A method and system for online measurement of springback in automotive door panel stampings.

Abstract

This invention discloses an online measurement method and system for the springback of automotive door panel stampings, relating to the field of automotive stamping parts inspection. Utilizing a compliance coefficient matrix and a first constraint force vector, the full-field elastic deformation field caused by the first constraint force group is calculated. The elastic deformation field is then subtracted from the first hot-state point cloud to obtain an equivalent cold-state free-state point cloud. This equivalent cold-state free-state point cloud is aligned with a pre-stored cold-state theoretical shape point cloud, and the normal deviation of each point is calculated as the springback amount. This invention employs a dual-constraint force group switching detection mode under constant temperature conditions, which can completely eliminate the interference of temperature deformation on springback measurement, ensuring consistent data acquisition conditions for both sets of point clouds and improving the uniformity of the detection benchmark. By online identification of the compliance relationship between force and displacement under hot conditions to construct a compliance coefficient matrix, the full-field elastic deformation caused by tooling constraints is accurately isolated, effectively solving the industry pain point that traditional detection methods cannot distinguish between elastic deformation and actual springback deformation.

Description

Technical Field

[0001] This invention relates to the field of automotive stamping parts testing technology, and in particular to an online method and system for measuring the springback of automotive door panel stamping parts. Background Technology

[0002] Automotive door panels are key stamped body panels, and their forming accuracy directly determines the assembly clearance and appearance quality of the entire vehicle. During the sheet metal stamping process, multiple factors such as material mechanical properties, stamping process parameters, tooling constraints, and ambient temperature can cause the stamped parts to easily spring back after demolding, resulting in deviations between the actual formed shape and the theoretical design shape. This is the core issue restricting the improvement of door panel stamping accuracy.

[0003] Currently, the mainstream springback detection methods in the industry are mostly offline static inspections. These involve cooling the stamped part to room temperature, removing it from the tooling constraint, and then using a 3D scanning device to collect shape data and calculate the springback amount. This method has a significant time lag and cannot meet the online inspection requirements of stamping production lines, making it difficult to provide real-time feedback on springback fluctuations during production. Furthermore, existing detection technologies do not fully consider the coupled effects of the thermal conditions after stamping and the elastic deformation caused by tooling constraints. The residual temperature after stamping can cause minor thermal deformation, and the tooling constraints can produce elastic deformation. Both types of deformation are superimposed on the final detection result, leading to low accuracy in springback measurement.

[0004] Furthermore, traditional testing methods cannot effectively isolate the interference of constraint elastic deformation and temperature deformation on springback measurement, making it difficult to accurately obtain the true free-state springback data of stamped parts. This results in unstable pass rates for door panel stamping products, long production debugging cycles, and difficulty in improving production efficiency. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides an online method and system for measuring the springback of stamped automotive door panels. The technical solution is as follows:

[0006] A method for online measurement of springback in automotive door panel stampings includes the following steps:

[0007] Step 1: Clamp the car door panel stamping part into the constraint simulation device, apply the first constraint force group of the simulated welding fixture constraint, and record the first constraint force vector;

[0008] Step 2: Obtain the first thermal point cloud of the automotive door panel stamping part under the first constraint force group, and simultaneously obtain the first temperature field;

[0009] Step 3: Keeping the clamping state unchanged, quickly switch the clamping force to the second constraint force group through the constraint simulation device, and record the second constraint force vector;

[0010] Step 4: Obtain the second thermal point cloud of the automobile door panel stamping part under the second constraint force group, and simultaneously obtain the second temperature field, and control the difference between the second temperature field and the first temperature field to be within a preset threshold.

[0011] Step 5: Based on multiple sets of thermal point clouds and corresponding constraint force vectors obtained by switching constraint force groups at least once under constant temperature conditions, identify the force and displacement compliance relationship of the automotive door panel stamping part under the current thermal state online and obtain the compliance coefficient matrix.

[0012] Step 6: Using the compliance coefficient matrix and the first constraint force vector, calculate the full-field elastic deformation field caused by the first constraint force group, and subtract the elastic deformation field from the first hot point cloud to obtain the equivalent cold free point cloud;

[0013] Step 7: Align the equivalent cold free state point cloud with the pre-stored cold theoretical shape point cloud, and calculate the normal deviation of each point as the rebound amount.

[0014] Optionally, in step 1, multiple constraint clamping points are preset based on the positioning reference of the welding tooling of the automotive door panel stamping, the clamping point position and the actual production constraint conditions. The constraint clamping points cover the door panel edge, the mounting bracket and the critical stress area of ​​the flange.

[0015] The servo actuators at each clamping point of the control constraint simulation device output preset clamping force and clamping displacement to simulate the combined rigid and elastic constraints of the actual welding fixture.

[0016] Force sensors integrated into each clamping point collect clamping force data in real time, integrate the force values ​​and direction parameters of all clamping points, construct the first constraint force vector with dimension matching, and store the record.

[0017] Optionally, the method for synchronously acquiring the first thermal point cloud and the first temperature field in step 2 is:

[0018] A 3D laser scanning module is used to perform a full-area scan on the automotive door panel stamping part constrained by the first constraint force group, and high-density 3D coordinate point cloud data of the stamping part surface is collected to form a first thermal point cloud. An infrared thermal imaging module is used to simultaneously collect full-area temperature data of the stamping part to construct a first temperature field corresponding to pixels. The acquisition timing of the 3D laser scanning module and the infrared thermal imaging module is bound by a timing synchronization module to ensure that the acquisition time difference between the first thermal point cloud and the first temperature field is less than 10ms.

[0019] Optionally, the control method for rapid switching of constraint force groups in step 3 includes:

[0020] The positioning reference, clamping points, and clamping posture of the automotive door panel stamping part remain completely unchanged. Only the output force values ​​of each servo actuator of the constraint simulation device are adjusted. The force value parameters of each point of the second constraint force group are preset. The constraint stiffness and clamping force of the second constraint force group form a gradient difference with the first constraint force group, and there are no zero-constraint suspended points. The clamping force is dynamically and quickly switched through servo closed-loop control. The time to complete the switching of a single constraint force group is no more than 0.5s. There is no mechanical impact or workpiece displacement during the switching process. After the switching is completed, the data acquisition is performed after stabilizing for 1-2s.

[0021] Optionally, the preset threshold and control method for temperature field difference control in step 4 are as follows:

[0022] The preset thresholds for temperature field differences include a global average temperature difference threshold and a local maximum temperature difference threshold, wherein the global average temperature difference is less than or equal to 0.3℃, and the maximum temperature difference in any local area is less than or equal to 0.5℃; the global temperature distribution data of the second temperature field and the first temperature field are compared in real time. If the temperature difference exceeds the preset threshold, the constant temperature compensation module is activated to perform micro-temperature adjustment on the stamped part; after the temperature field difference falls back to the preset threshold range and the temperature distribution stabilizes, the second temperature field data is solidified and the second thermal point cloud acquisition is completed.

[0023] Optionally, the online identification method for the compliance coefficient matrix in step 5 includes the following sub-steps:

[0024] Step 501: Under constant temperature conditions, repeat the constraint force group switching operation at least twice to obtain multiple sets of one-to-one corresponding constraint force vectors and thermal displacement point cloud data, and construct a force and displacement sample dataset.

