle are algebraically superimposed to obtain the target bending angle.

[0086] Specifically, the preset bending angle is the theoretical angle required by the product design, while the springback compensation angle is a correction value predicted based on material properties and the model. The central processing unit calculates independently for each process and adds the compensation angle to the theoretical angle (usually added for bending, depending on the specific definition).

[0087] For example, for the first bending line, the theoretical angle is 45°, the compensation angle is 1.2°, and the target angle is set to 46.2°; for the second line, due to the work hardening effect of the material, the compensation angle is dynamically adjusted to 1.5°, and the target angle is set to 46.5°.

[0088] This step enables precise angle control for each pass, adapting to the dynamic changes in material properties during processing.

[0089] Step S202C4: Based on the springback compensation angle and the effective length of the current bending line, match the corresponding bending pressure compensation coefficient in the preset process expert database, and use the bending pressure compensation coefficient to correct the preset bending pressure parameters.

[0090] In this embodiment of the invention, the bending pressure compensation coefficient is a proportional factor used to adjust the output tonnage of the hydraulic system to ensure sufficient plastic deformation.

[0091] Specifically, the effective length refers to the length of the plate segment involved in deformation. The longer the length, the greater the total pressure required, but the pressure distribution per unit length is affected differently by springback.

[0092] In one feasible implementation, the optimal pressure coefficient is obtained by querying a database of process experts that stores a large number of successful cases, using the springback compensation angle and effective length as a joint index.

[0093] For example, in the case of a long bending line and a large springback compensation angle, the database matches a pressure coefficient of 1.15, and the system corrects the preset pressure of 100 tons to 115 tons to ensure sufficient plastic deformation and reduce springback.

[0094] This step combines expert experience with real-time data to ensure the scientific validity and safety of the pressure parameters.

[0095] Step S202C5: According to the execution order, the spatial coordinates, preset bending direction, target bending angle and corrected bending pressure parameters corresponding to each bending line are sequentially arranged and encapsulated to generate an automated bending operation program containing multi-step sequential execution instructions.

[0096] In this embodiment of the invention, the automated bending operation program is the final set of instructions that drives the machine tool to complete the entire set of actions.

[0097] Specifically, the timing arrangement ensures that each step is executed strictly in logical order, while encapsulation packages discrete parameters into a standard format file.

[0098] This embodiment implements personalized angle and pressure compensation for the specific working conditions (length, position, sequence) of each bending line. This fine-grained control strategy significantly improves the forming accuracy of complex subframe parts, reduces scrap caused by improper process arrangement, and achieves efficient and high-quality automated bending production.

[0099] Based on the above embodiments, the actual bending angle and surface distortion are extracted from the point cloud data of the semi-finished sheet material to obtain the first processing data package, which includes: Step S203A: Spatial registration is performed between the point cloud data of the semi-finished sheet material and the theoretical 3D model of the semi-finished sheet material to eliminate rigid body displacement errors.

[0100] In this embodiment of the invention, rigid body displacement error refers to the overall translation or rotation caused by inaccurate placement of the workpiece on the inspection table. The theoretical three-dimensional model is a digital representation of the CAD design data.

[0101] Specifically, the Iterative Closest Point (ICP) algorithm or a feature point-based matching algorithm is used to calculate the optimal rotation and translation matrix of the measured point cloud relative to the theoretical model, and the measured data is then subjected to coordinate transformation.

[0102] Step S203B: Calculate the normal distance and direction vector from each data point in the registered point cloud to the surface of the theoretical 3D model to obtain the deviation mapping field characterizing the local deformation distribution of the plate.

[0103] In this embodiment of the invention, the deviation mapping field is a data field that visually displays the degree of deviation of each point on the workpiece surface from its ideal position. The normal distance refers to the shortest distance from the point cloud to the model surface, and the direction vector indicates the direction of deviation (inward or outward).

[0104] Specifically, the central processing unit traverses each registered point, calculates its distance to the nearest model surface, and uses color spectra (such as red positive and blue negative) for visualization encoding.

[0105] For example, in the bend and rounded corner area, if a series of points show a regular positive distance, it indicates that the rebound is large; if they show disordered fluctuations, there may be local wrinkling. This step transforms the abstract point cloud data into an intuitive deformation distribution map, laying the foundation for quantitative analysis.