[0025] Step 502: Denoise and register preprocessing is performed on multiple sets of thermal point clouds, and displacement variables of each constraint point and global measurement point are extracted. A linear force and displacement balance equation is constructed by combining the corresponding constraint force vector.

[0026] Step 503: The equilibrium equation is solved iteratively using the least squares method. The solution results are combined with the constraint correction results of the mechanical properties of the stamped sheet to identify the compliance coefficient matrix corresponding to the whole domain under the current hot state. The compliance coefficient matrix is ​​a symmetric positive definite matrix that represents the elastic displacement response corresponding to the unit constraint force.

[0027] Optionally, the method for obtaining the equivalent cold free-state point cloud in step 6 includes the following sub-steps:

[0028] Step 601: Substitute the first constraint force vector into the compliance coefficient matrix, and solve for the elastic displacement of each measuring point in the entire domain of the stamping part under the action of the first constraint force through matrix forward operation, and fit to generate a continuous full-field elastic deformation field.

[0029] Step 602: Decompose the coordinates of the first hot point cloud, deduct the elastic deformation displacement components at the corresponding positions point by point, and eliminate the influence of elastic deformation caused by tooling constraints.

[0030] Step 603: Based on the constant characteristics of the two temperature fields, eliminate the temperature deformation error and reconstruct the equivalent cold free state point cloud without constraints and elastic deformation.

[0031] Optionally, the point cloud alignment preprocessing method in step 7 is as follows: the pre-stored cold-state theoretical shape point cloud is the standard point cloud corresponding to the design model of the automotive door panel stamping part, which is the theoretical shape data in the unconstrained free state at room temperature; the ICP iterative nearest point algorithm is used to perform coarse and fine registration of the equivalent cold-state free state point cloud and the cold-state theoretical shape point cloud, and the registration process constrains the translation and rotation degrees of freedom to ensure the consistency of the benchmark; after the registration is completed, abnormal noise points and invalid measurement points at the edge of the point cloud are removed, and the point cloud data of the effective forming area of ​​the stamping part is retained to ensure that the measurement points of the two sets of point clouds correspond one-to-one.

[0032] Optionally, the method for calculating and outputting the rebound amount in step 7 includes the following sub-steps:

[0033] Step 701: For the two sets of registered point clouds, calculate the deviation value of the measuring point along the normal direction of the stamping surface point by point, and correspond the positive and negative deviations of the normal direction to tensile springback and compression springback, respectively.

[0034] Step 702: Calculate the maximum, minimum, and average springback values ​​and root mean square error of the springback at all measurement points in the entire area, and generate a cloud map of the springback distribution of the door panel stamping part in the entire area.

[0035] Step 703: Extract local springback data separately for key forming areas such as door panel flanges, curved surfaces, and installation points, and output the regional springback accuracy evaluation results.

[0036] An online springback measurement system for automotive door panel stampings is provided to realize an online method for measuring the springback of automotive door panel stampings. The system includes a constraint simulation device, a data acquisition device, a temperature control device, and a data processing terminal.

[0037] The constraint simulation device includes a multi-point servo clamping mechanism, a force sensing module, and a drive control module. The multi-point servo clamping mechanism is used to position and clamp automotive door panel stamping parts and output constraint force groups of different gradients. The force sensing module is used to collect force value data at each clamping point in real time and construct constraint force vectors. The drive control module is used to realize the rapid and smooth switching of constraint force groups.

[0038] The data acquisition device includes a three-dimensional laser scanning module and an infrared thermal imaging module. The three-dimensional laser scanning module is used to acquire thermal point cloud data of stamped parts under different constraint states, and the infrared thermal imaging module is used to synchronously acquire temperature field data under corresponding working conditions. The two are matched in time and space through a time synchronization module.

[0039] The temperature control device includes a constant temperature compensation module and a temperature monitoring module. The temperature monitoring module compares the difference between the two temperature fields in real time, while the constant temperature compensation module is used to perform micro-temperature control on the stamped parts to ensure that the difference between the two temperature fields is within a preset threshold.

[0040] The data processing terminal has built-in point cloud preprocessing module, flexibility matrix identification module, elastic deformation solving module, point cloud registration module, and springback calculation module, which are used to complete force and displacement relationship identification, elastic deformation deduction, point cloud alignment, and full-domain springback calculation, so as to realize online accurate measurement of the springback of stamped parts.

[0041] In summary, the present invention has at least one of the following beneficial technical effects:

[0042] This invention provides an online measurement method and system for the springback of stamped automotive door panels. It employs a dual-constraint force switching detection mode under constant temperature conditions, completely eliminating the interference of temperature deformation on springback measurement and ensuring consistent point cloud data acquisition conditions, thus improving the uniformity of the detection benchmark. By constructing a compliance coefficient matrix through online identification of the compliance relationship between force and displacement under hot conditions, it can accurately isolate the full-field elastic deformation caused by tooling constraints, effectively solving the industry pain point that traditional detection methods cannot distinguish between elastic deformation and actual springback deformation, and obtaining high-precision equivalent cold-state free-state point cloud data.

[0043] It enables real-time online measurement of springback in stamped parts, eliminating the need for offline cooling and resting of the workpiece. This aligns with the continuous production and inspection requirements of automated stamping lines, significantly shortening the inspection cycle and improving production and inspection efficiency. Furthermore, through full-domain point cloud registration and normal deviation calculation, it achieves precise quantification of springback across the entire stamped part and in key areas, providing intuitive feedback on springback distribution. This offers accurate and reliable data support for optimizing stamping process parameters, adjusting tooling structures, and designing springback compensation, effectively improving the forming accuracy and product consistency of automotive door panel stamped parts and reducing the defect rate. Attached Figure Description

[0044] Figure 1 This is a flowchart illustrating an online method for measuring the springback of stamped automotive door panels according to the present invention.

[0045] Figure 2 This is a schematic diagram of a regional rebound accuracy evaluation report according to a specific embodiment of the present invention;

[0046] Figure 3This is a global rebound distribution cloud map of a specific embodiment of the present invention;

[0047] Figure 4 This is a schematic diagram of the component connection principle of an online measurement system for the springback amount of automotive door panel stamping parts according to the present invention.

[0048] Figure reference numerals: 11. Multi-point servo clamping mechanism; 12. Force sensing module; 13. Drive control module; 2. Data acquisition device; 21. Three-dimensional laser scanning module; 22. Infrared thermal imaging module; 31. Constant temperature compensation module; 32. Temperature monitoring module; 4. Data processing terminal; 41. Point cloud preprocessing module; 42. Flexibility matrix identification module; 43. Elastic deformation solving module; 44. Point cloud registration module; 45. Springback calculation module. Detailed Implementation

[0049] The present invention will be further described in detail below with reference to the accompanying drawings.

[0050] This invention discloses an online method and system for measuring the springback of stamped automotive door panels.

[0051] Reference Figures 1-4 Example 1: An online method for measuring the springback of stamped automotive door panels, comprising the following steps:

[0052] Step 1: Clamp the car door panel stamping part into the constraint simulation device, apply the first constraint force group of the simulated welding fixture constraint, and record the first constraint force vector;

[0053] Step 2: Obtain the first thermal point cloud of the automotive door panel stamping part under the first constraint force group, and simultaneously obtain the first temperature field;

[0054] Step 3: Keeping the clamping state unchanged, quickly switch the clamping force to the second constraint force group through the constraint simulation device, and record the second constraint force vector;

[0055] Step 4: Obtain the second thermal point cloud of the automobile door panel stamping part under the second constraint force group, and simultaneously obtain the second temperature field, and control the difference between the second temperature field and the first temperature field to be within a preset threshold.

[0056] Step 5: Based on multiple sets of thermal point clouds and corresponding constraint force vectors obtained by switching constraint force groups at least once under constant temperature conditions, identify the force and displacement compliance relationship of the automotive door panel stamping part under the current thermal state online and obtain the compliance coefficient matrix.

[0057] Step 6: Using the compliance coefficient matrix and the first constraint force vector, calculate the full-field elastic deformation field caused by the first constraint force group, and subtract the elastic deformation field from the first hot point cloud to obtain the equivalent cold free point cloud;

[0058] Step 7: Align the equivalent cold free state point cloud with the pre-stored cold theoretical shape point cloud, and calculate the normal deviation of each point as the rebound amount.

[0059] By adopting the above technical solution, the stamped part of the automobile door panel is placed on the worktable of the constraint simulation device. Using the positioning holes of the welding fixture as a reference, the part is clamped by the positioning pins and support blocks of the multi-point servo clamping mechanism 11, ensuring that the clamping state is consistent with the subsequent welding fixture. The constraint simulation device applies the first constraint force group according to a preset program. Each clamping point synchronously outputs the set clamping force. The force sensing module 12 collects the force value and direction cosine of each point in real time, summarizing them into a first constraint force vector. The vector dimension is equal to the total number of clamping points multiplied by the number of force components at a single point. The data is stored in the real-time database of the data processing terminal 4 in the form of a double-precision floating-point array. The data acquisition device 2 is activated, and the 3D laser scanning module 21 performs a full-area scan of the upper surface of the automotive door panel stamping part in line laser scanning mode to acquire a high-density point cloud with a point spacing of no more than 0.2 mm, forming the first thermal point cloud. At the same time, the infrared thermal imaging module 22 acquires the surface temperature distribution of the stamping part at a resolution of 640×512 pixels to form the first temperature field. The timing synchronization module ensures that the deviation between the start time of acquisition of the two is less than 10 ms through hardware triggering, so as to achieve data spatiotemporal matching. Keeping the clamping posture unchanged, the drive control module 13 controls the output force of each clamping point to switch instantly from the first constraint force group to the second constraint force group through the torque loop control of the servo motor. The switching process adopts an S-shaped acceleration and deceleration curve to avoid force impact. The switching completion time does not exceed 0.5 s. After the switching, it stabilizes for 2 s to allow the structural response to decay, and the second constraint force vector at this time is recorded. Data acquisition device 2 scans again in the same manner to acquire the second thermal point cloud and the second temperature field. Temperature monitoring module 32 calculates the global average temperature difference and the local maximum temperature difference between the two temperature fields in real time. If the temperature difference exceeds the preset threshold of 0.3℃ for the global average temperature difference or 0.5℃ for the local maximum temperature difference, the constant temperature compensation module 31 is activated to compensate for heating or cooling of the stamped part through the array of semiconductor temperature control units until the temperature difference falls back to within the threshold and stabilizes, and then the acquisition is completed. Based on the obtained at least two sets of constraint force vectors and the corresponding thermal point cloud, under the premise of constant temperature, compliance matrix identification module 42 performs noise reduction and registration based on the reference hole on the point cloud, extracts the displacement increment of each measuring point, constructs a linear force and displacement balance equation, iteratively solves it using the weighted least squares method, and introduces the thin plate bending strain energy constraint term to correct the solution, thereby obtaining a symmetric positive definite compliance coefficient matrix. Elastic deformation solution module 43 reads the compliance coefficient matrix and the first constraint force vector, calculates the elastic displacement vector of all measuring points under the action of the first constraint force through matrix multiplication, and generates a global elastic deformation field. The point cloud preprocessing module 41 subtracts the elastic displacement component of the corresponding position from the coordinates of each measuring point in the first hot point cloud. Since the first temperature field is consistent with the second temperature field, it can be directly assumed that the thermal expansion displacement has been canceled in the differential calculation of the two scans, thus obtaining the equivalent cold free point cloud.The point cloud registration module 44 aligns the equivalent cold-state free-state point cloud with the cold-state theoretical shape point cloud derived from the design model. Using the ICP iterative nearest point algorithm, with the stamping part's reference surface as the initial alignment reference, after coarse and fine registration, the two sets of point clouds achieve point correspondence in a unified coordinate system. The springback calculation module 45 calculates the deviation along the normal to the theoretical surface for each corresponding point. This deviation value is defined as the springback amount at that point; a positive deviation indicates tensile springback, and a negative deviation indicates compressive springback. The results are output in attributed point cloud format.

[0060] In Example 2, in step 1, multiple constraint clamping points are preset based on the positioning reference of the welding tooling of the automotive door panel stamping part, the clamping point position and the actual production constraint conditions. The constraint clamping points cover the edge of the door panel, the mounting bracket and the key stress area of ​​the flange.

[0061] The servo actuators at each clamping point of the control constraint simulation device output preset clamping force and clamping displacement to simulate the combined rigid and elastic constraints of the actual welding fixture.

[0062] Force sensors integrated into each clamping point collect clamping force data in real time, integrate the force values ​​and direction parameters of all clamping points, construct the first constraint force vector with dimension matching, and store the record.

[0063] By adopting the above technical solution, the preset constraint clamping points in step 1 are based on the welding fixture drawings of the stamping parts. The position coordinates of all positioning pins, support surfaces, and clamps are extracted, and key forming areas such as the edge of the door panel and window frame, hinge mounting bracket, door lock mounting hole, and surrounding flange line are added as supplementary constraint points. The total number of constraint clamping points is between 20 and 35. Each point of the multi-point servo clamping mechanism 11 of the constraint simulation device includes a servo electric cylinder and a ball head clamping unit. The stroke of the servo electric cylinder is 50mm, and the repeatability is ±5μm. It can realize position control and force control simultaneously. According to the actual working conditions of the welding fixture, some points are set to rigid constraint mode, that is, the position is maintained in a closed loop after the clamping displacement is controlled to the contact surface; some points are set to elastic constraint mode, that is, a constant clamping force is output by force closed-loop control, with a force value range of 50N to 800N. The force sensing module 12 uses strain gauge force sensors. Each sensor outputs three orthogonal force components, which are aggregated to the drive control module 13 via the CAN bus. The first constraint force vector is constructed as follows: the measured force components of all points are arranged into a column vector according to the point number. The vector elements are in the order of X, Y, Z force of point 1, X, Y, Z force of point 2, and so on. The total length of the vector is equal to 3 times the number of clamping points. When storing, a timestamp and point status information are attached.