[0106] Step S203C: Based on the deviation mapping field, extract the feature plane normal vectors of the regions on both sides of the bending line, and calculate the included angle to obtain the actual bending angle.

[0107] In this embodiment of the invention, the feature plane normal vector is a vector that describes the spatial orientation of the flat plate portions on both sides of the bend.

[0108] The extraction area is usually selected from the flat part away from the rounded corner to avoid interference from the rounded corner transition area.

[0109] Specifically, select specific windows on both sides of the bending line in the deviation mapping field, fit two optimal planes using the least squares method, and calculate the angle between the normal vectors of these two planes.

[0110] For example, the normal vector of the left plane is (0,0,1) and the normal vector of the right plane is (0.707,0,0.707). The calculated angle is 45.2°, which is the actual bending angle.

[0111] This step eliminates surface roughness noise through plane fitting, resulting in high-precision angle measurements.

[0112] Step S203D: Statistically map the fluctuation characteristics of the deviation field along the bending line direction, and quantify the fluctuation characteristics into surface distortion.

[0113] In this embodiment of the invention, the wave characteristics reflect the inconsistency of the height of the sheet metal along the bending axis, which is a direct manifestation of torsional deformation.

[0114] The quantification process transforms these irregular fluctuations into a specific numerical indicator (such as the difference between the maximum peak and trough or the root mean square error).

[0115] Specifically, the profile curve is cut along the bending line, its straightness or flatness deviation is analyzed, and its standard deviation or extreme value difference is statistically analyzed as the torsion degree.

[0116] For example, if the deviation fluctuates between -0.5mm and +0.8mm along a 1-meter-long bend, the quantified surface distortion is 1.3mm.

[0117] This step accurately captures spiral or wavy deformations that traditional angle measurements cannot detect, providing a crucial dimension for subsequent cutting compensation.

[0118] Step S203E: Encapsulate the actual bending angle and surface distortion into a first processing data packet.

[0119] In this embodiment of the invention, the first processing data packet is a standardized data structure carrying the detection results. Specifically, the encapsulation format can be JSON, XML, or a binary stream, and it includes angle values, distortion values, confidence levels, and timestamps.

[0120] This step enables the structured output of the detection data, ensuring the reliability and compatibility of data transmission between different subsystems.

[0121] This embodiment accurately separates two different deformation modes: angular deviation and surface distortion. In particular, the proposed surface distortion quantification method fills a gap in traditional detection techniques, providing rich and reliable deformation data support for subsequent processes and ensuring the effectiveness of the adaptive compensation strategy.

[0122] Based on the above embodiments, the preset cutting parameters of the cutting equipment are corrected according to the first processing data package to generate an adaptive cutting path that adapts to the actual posture of the current semi-finished board material, including: Step S204A: Based on the actual bending angle and surface distortion in the first processing data package, construct a nonlinear spatial transformation matrix from the theoretical coordinate system to the actual deformed coordinate system.

[0123] In this embodiment of the invention, the nonlinear spatial transformation matrix is ​​a mathematical tool used to describe the complex mapping relationship between the theoretical shape and the actual deformed shape, especially for nonlinear deformations such as twisting.

[0124] Specifically, the matrix includes not only rotation and translation but also scaling and shearing components to simulate real physical deformation.

[0125] The central processing unit corrects the rotation component using the actual bending angle and constructs the shear component that varies along the axial direction using the surface twist, combining them into a high-order transformation matrix.

[0126] For example, for a sheet material with distortion, the transformation parameters of the matrix are different at the beginning and end of the bending line, thus forming a gradually changing transformation field.

[0127] Step S204B: Using a nonlinear spatial transformation matrix, the coordinate points of the original cutting trajectory in the preset cutting parameters are mapped and corrected point by point to obtain the corrected spatial trajectory.

[0128] In this embodiment of the invention, point-by-point mapping correction is the process of projecting each discrete point on the theoretical path onto the actual workpiece surface through a transformation matrix. Specifically, the original cutting trajectory is generated based on a perfect model, while the corrected spatial trajectory is a path that conforms to the actual deformed surface.