[0064] Example 3, the method for synchronously acquiring the first thermal point cloud and the first temperature field in step 2 is as follows:

[0065] A three-dimensional laser scanning module 21 is used to perform a full-area scan on the automotive door panel stamping part constrained by the first constraint force group, and high-density three-dimensional coordinate point cloud data of the stamping part surface is collected to form a first thermal point cloud; an infrared thermal imaging module 22 is used to synchronously collect the full-area temperature data of the stamping part to construct a first temperature field corresponding to the pixel level; the acquisition timing of the three-dimensional laser scanning module 21 and the infrared thermal imaging module 22 is bound by a timing synchronization module to ensure that the acquisition time difference between the first thermal point cloud and the first temperature field is less than 10ms.

[0066] By adopting the above technical solution, in step 2, the 3D laser scanning module 21 uses a blue laser scanner with a laser line width of 0.1mm, a scanning rate of 300 contours per second, a single scanning range of 400mm×300mm, and a depth resolution of 0.01mm. During scanning, the scanning head is held by a robot and moves along the surface of the stamped part in a serpentine path at a speed of 100mm / s, covering the entire formed area, acquiring approximately 2 million to 5 million points in a single scan. The infrared thermal imaging module 22 uses an uncooled focal plane detector. The timing synchronization module adopts a combination of the IEEE 1588 precision time protocol and external hardware triggering: the 3D laser scanning module 21 outputs a pulse signal for each line of contour scanned, and each frame exposure of the infrared thermal imaging module 22 is triggered by this pulse signal, while recording a precise timestamp to ensure the correspondence between each point cloud data and temperature image pixels, with the time difference controlled within 10ms. The first temperature field data is a temperature mapping within the same spatial range as the point cloud. Using the point cloud coordinates as a reference, the temperature values ​​are mapped onto the point cloud surface through the internal and external parameter calibration and coordinate transformation of the infrared thermal imaging module 22, thus constructing a pixel-level first temperature field matrix.

[0067] Example 4, the control method for rapid switching of constraint force groups in step 3 includes:

[0068] The positioning reference, clamping points, and clamping posture of the automotive door panel stamping part remain completely unchanged. Only the output force values ​​of each servo actuator of the constraint simulation device are adjusted. The force value parameters of each point of the second constraint force group are preset. The constraint stiffness and clamping force of the second constraint force group form a gradient difference with the first constraint force group, and there are no zero-constraint suspended points. The clamping force is dynamically and quickly switched through servo closed-loop control. The time to complete the switching of a single constraint force group is no more than 0.5s. There is no mechanical impact or workpiece displacement during the switching process. After the switching is completed, the data acquisition is performed after stabilizing for 1-2s.

[0069] By adopting the above technical solution, the rapid switching of constraint force groups in step 3 is implemented using the following method: The drive control module 13 internally presets two sets of force parameter tables for each servo clamping point, corresponding to the first and second constraint force groups respectively. After the switching command is triggered, the servo driver directly reads the target force value of the second constraint force group from the torque loop control mode. Through a feedforward + feedback composite control algorithm, the current loop output is adjusted to the corresponding value within 10ms, and the force rise time does not exceed 0.3s. To prevent impact, the force change slope is limited to within 2000N / s, and a notch filter is set to suppress mechanical resonance. During the switching process, the position loop maintains the position command of the previous state unchanged; only the force control command is switched, therefore the spatial pose of the stamped part does not change. The force value differences at each point are designed as follows: the first constraint force group simulates the actual clamping force distribution during welding, with a maximum force value of approximately 600N; the second constraint force group increases the force value at some key points by 20% to 30% and decreases it at others by 10% to 15%, but never reaches a zero-force state, with the minimum force value maintained above 80N, forming a clear stiffness gradient. After the switch is completed, the force sensing module 12 monitors the stability of the force value. If the fluctuation amplitude is less than ±1N and lasts for 1.5s, it is determined to be in a stable state, and data acquisition is then triggered.

[0070] Example 5, the preset threshold and control method for temperature field difference control in step 4 are as follows:

[0071] The preset threshold for temperature field difference includes a global average temperature difference threshold and a local maximum temperature difference threshold, wherein the global average temperature difference is less than or equal to 0.3℃, and the maximum temperature difference in any local area is less than or equal to 0.5℃; the global temperature distribution data of the second temperature field and the first temperature field are compared in real time. If the temperature difference exceeds the preset threshold, the constant temperature compensation module 31 is activated to perform micro-temperature adjustment on the stamping part; after the temperature field difference falls back to the preset threshold range and the temperature distribution stabilizes, the second temperature field data is solidified and the second thermal point cloud acquisition is completed.

[0072] By adopting the above technical solution, the preset threshold for temperature field difference control in step 4 is specifically set as: the average temperature difference across the entire domain. The maximum temperature difference in a local area at any 3×3 measuring point The temperature monitoring module 32 acquires real-time temperature field data from the infrared thermal imaging module 22 and performs pixel-by-pixel difference calculations with the stored first temperature field. If the difference exceeds the limit, the constant temperature compensation module 31 is activated. The constant temperature compensation module 31 consists of 16 independently controlled far-infrared radiation heating plates and 4 eddy current cooling pipes, arranged approximately 300mm above and below the stamped part. The heating plate power is adjustable from 0-200W, and the cooling pipes are adjusted by compressed air flow to achieve non-contact temperature control. The control algorithm adopts fuzzy PID, with temperature difference value and temperature difference change rate as input, an adjustment cycle of 0.5s, and a temperature adjustment resolution of 0.05℃. After the global average temperature difference between the second temperature field and the first temperature field drops below 0.2℃ and the local maximum temperature difference drops below 0.4℃ and remains stable for 3 sampling cycles, the second temperature field data is solidified, and then the three-dimensional laser scanning module 21 is executed to scan and obtain the second thermal point cloud.

[0073] Example 6, the online identification method for the compliance coefficient matrix in step 5 includes the following sub-steps:

[0074] Step 501: Under constant temperature conditions, repeat the constraint force group switching operation at least twice to obtain multiple sets of one-to-one corresponding constraint force vectors and thermal displacement point cloud data, and construct a force and displacement sample dataset.

[0075] Step 502: Denoise and register preprocessing is performed on multiple sets of thermal point clouds, and displacement variables of each constraint point and global measurement point are extracted. A linear force and displacement balance equation is constructed by combining the corresponding constraint force vector.

[0076] Step 503: The equilibrium equation is solved iteratively using the least squares method. The solution results are combined with the constraint correction results of the mechanical properties of the stamped sheet to identify the compliance coefficient matrix corresponding to the whole domain under the current hot state. The compliance coefficient matrix is ​​a symmetric positive definite matrix that represents the elastic displacement response corresponding to the unit constraint force.

[0077] By adopting the above technical solution, in step 501, under the isothermal condition that confirms the difference between the first temperature field and the second temperature field does not exceed the limit, at least three constraint force group switching operations are performed. Each switching generates a pair of constraint force vectors and a corresponding thermal point cloud. Combined with the initial first set of data, at least four sets of paired data are obtained, forming a force and displacement sample dataset. Each set of data contains one constraint force vector. and the corresponding point cloud .

[0078] In step 502, the first group of point clouds is used. Based on this, the remaining point clouds were analyzed. Rigid body registration based on the reference hole is performed to eliminate overall displacement, and then the displacement difference between each measuring point and the reference point cloud is calculated. , forming a displacement vector The displacement at the constraint point is selected as the driving displacement, combined with the force vector. and The difference Establish a system of linear equations: ;in The stiffness matrix is ​​used, but it is more convenient to directly identify the flexibility matrix. Therefore, the form is adopted. Combine all sets of data to form an overdetermined system of equations, in matrix form. .

[0079] In step 503, the compliance coefficient matrix C is solved using the least squares method with Tikhonov regularization. The regularization parameter is selected using the L-curve method to balance the smoothness of the fitting residuals and the solution. Simultaneously, considering the mechanical properties of the thin sheet metal of the stamped part, a constraint is introduced: the compliance matrix C must satisfy symmetry. Furthermore, based on the moment-curvature relationship derived from the thin plate bending theory, the consistency constraint of the magnitude of the compliance term is used to project and correct the solution results, ultimately obtaining a symmetric positive definite compliance coefficient matrix, whose physical meaning is the elastic displacement generated in the direction of point i when a unit force is applied in the direction of point j.

[0080] Example 7, the method for obtaining the equivalent cold free state point cloud in step 6 includes the following sub-steps:

[0081] Step 601: Substitute the first constraint force vector into the compliance coefficient matrix, and solve for the elastic displacement of each measuring point in the entire domain of the stamping part under the action of the first constraint force through matrix forward operation, and fit to generate a continuous full-field elastic deformation field.

[0082] Step 602: Decompose the coordinates of the first hot point cloud, deduct the elastic deformation displacement components at the corresponding positions point by point, and eliminate the influence of elastic deformation caused by tooling constraints.

[0083] Step 603: Based on the constant characteristics of the two temperature fields, eliminate the temperature deformation error and reconstruct the equivalent cold free state point cloud without constraints and elastic deformation.

[0084] By adopting the above technical solution, the acquisition of the equivalent cold free state point cloud in step 6 is implemented through the following process: the elastic deformation solution module 43 reads in the stored first constraint force vector F1, multiplies it with the identified compliance coefficient matrix C, and obtains the elastic displacement field of all measuring points on the surface of the stamped part under the action of the first constraint force. The displacement field is a three-dimensional vector field, with displacement components in the x, y, and z directions at each measuring point. To obtain a continuous full-field elastic deformation field, bicubic spline interpolation is used to fit the elastic displacement of discrete measuring points, generating a deformation field mesh that matches the density of the first hot-state point cloud. Subsequently, the point cloud preprocessing module 41 traverses each point of the first hot-state point cloud, subtracting the corresponding elastic deformation displacement component from the three-dimensional coordinates of that point. This operation is directly performed using point coordinate vector subtraction. Since the first and second temperature fields are obtained under the same isothermal conditions and the temperature fields are consistent during the two scans, the thermal expansion and contraction of the stamped part are approximately equal in the two point clouds, naturally canceling each other out during the subtraction process. Therefore, there is no need to calculate the thermal deformation separately. The final output is an equivalent cold-state free-state point cloud without clamping constraints, elastic deformation, and with the influence of thermal deformation eliminated. This point cloud represents the geometric state of the stamped part, which only includes the shape deviation of plastic forming.

[0085] In Example 8, the point cloud alignment preprocessing method in step 7 is as follows: the pre-stored cold-state theoretical shape point cloud is the standard point cloud corresponding to the design model of the automotive door panel stamping part, which is the theoretical shape data in the unconstrained free state at room temperature; the ICP iterative nearest point algorithm is used to perform coarse and fine registration of the equivalent cold-state free state point cloud and the cold-state theoretical shape point cloud, and the registration process constrains the translation and rotation degrees of freedom to ensure the consistency of the benchmark; after the registration is completed, abnormal noise points and invalid measurement points at the edge of the point cloud are removed, and the point cloud data of the effective forming area of ​​the stamping part is retained to ensure that the measurement points of the two sets of point clouds correspond one-to-one.

[0086] By adopting the above technical solution, the cold-state theoretical shape point cloud is generated from the CAD digital model through STL surface discretization, with a discretization density set to 4 points per square millimeter and the direction being the outward normal of the shape surface. The equivalent cold-state free-state point cloud and the cold-state theoretical shape point cloud are first coarsely aligned using three reference points: the centers or edge corners of three non-collinear reference holes on the stamped part are extracted from the point cloud, and the rigid body transformation matrix is ​​calculated to achieve initial posture uniformity. Then, the ICP iterative nearest-point algorithm is used for fine registration, with KD-tree acceleration used for the search strategy. The distance threshold for matching point pairs is set to 0.5 mm, and point pairs exceeding the threshold are removed. The iteration termination condition is that the root mean square error change between two consecutive iterations is less than 10^{-5} mm or the maximum number of iterations (50) is reached. During the registration process, the constrained model can only translate along the X, Y, and Z axes and rotate around them, without applying scaling. After registration, the two sets of point clouds are cropped at the edges. The boundary noise and invalid measurement points caused by scanning occlusion are identified and removed using the normal vector gradient mutation feature. Only the data of the effective forming area of ​​the stamping part are retained to ensure that the number and position of measurement points of the two point clouds are completely corresponding, providing conditions for point-by-point deviation calculation.

[0087] Example 9, the method for calculating and outputting the rebound amount in step 7 includes the following sub-steps:

[0088] Step 701: For the two sets of registered point clouds, calculate the deviation value of the measuring point along the normal direction of the stamping surface point by point, and correspond the positive and negative deviations of the normal direction to tensile springback and compression springback, respectively.

[0089] Step 702: Calculate the maximum, minimum, and average springback values ​​and root mean square error of the springback at all measurement points in the entire area, and generate a cloud map of the springback distribution of the door panel stamping part in the entire area.

[0090] Step 703: Extract local springback data separately for key forming areas such as door panel flanges, curved surfaces, and installation points, and output the regional springback accuracy evaluation results.

[0091] By adopting the above technical solution, for the registered point cloud, each measurement point i finds the three nearest points in the theoretical shape point cloud to form a local plane, and calculates the unit normal vector of this plane. The direction points outward from the surface. Calculate the corresponding point in the equivalent cold free-state point cloud. To the theoretical point Directed distance This is the normal springback at that point; a positive value indicates springback under convex tension, and a negative value indicates springback under concave compression. Statistically analyze the springback at all valid measuring points and find the maximum value. Minimum value Calculate the average rebound amount And the root mean square error (RMSE). When generating the global rebound distribution cloud map, the cold-state theoretical shape point cloud is used as the carrier. Values ​​are mapped to colors, with the color spectrum ranging from blue (corresponding to maximum negative springback) to red (corresponding to maximum positive springback). Areas outside the tolerance range are highlighted. For critical forming areas such as door panel flanges, curved surface transition zones, and mounting supports, a predefined boundary selection range is used. Points within each area are extracted for the aforementioned statistical analysis. The output includes the area's maximum springback, springback pass rate, and CPK value, providing an evaluation result of the area's springback accuracy. The results are output in PDF and CSV data file formats.

[0092] Example 10: An online springback measurement system for automotive door panel stampings, used to realize an online springback measurement method for automotive door panel stampings. The system includes a constraint simulation device, a data acquisition device 2, a temperature control device, and a data processing terminal 4.