[0129] By traversing thousands of coordinate points on the original trajectory and multiplying them by a nonlinear spatial transformation matrix, their new coordinates in the actual deformed coordinate system are calculated.

[0130] This step allows the cutting path to follow the natural shape, ensuring the quality of the cut.

[0131] Step S204C: Based on the corrected spatial trajectory, replan the feed speed of the cutting head and the laser focus position to generate an adaptive cutting path.

[0132] In this embodiment of the invention, the adaptive cutting path includes not only spatial coordinates but also dynamic process parameters. Specifically, the feed rate needs to be adjusted according to the trajectory curvature and changes in the sheet thickness, and the laser focus position needs to be tracked in real time according to the undulations of the sheet surface.

[0133] In one example, the central processing unit analyzes the curvature changes of the corrected trajectory and automatically reduces speed at sharp turns to prevent overheating; at the same time, it generates Z-axis follow-up commands based on the surface height changes derived from the transformation matrix to keep the focus constant.

[0134] For example, when the cutting path passes through a bulging area caused by twisting, the system instructs the cutting head to decelerate and raise the focal point to ensure stable energy density.

[0135] This embodiment achieves height adaptation of the cutting path to the actual deformation of the sheet metal by constructing a nonlinear spatial transformation matrix and implementing point-by-point mapping correction. This solves the problem of cutting trajectory deviation caused by bending and twisting, ensuring the geometric consistency of the cutting edge with the theoretical design. Simultaneously, dynamically adjusted feed speed and focal position guarantee uniformity of cutting quality in different deformation zones, significantly reducing defects such as skew cutting and slag buildup, and improving the fitting accuracy of subframe components.

[0136] Based on the above embodiments, and based on the actual pose and a preset assembly model, the subframe components are assembled to obtain a preliminary subframe assembly, including: Step S205A: In the virtual assembly space, with the goal of minimizing the gap between adjacent parts, the optimal pose transformation matrix of each subframe part relative to the preset assembly model is calculated using the iterative nearest point matching algorithm.

[0137] In this embodiment of the invention, the virtual assembly space is a digital simulation environment constructed in computer memory, and the optimal pose transformation matrix is ​​the movement parameter that enables the parts to achieve the best fit.

[0138] Specifically, we construct an objective function, whose optimization objective is defined as minimizing the sum of the squares of all butt joint widths.

[0139] The ICP algorithm is used to continuously fine-tune the pose of each component in virtual space until the point cloud distance at the seam is minimized.

[0140] For example, if there is a wedge-shaped gap between two beams, the algorithm will automatically calculate a small rotation angle and translation amount to make the two beams fit together most tightly.

[0141] This step uses a global optimization algorithm to find the theoretically optimal assembly scheme, overcoming the limitations of single-reference positioning.

[0142] Step S205B: Based on the optimal pose transformation matrix, adjust the pose of each subframe component on the assembly platform to assemble the subframe components and obtain the prototype of the subframe assembly.

[0143] In this embodiment of the invention, pose adjustment is the process of implementing virtual calculation results into a physical object through a physical actuator.

[0144] Specifically, the central processing unit decomposes the optimal pose transformation matrix into action commands for each actuator, driving cylinders or servo motors to move the components to the calculated optimal position.

[0145] For example, the system controls the left clamp to move 2mm to the right and the right clamp to rise 0.5mm, so that the originally misaligned parts are perfectly joined together to form the prototype of the assembly to be welded.

[0146] This embodiment achieves flexible and high-precision assembly of subframe components through a close integration of virtual simulation and physical adjustment. It can automatically compensate for accumulated geometric errors from previous processes (bending, cutting), enabling precise assembly of multiple parts without the need for dedicated rigid fixtures. This not only reduces tooling costs but also significantly improves assembly efficiency and joint quality, providing an ideal joint condition for subsequent welding processes and effectively reducing welding deformation and residual stress.

[0147] To further improve assembly accuracy, after the cutting operation in step S204, a data acquisition step is added: the actual contours of each subframe component after cutting are collected by the visual recognition device at the cutting equipment, and the actual contours are compared with the expected contours after the cutting parameters are corrected. A second processing data package containing information on cutting edge deviation, cut perpendicularity and local thermal deformation, as well as a deviation report, are generated and stored in the pre-built process database.