[0093] The constraint simulation device includes a multi-point servo clamping mechanism 11, a force sensing module 12, and a drive control module 13. The multi-point servo clamping mechanism 11 is used to position and clamp the automotive door panel stamping and output constraint force groups of different gradients. The force sensing module 12 is used to collect force value data of each clamping point in real time and construct constraint force vectors. The drive control module 13 is used to realize the rapid and smooth switching of constraint force groups.

[0094] The data acquisition device 2 includes a three-dimensional laser scanning module 21 and an infrared thermal imaging module 22. The three-dimensional laser scanning module 21 is used to acquire thermal point cloud data of stamping parts under different constraint states, and the infrared thermal imaging module 22 is used to synchronously acquire temperature field data under corresponding working conditions. The two are matched in time and space through a time synchronization module.

[0095] The temperature control device includes a constant temperature compensation module 31 and a temperature monitoring module 32. The temperature monitoring module 32 compares the difference between the two temperature fields in real time, and the constant temperature compensation module 31 is used to perform micro-temperature control on the stamped parts to ensure that the difference between the two temperature fields is within a preset threshold.

[0096] The data processing terminal 4 has a built-in point cloud preprocessing module 41, a flexibility matrix identification module 42, an elastic deformation solving module 43, a point cloud registration module 44, and a springback calculation module 45, which are used to complete the identification of force and displacement relationship, elastic deformation deduction, point cloud alignment, and full-domain springback calculation, so as to realize the online accurate measurement of the springback of stamped parts.

[0097] By adopting the above technical solution, the core of the constraint simulation device is a multi-point servo clamping mechanism 11, whose base is made of a cast iron welded platform, on which 30 servo electric cylinders are arranged in an array. The strain-type three-dimensional force sensor of the force sensing module 12 is integrated at the front end of the cylinder body. The drive control module 13 is a 19-inch rack-mounted servo drive system, which uses EtherCAT industrial Ethernet bus to realize synchronous control of each axis, with a communication cycle of 250μs, and is used for multi-axis collaborative force / position hybrid control switching. The three-dimensional laser scanning module 21 of the data acquisition device 2 is installed at the end of a 6-axis industrial robot, and the robot drives the scanning head to move along the planned path; the infrared thermal imaging module 22 is fixed on the top of the gantry, and the field of view covers the entire stamped part. The time synchronization module distributes the horizontal synchronization signal of the scanning head to the external trigger input port of the infrared thermal imager through hard wiring. The temperature monitoring module 32 of the temperature control device is an industrial computer that runs a real-time temperature measurement and comparison program based on LabVIEW. The actuator of the constant temperature compensation module 31 is driven by a relay matrix controlled by PLC, which can independently adjust each heating and cooling unit. Data processing terminal 4 is a high-performance workstation equipped with a multi-core CPU and a professional graphics processing unit. Point cloud preprocessing module 41 performs denoising, sampling, and normal vector estimation; flexibility matrix identification module 42 encapsulates a least-squares solver; elastic deformation solution module 43 uses a sparse matrix operation library; point cloud registration module 44 is based on the PCL point cloud library; and springback calculation module 45 integrates a visualization rendering engine. All modules communicate via gigabit Ethernet and TCP / IP protocols. Data processing terminal 4 deploys a web-based human-machine interface that displays the measurement progress and springback results in real time, while historical data is stored in an SQL database.

[0098] The following specific embodiments illustrate the implementation principle of the present invention:

[0099] A vehicle manufacturing company's stamping workshop produces the outer left front door panel of a certain model. The material is DC06 cold-rolled steel sheet, 0.7mm thick, and the stamping process is drawing, trimming, and flanging. During the mold debugging phase, it was found that the springback of the door panel window frame area and the flanging line area exceeded the tolerance. The maximum springback at the window frame corner reached 1.8mm, exceeding the ±0.5mm tolerance range. It is necessary to perform online precise measurement of the springback of the stamped parts to guide mold correction.

[0100] During implementation, the stamped left front door panel after production is placed on the worktable of the constraint simulation device. Using the three welding fixture positioning holes A1, A2 and B reference holes on the door panel as the clamping reference, the multi-point servo clamping mechanism 11 is configured with a total of 28 clamping points, including 14 points evenly distributed around the perimeter of the door panel, 6 points on the inner edge of the window frame, 4 points in the hinge mounting support area, 2 points in the door lock mounting hole area, and 2 points in the lower edge area. Each point is clamped by positioning pins and ball head clamping units.

[0101] The drive control module 13 outputs the first constraint force group according to preset parameters to simulate the constraint state of the actual welding fixture. The peripheral points use force closed-loop control, with force values ​​set from 80N to 450N; the four points of the hinge mounting support use position closed-loop control, holding the force after it has moved to the contact surface; the six points of the inner edge of the window frame use force closed-loop control, with force values ​​set from 120N to 350N. The force sensing module 12 collects 84 force components from 28 points at a sampling frequency of 1kHz, summarizes them to construct the first constraint force vector F1 with a vector dimension of 84, and stores it in the data processing terminal 4.

[0102] Subsequently, data acquisition device 2 is activated. The blue laser scanner of the 3D laser scanning module 21, held by a 6-axis industrial robot, scans along the door panel surface in a serpentine path at a speed of 100 mm / s, acquiring approximately 3.8 million points per scan with a point spacing of 0.15 mm, forming the first thermal point cloud. The infrared thermal imaging module 22 simultaneously acquires temperature field data. The detector resolution is 640×512, the frame rate is 30Hz, and the timing synchronization module ensures an 8ms acquisition time difference through hardware triggering of the line scanning pulse. The first temperature field displays an average temperature of 28.5℃ across the entire door panel, completing the pixel-level temperature mapping of the first temperature field matrix.

[0103] After the initial data acquisition, the drive control module 13 performs a constraint force group switch. The second constraint force group parameters are preset for each servo clamping point: the force values ​​at the four key points on the inner edge of the window frame are increased by 25%, from 350N to 437N; the hinge mounting support area is switched from position control to force control, with a force value of 300N; the force values ​​at the six peripheral points are reduced by 12%, from 200N to 176N; the force values ​​at the remaining points are maintained or slightly adjusted, with a minimum force value of 100N. The switch employs a feedforward and feedback composite control, limiting the force change slope to within 1800N / s. The 25 points complete the switch within 0.4s, maintaining the position command unchanged during the switch, and ensuring no spatial orientation shift in the stamped part. After stabilization for 1.8s, the force sensing module 12 confirms that the force value fluctuation at each point is less than ±0.8N and records the second constraint force vector F2.

[0104] The second scan was then prepared, with temperature monitoring module 32 comparing the second temperature field with the first temperature field in real time. The comparison results showed that the maximum local temperature difference in the window frame area reached 0.7℃, exceeding the 0.5℃ threshold. The constant temperature compensation module 31 was activated, with four far-infrared radiation heating plates in the corresponding area heating at 120W power and two eddy current cooling pipes cooling at 30% flow rate. The fuzzy PID adjustment cycle was 0.5s. After five cycles, the maximum local temperature difference in the window frame area decreased to 0.3℃, and the average temperature difference across the entire area was 0.15℃. After stabilizing for three sampling cycles, the second temperature field data was solidified, with an average temperature of 28.6℃ across the entire area. The three-dimensional laser scanning module 21 completed the second scan, acquiring the second thermal point cloud with 3.79 million points.

[0105] To construct the force and displacement sample dataset, the constraint force group switching operation is repeated twice, generating the third constraint force vector F3 and its corresponding third thermal point cloud, and the fourth constraint force vector F4 and its corresponding fourth thermal point cloud, respectively. The four switching operations form four sets of force vector and thermal point cloud pairing data. The temperature monitoring module 32 confirms that the temperature field meets the preset threshold in all four acquisitions, thus constituting the force and displacement sample dataset under isothermal conditions.

[0106] The compliance matrix identification module 42 uses the first set of point clouds as a reference and performs rigid body registration on the remaining three sets of point clouds based on positioning holes A1, A2, and B. After eliminating the overall rigid body displacement, it calculates the displacement increment of each measuring point. A linear force-displacement equilibrium equation is constructed using the changes in 84 force components and corresponding displacements at 28 clamping points. Iterative solutions are obtained using a weighted least squares method with Tikhonov regularization. The L-curve determines the regularization parameter to be 0.15, and compliance matrix symmetry constraints are applied. equal By constraining the magnitude of the bending compliance term of the thin plate, an 84th-order symmetric positive definite compliance coefficient matrix C is identified, with a matrix condition number of 36.2, which satisfies the stability requirements of the solution.

[0107] The elastic deformation solution module 43 will calculate the first constraint force vector. Substituting the flexibility coefficient matrix C, we can use matrix multiplication to find that Delastic equals C multiplied by C. The full-field elastic deformation field under the action of the first constraint force group was calculated, with displacement components including the x, y, and z directions. The maximum elastic displacement at the window frame corner was calculated to be 0.52 mm. Bicubic spline interpolation was performed on the discrete elastic displacement to generate a deformation field mesh with the same density as the first thermal point cloud.

[0108] The point cloud preprocessing module 41 traverses each measurement point of the first hot-state point cloud, directly subtracting the elastic deformation displacement component at the corresponding position from the point coordinate vector, thus deducting the influence of elastic deformation point by point. Since the first temperature field of 28.5℃ and the second temperature field of 28.6℃, as well as the subsequent two temperature fields, are all between 28.5℃ and 28.7℃, the thermal expansion and contraction are approximately equal and naturally cancel each other out in multiple scans, resulting in an equivalent cold-state free-state point cloud with approximately 3.78 million points, representing the shape deviation of the stamped part that only includes plastic springback.

[0109] The point cloud registration module 44 derives the cold-state theoretical shape point cloud from the design digital model, with a discrete density of 4 points per square millimeter, totaling approximately 3.92 million points. Coarse registration is performed using the coordinates of the center points of positioning holes A1, A2, and B, and the rigid body transformation matrix is ​​calculated to achieve initial alignment. Subsequently, fine registration is performed using the ICP iterative nearest-point algorithm with a KD-tree search strategy and a matching distance threshold of 0.5 mm. After the 23rd iteration, the root mean square error decreases to 8.6 x 10⁻⁶ mm, satisfying the termination condition. The registration process constrains three-axis translation and rotation, without applying scaling. After registration, edge noise and scanned occlusion areas are trimmed using the normal vector gradient abrupt change feature, retaining approximately 3.65 million corresponding measurement points within the effective forming area of ​​the stamped part.

[0110] The rebound calculation module 45 calculates the normal rebound amount point-by-point. For each measurement point, the system searches for its three nearest neighbors in the theoretical point cloud to construct a local plane and determine the direction of the unit normal vector. It then calculates the directed distance from the corresponding point in the equivalent cold-state free-state point cloud to the theoretical point. Statistical results show that the maximum positive rebound amount across the entire area is 1.52 mm, located at the upper left corner of the window frame; the maximum negative rebound amount is -0.38 mm, located in the middle of the downturned edge; the average rebound amount is 0.21 mm; and the root mean square error (RMSE) is 0.34 mm. A global rebound distribution cloud map is generated, as shown below. Figure 3 As shown, using the point cloud of the cold-state theoretical shape as the carrier, the positive rebound area is shown in red, the negative rebound area is shown in blue, and the out-of-tolerance area is highlighted.

[0111] Local springback data was extracted separately for three key forming areas: the window frame area, the flange line area, and the hinge mounting bracket area. The maximum springback in the window frame area was 1.52mm, with a springback pass rate of 62.3%; the maximum springback in the flange line area was 0.41mm, with a springback pass rate of 94.8%; and the maximum springback in the hinge mounting bracket area was 0.28mm, with a springback pass rate of 100%. The output includes a regional springback accuracy evaluation report containing regional springback statistics, CPK values, and the distribution of out-of-tolerance points. Figure 2 As shown, submit mold correction references in PDF and CSV formats.

[0112] The entire measurement process, from clamping to outputting the springback report, takes approximately 11 minutes, matching the sampling inspection cycle of stamped parts, enabling online accurate measurement of the springback amount of door panel stamped parts and evaluation of the accuracy of key areas.

[0113] The above are all preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape and principle of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A method for online measurement of springback in automotive door panel stampings, characterized in that, Includes the following steps: Step 1: Clamp the car door panel stamping part into the constraint simulation device, apply the first constraint force group of the simulated welding fixture constraint, and record the first constraint force vector; Step 2: Obtain the first thermal point cloud of the automotive door panel stamping part under the first constraint force group, and simultaneously obtain the first temperature field; Step 3: Keeping the clamping state unchanged, quickly switch the clamping force to the second constraint force group through the constraint simulation device, and record the second constraint force vector; Step 4: Obtain the second thermal point cloud of the automobile door panel stamping part under the second constraint force group, and simultaneously obtain the second temperature field, and control the difference between the second temperature field and the first temperature field to be within a preset threshold. Step 5: Based on multiple sets of thermal point clouds and corresponding constraint force vectors obtained by switching constraint force groups at least once under constant temperature conditions, identify the force and displacement compliance relationship of the automotive door panel stamping part under the current thermal state online and obtain the compliance coefficient matrix. Step 6: Using the compliance coefficient matrix and the first constraint force vector, calculate the full-field elastic deformation field caused by the first constraint force group, and subtract the elastic deformation field from the first hot point cloud to obtain the equivalent cold free point cloud; Step 7: Align the equivalent cold free state point cloud with the pre-stored cold theoretical shape point cloud, and calculate the normal deviation of each point as the rebound amount.

2. The method for online measurement of springback of stamped automotive door panels according to claim 1, characterized in that, In step 1, based on the positioning reference of the welding tooling of the automotive door panel stamping, the clamping point and the actual production constraint conditions, multiple constraint clamping points are preset, and the constraint clamping points cover the door panel edge, mounting bracket and key stress area of ​​the flange. The servo actuators at each clamping point of the control constraint simulation device output preset clamping force and clamping displacement to simulate the combined rigid and elastic constraints of the actual welding fixture. Force sensors integrated into each clamping point collect clamping force data in real time, integrate the force values ​​and direction parameters of all clamping points, construct the first constraint force vector with dimension matching, and store the record.