[0148] Subsequently, during assembly, the central processing unit not only uses the macroscopic actual pose collected by the vision recognition device, but also integrates the microscopic contour deviation information in the second processing data packet to perform a secondary correction on the optimal pose transformation matrix of the parts.

[0149] Specifically, if the second processing data package shows that the cutting edge of a part has local wavy deformation or dimensional shrinkage, the system will include this local deformation in the constraint conditions when calculating the optimal pose transformation matrix, and appropriately adjust the placement angle or position of the part to compensate for the edge defects and ensure the best fit at the joint.

[0150] Specifically, a multi-objective optimization algorithm can be used to take both macroscopic pose matching and microscopic edge matching as objective functions to solve for the comprehensive optimal pose transformation matrix.

[0151] For example, if the edge of a longitudinal beam part shrinks inward by 0.2mm after cutting, the system will control the robot to push it outward by 0.2mm during assembly, or finely adjust the rotation angle so that even though the part itself has cutting errors, it can still form a perfect docking gap with other parts at the assembly interface.

[0152] This supplementary embodiment establishes a deep data association between the cutting and assembly processes by introducing a second processing data package. Specifically, it uses adjustments to the assembly posture to compensate for microscopic edge defects generated during the cutting process. This cross-process collaborative optimization further improves the assembly accuracy of the subframe assembly, ensuring high-quality joints even with slight cutting deformation of components, thereby significantly improving the structural performance and dimensional stability of the final welded product.

[0153] Based on the above embodiments, such as Figure 3 As shown, based on the joint state of the subframe assembly prototype, initial welding parameters are generated, and based on these initial welding parameters, the welding equipment is controlled to weld each joint, including: Step S206A: The visual recognition device at the welding equipment acquires images of the seams between the subframe components.

[0154] In this embodiment of the invention, the seam image is high-resolution optical data reflecting the weld root gap, misalignment, and surface oxidation.

[0155] Specifically, the visual recognition device is usually a high frame rate industrial camera installed next to the welding torch, in conjunction with a light source of a specific wavelength to eliminate arc light interference.

[0156] Before welding begins or during interpass pauses, the camera is aimed at the area to be welded to capture high-resolution images, which are then pre-processed to enhance edge features.

[0157] Step S206B: Based on the joint image, generate initial welding parameters using a large model; the initial welding parameters include at least welding current, welding speed, and welding direction.

[0158] In this embodiment of the invention, a large model refers to a deep learning model (such as the Transformer architecture) trained with massive welding process data, which has powerful feature recognition and parameter inference capabilities.

[0159] Specifically, the large model can understand the nonlinear relationship between geometric features in the image and weld quality. By inputting the joint image into the large model, the model outputs a recommended combination of process parameters.

[0160] For example, in response to the wide gap and misalignment images mentioned above, the large model determines that a larger amount of welding is required, and therefore generates a parameter combination of "current 220A, speed 40cm / min, with slight oscillation".

[0161] This step leverages the generalization capabilities of artificial intelligence, enabling it to handle various complex and changing joint conditions, outperforming traditional rule-based matching.

[0162] Step S206C: Control the welding equipment to weld along the joint trajectory according to the initial welding parameters; and monitor the positional deviation between the welding torch and the joint in the welding equipment in real time during the welding process, so as to adjust the direction of the welding torch in real time.

[0163] In this embodiment of the invention, real-time monitoring and adjustment constitute a closed-loop feedback control of the welding process. Specifically, positional deviation refers to the real-time distance between the welding torch centerline and the weld centerline.

[0164] Specifically, an arc sensor or vision sensor is used to collect deviation signals during welding, and the robot's posture is corrected in real time through a PID control algorithm.

[0165] For example, if thermal deformation during welding causes the weld to drift 2mm to the left, the system immediately instructs the welding torch to follow 2mm to the left, ensuring that it always travels along the center of the weld.

[0166] This step enables dynamic correction, effectively preventing defects such as weld misalignment and undercut, and ensuring the internal quality of the weld.

[0167] This embodiment achieves intelligent and adaptive welding processes. It can dynamically match the optimal welding parameters according to the actual condition of the joint, and overcome interference such as thermal deformation in real time during the welding process, ensuring that the weld is aesthetically pleasing and free of internal defects.

[0168] Based on the same technical concept, embodiments of the present invention also provide an electronic device, such as... Figure 4 As shown, it includes a processor 401, a communication interface 402, a memory 403, and a communication bus 404, wherein the processor 401, the communication interface 402, and the memory 403 communicate with each other through the communication bus 404.

[0169] Memory 403 is used to store computer programs; The processor 401 is used to execute the program stored in the memory 403 to implement the steps of the automatic processing method for the subframe.

[0170] The communication bus mentioned in the above electronic devices can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus can be divided into address bus, data bus, control bus, etc. For ease of illustration, only one thick line is used to represent it in the diagram, but this does not indicate that there is only one bus or one type of bus.

[0171] The communication interface is used for communication between the aforementioned electronic devices and other devices.

[0172] The memory may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.

[0173] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0174] The apparatus for the automatic subframe processing method provided in this embodiment of the invention can be specific hardware on the equipment or software or firmware installed on the equipment. The implementation principle and technical effects of the apparatus provided in this embodiment of the invention are the same as those in the foregoing method embodiments. For the sake of brevity, any parts not mentioned in the apparatus embodiments can be referred to the corresponding content in the foregoing method embodiments. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, apparatuses, and units described above can all be referred to the corresponding processes in the above method embodiments, and will not be repeated here.

[0175] In the embodiments provided by this invention, it should be understood that the disclosed apparatus and method can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0176] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0177] In addition, the functional units in the embodiments provided by the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0178] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0179] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. In addition, the terms "first", "second", "third", etc. are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0180] Finally, it should be noted that the above-described embodiments are merely specific implementations of the present invention, used to illustrate the technical solutions of the present invention, and not to limit it. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments within the technical scope disclosed in the present invention, or make equivalent substitutions for some of the technical features; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention. All should be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An automated manufacturing method for subframes, characterized in that, The method is applied to a subframe processing system, which includes a central processing unit, bending equipment, cutting equipment, assembly platform, welding equipment, and visual recognition devices installed at each processing step; the method includes: Obtain the property data of the sheet metal to be processed for manufacturing the subframe; Based on the attribute data, a bending program for the bending equipment is generated; and the bending program is used to drive the bending equipment to perform a bending operation to obtain a plastically deformed sheet metal semi-finished product. In response to the completion of the bending operation, the visual recognition device at the bending process collects the point cloud data of the semi-finished sheet material, and extracts the actual bending angle and surface distortion based on the point cloud data of the semi-finished sheet material to obtain the first processing data package. Based on the first processing data packet, the preset cutting parameters of the cutting device are corrected to generate an adaptive cutting path that adapts to the actual posture of the current semi-finished sheet material; and the cutting device is driven to perform a cutting operation on the semi-finished sheet material using the adaptive cutting path to obtain subframe components; In response to the end of the cutting operation, the actual pose of each of the subframe components is collected by the vision recognition device above the assembly platform; and based on the actual pose and the preset assembly model, the subframe components are assembled to obtain a preliminary subframe assembly. Based on the joint state of the subframe assembly prototype, initial welding parameters are generated, and based on the initial welding parameters, the welding equipment is controlled to weld each joint to complete the final forming of the subframe.

2. The method according to claim 1, characterized in that, The step of generating the bending program for the bending device based on the attribute data includes: The attribute data of the sheet material to be processed and the preset bending parameters are input into the pre-trained springback prediction model, and the predicted springback deformation for the current sheet material to be processed is output; the attribute data includes at least the material and thickness of the sheet material; The rebound compensation angle is determined based on the difference between the predicted rebound deformation and the theoretical rebound deformation. The preset bending parameters are corrected based on the springback compensation angle, and a bending program is generated based on the corrected preset bending parameters.