3. The method for online measurement of springback of stamped automotive door panels according to claim 2, characterized in that, The method for synchronously acquiring the first thermal point cloud and the first temperature field in step 2 is as follows: A three-dimensional laser scanning module (21) is used to perform a full-domain scan on the automotive door panel stamping part constrained by the first constraint force group, and high-density three-dimensional coordinate point cloud data of the stamping part surface is collected to form a first thermal point cloud; an infrared thermal imaging module (22) is used to synchronously collect the full-domain temperature data of the stamping part to construct a first temperature field corresponding to the pixel level; the acquisition timing of the three-dimensional laser scanning module (21) and the infrared thermal imaging module (22) is bound by the timing synchronization module to ensure that the acquisition time difference between the first thermal point cloud and the first temperature field is less than 10ms.

4. The method for online measurement of springback of stamped automotive door panels according to claim 3, characterized in that, The control method for rapid switching of constraint force groups in step 3 includes: The positioning reference, clamping points, and clamping posture of the automotive door panel stamping part remain completely unchanged. Only the output force values ​​of each servo actuator of the constraint simulation device are adjusted. The force value parameters of each point of the second constraint force group are preset. The constraint stiffness and clamping force of the second constraint force group form a gradient difference with the first constraint force group, and there are no zero-constraint suspended points. The clamping force is dynamically and quickly switched through servo closed-loop control. The time to complete the switching of a single constraint force group is no more than 0.5s. There is no mechanical impact or workpiece displacement during the switching process. After the switching is completed, the data acquisition is performed after stabilizing for 1-2s.

5. The method for online measurement of springback of stamped automotive door panels according to claim 4, characterized in that, The preset threshold and control method for temperature field difference control in step 4 are as follows: The preset threshold for temperature field difference includes the global average temperature difference threshold and the local maximum temperature difference threshold, wherein the global average temperature difference is less than or equal to 0.3℃ and the maximum temperature difference in any local area is less than or equal to 0.5℃; the global temperature distribution data of the second temperature field and the first temperature field are compared in real time. If the temperature difference exceeds the preset threshold, the constant temperature compensation module (31) is activated to perform micro-temperature control on the stamping part. Once the temperature field difference falls back to the preset threshold range and the temperature distribution stabilizes, the second temperature field data is solidified and the second thermal point cloud acquisition is completed.

6. The method for online measurement of springback of stamped automotive door panels according to claim 5, characterized in that, The online identification method for the compliance coefficient matrix in step 5 includes the following sub-steps: Step 501: Under constant temperature conditions, repeat the constraint force group switching operation at least twice to obtain multiple sets of one-to-one corresponding constraint force vectors and thermal displacement point cloud data, and construct a force and displacement sample dataset. Step 502: Denoise and register preprocessing is performed on multiple sets of thermal point clouds, and displacement variables of each constraint point and global measurement point are extracted. A linear force and displacement balance equation is constructed by combining the corresponding constraint force vector. Step 503: The equilibrium equation is solved iteratively using the least squares method. The solution results are combined with the constraint correction results of the mechanical properties of the stamped sheet to identify the compliance coefficient matrix corresponding to the whole domain under the current hot state. The compliance coefficient matrix is ​​a symmetric positive definite matrix that represents the elastic displacement response corresponding to the unit constraint force.

7. The method for online measurement of springback of stamped automotive door panels according to claim 6, characterized in that, Step 6, the method for obtaining the equivalent cold free state point cloud, includes the following sub-steps: Step 601: Substitute the first constraint force vector into the compliance coefficient matrix, and solve for the elastic displacement of each measuring point in the entire domain of the stamping part under the action of the first constraint force through matrix forward operation, and fit to generate a continuous full-field elastic deformation field. Step 602: Decompose the coordinates of the first hot point cloud, deduct the elastic deformation displacement components at the corresponding positions point by point, and eliminate the influence of elastic deformation caused by tooling constraints. Step 603: Based on the constant characteristics of the two temperature fields, eliminate the temperature deformation error and reconstruct the equivalent cold free state point cloud without constraints and elastic deformation.

8. The method for online measurement of springback of stamped automotive door panels according to claim 7, characterized in that, The point cloud alignment preprocessing method in step 7 is as follows: the pre-stored cold-state theoretical shape point cloud is the standard point cloud corresponding to the design model of the automotive door panel stamping part, which is the theoretical shape data in the unconstrained free state at room temperature; the ICP iterative nearest point algorithm is used to perform coarse registration and fine registration of the equivalent cold-state free state point cloud and the cold-state theoretical shape point cloud, and the registration process constrains the translation and rotation degrees of freedom to ensure the consistency of the reference. After registration, abnormal noise points and invalid measurement points at the edge of the point cloud are removed, and the point cloud data of the effective forming area of ​​the stamped part is retained to ensure that the measurement points of the two sets of point clouds correspond one-to-one.

9. The method for online measurement of springback of stamped automotive door panels according to claim 8, characterized in that, The method for calculating and outputting the rebound amount in step 7 includes the following sub-steps: Step 701: For the two sets of registered point clouds, calculate the deviation value of the measuring point along the normal direction of the stamping surface point by point, and correspond the positive and negative deviations of the normal direction to tensile springback and compression springback, respectively. Step 702: Calculate the maximum, minimum, and average springback values ​​and root mean square error of the springback at all measurement points in the entire area, and generate a cloud map of the springback distribution of the door panel stamping part in the entire area. Step 703: Extract local springback data separately for key forming areas such as door panel flanges, curved surfaces, and installation points, and output the regional springback accuracy evaluation results.

10. An online measurement system for the springback of stamped automotive door panels, characterized in that, The system for implementing the online measurement method for the springback of a stamped automotive door panel as described in claim 9 includes a constraint simulation device, a data acquisition device (2), a temperature control device, and a data processing terminal (4). The constraint simulation device includes a multi-point servo clamping mechanism (11), a force sensing module (12), and a drive control module (13). The multi-point servo clamping mechanism (11) is used to position and clamp the automotive door panel stamping and output constraint force groups of different gradients. The force sensing module (12) is used to collect the force value data of each clamping point in real time and construct the constraint force vector. The drive control module (13) is used to realize the rapid and smooth switching of constraint force groups. The data acquisition device (2) includes a three-dimensional laser scanning module (21) and an infrared thermal imaging module (22). The three-dimensional laser scanning module (21) is used to collect thermal point cloud data of stamping parts under different constraint states, and the infrared thermal imaging module (22) is used to synchronously collect temperature field data under corresponding working conditions. The two are matched in time and space through a time synchronization module. The temperature control device includes a constant temperature compensation module (31) and a temperature monitoring module (32). The temperature monitoring module (32) compares the difference between the two temperature fields in real time. The constant temperature compensation module (31) is used to perform micro-temperature control on the stamping parts to ensure that the difference between the two temperature fields collected is within the preset threshold. The data processing terminal (4) has a built-in point cloud preprocessing module (41), a flexibility matrix identification module (42), an elastic deformation solving module (43), a point cloud registration module (44), and a springback calculation module (45), which are used to complete the identification of force and displacement relationship, elastic deformation deduction, point cloud alignment, and full-domain springback calculation, so as to realize the online accurate measurement of the springback of stamped parts.

Images