3. The method according to claim 2, characterized in that, The step of correcting the preset bending parameters based on the springback compensation angle and generating a bending program based on the corrected preset bending parameters includes: The visual recognition device at the feed point collects three-dimensional point cloud data of the board to be processed; and based on the three-dimensional point cloud data of the board to be processed, determines several bending lines and extracts the spatial coordinates of each bending line. Based on the distribution of the spatial coordinates of the bending lines, the execution order of each bending line is determined; For each bending line in the execution sequence, the springback compensation angle is algebraically superimposed with the preset bending angle to obtain the target bending angle; Based on the springback compensation angle and the effective length of the current bending line, the corresponding bending pressure compensation coefficient is matched in the preset process expert database, and the preset bending pressure parameters are corrected using the bending pressure compensation coefficient. According to the execution order, the spatial coordinates, preset bending direction, target bending angle and corrected bending pressure parameters corresponding to each bending line are sequentially arranged and encapsulated to generate an automated bending operation program containing multi-step sequential execution instructions.

4. The method according to claim 1, characterized in that, The extraction of the actual bending angle and surface distortion from the point cloud data of the semi-finished sheet material to obtain the first processing data package includes: The point cloud data of the semi-finished sheet material is spatially registered with the theoretical three-dimensional model of the semi-finished sheet material to eliminate rigid body displacement errors. The normal distance and direction vector from each data point in the registered point cloud to the surface of the theoretical 3D model are calculated to obtain the deviation mapping field characterizing the local deformation distribution of the plate. Based on the aforementioned deviation mapping field, the feature plane normal vectors of the regions on both sides of the bending line are extracted, and the included angle is calculated to obtain the actual bending angle. The fluctuation characteristics of the deviation mapping field along the bending line direction are statistically analyzed, and the fluctuation characteristics are quantified as surface distortion. The actual bending angle and the surface distortion are encapsulated into the first processing data package.

5. The method according to claim 1, characterized in that, The step of correcting the preset cutting parameters of the cutting equipment based on the first processing data packet to generate an adaptive cutting path that matches the actual posture of the current semi-finished board includes: Based on the actual bending angle and surface distortion in the first processing data packet, a nonlinear spatial transformation matrix is ​​constructed from the theoretical coordinate system to the actual deformed coordinate system; Using the nonlinear spatial transformation matrix, the original cutting trajectory coordinates in the preset cutting parameters are mapped and corrected point by point to obtain the corrected spatial trajectory. Based on the corrected spatial trajectory, the feed speed of the cutting head and the laser focus position are replanned to generate an adaptive cutting path.

6. The method according to claim 1, characterized in that, Based on the actual position and the preset assembly model, the subframe components are assembled to obtain a preliminary subframe assembly, including: In the virtual assembly space, with the goal of minimizing the gap between adjacent parts, the optimal pose transformation matrix of each subframe part relative to the preset assembly model is calculated using the iterative nearest point matching algorithm. Based on the optimal pose transformation matrix, the pose of each subframe component on the assembly platform is adjusted to assemble the subframe components and obtain a preliminary subframe assembly.

7. The method according to claim 1, characterized in that, The process of generating initial welding parameters based on the joint state of the subframe assembly prototype, and controlling the welding equipment to weld each joint based on the initial welding parameters, includes: Images of the seams between subframe components are captured using a visual recognition device at the welding equipment. Based on the joint image, initial welding parameters are generated using a large model; the initial welding parameters include at least welding current, welding speed, and welding direction. The welding equipment is controlled to weld along the joint trajectory according to the initial welding parameters; and the positional deviation between the welding torch and the joint in the welding equipment is monitored in real time during the welding process so as to adjust the direction of the welding torch in real time.

8. The method according to claim 1, characterized in that, The method further includes: In response to the end of the cutting operation, the actual contours of each of the subframe components are collected based on the visual recognition device at the cutting equipment. The actual contours of each of the subframe components are compared with the expected contours after the cutting parameters are corrected, and a second processing data package containing information on cutting edge deviation, cut perpendicularity, and local thermal deformation is generated and stored in a pre-built process database.

9. An automated subframe processing system, characterized in that, Includes a central processing unit, bending equipment, cutting equipment, assembly platform, welding equipment, and visual recognition devices installed at each processing step; The central processing unit is communicatively connected to the bending equipment, cutting equipment, hoisting and positioning equipment of the assembly platform, welding equipment, and visual recognition devices installed at each processing step; and executes the method described in any one of claims 1-8.

10. An electronic device, characterized in that, It includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; The memory is used to store computer programs; When the processor executes the program stored in the memory, it implements the method described in any one of claims 1-8.

Images