Automobile accessory welding fixture clamping control method and system

Abstract

The present application relates to the technical field of welding manufacturing, in particular to a kind of automobile accessory welding fixture clamping control method and system, the method includes: the key feature of the welding accessory is measured lightly, deviation sample is obtained;Using deviation sample to obtain the deviation prediction value of unmeasured position and corresponding prediction uncertainty, the clamping error distribution of the preset evaluation area and its each sub-region prediction uncertainty are formed by statistics;When sub-region prediction uncertainty exceeds threshold value, high-precision measurement is carried out to the sub-region to obtain retest deviation, and clamping error distribution is updated;According to the clamping error distribution after updating, the clamping execution parameters of each fixture are set, and clamping strategy is generated;Control drive each fixture to execute clamping action according to clamping strategy, and the clamping execution parameters are adaptively corrected according to sensing feedback.The present application can obtain reliable error distribution at lower measurement cost, reduce the risk of slip or bruise, and improve welding assembly consistency and efficiency.

Description

Technical Field

[0001] This invention relates to the field of welding manufacturing technology, and in particular to a clamping control method and system for the welding process of automotive parts. Background Technology

[0002] In automotive parts manufacturing, body structural components, bracket parts, thin-walled parts, and irregularly shaped parts typically require connection through processes such as spot welding, arc welding, or laser welding. Welding quality and assembly accuracy are highly dependent on the pre-welding clamping and positioning. If the workpiece has issues such as hole position deviation, boundary contour deviation, gap changes, or misalignment in the fixture, it will directly lead to weld position shift, uneven penetration, stress concentration, and even post-weld assembly deviations and rework.

[0003] Currently, most welding fixtures used in production lines employ fixed positioning elements and pneumatic or electric clamping mechanisms to clamp workpieces. To improve clamping consistency, some solutions introduce vision or sensor measurement to detect the workpiece positioning status and perform position compensation or clamping adjustment accordingly. However, existing technologies still generally suffer from the following problems: 1. The disconnect between measurement and control makes it difficult to characterize regional error distributions. Common inspection methods often only measure at a few points or only output pass / fail judgments, making it difficult to obtain the continuous deviation distribution within the evaluation area, and even more difficult to formulate differentiated clamping control strategies for the error characteristics of different sub-regions. When the workpiece has local warping, edge-sensitive zone offset, or weld seam misalignment, control based on a few points is prone to introducing new local errors through local corrections.

[0004] 2. High-precision measurement is costly, and comprehensive retesting impacts cycle time; lightweight measurement has limited reliability. High-precision 3D measurement or high-resolution visual measurement can provide more accurate deviation data, but it is time-consuming, expensive, and sensitive to on-site occlusion / reflection, making it difficult to use frequently across the entire area in welding stations with limited cycle time. In contrast, lightweight measurement has a speed advantage, but it is prone to increased uncertainty and greater fluctuations in repeated measurements on complex surfaces, in areas with occlusion interference, or in areas with indistinct features, leading to risks in directly controlling clamping actions based on measurement results.

[0005] 3. Multi-grip point coordinated control is difficult, easily leading to over-constraint, slippage, or crushing risks. Automotive parts welding fixtures often have multiple clamping points, with clamping force and displacement adjustment being coupled. Without a coordinated setting mechanism, multiple fixtures in the same sub-region may produce mutually canceling or over-constraining clamping effects, causing micro-slippage, torsional deformation, or localized crushing of the workpiece. Simultaneously, if the feedback adjustment of clamping force or displacement lacks boundary constraints and anomaly suppression mechanisms, it may lead to parameter overshoot and vibration amplification, thereby affecting welding stability and equipment safety.

[0006] 4. Thermal deformation during welding causes drift, making it difficult to maintain consistent accuracy with traditional single-clamping methods. Heat input during welding causes material expansion and contraction, as well as changes in structural stiffness, leading to drift in the clamping state over time. Traditional methods often rely on pre-weld positioning and clamping, which makes it difficult to promptly identify and suppress the accumulation of deviations formed during welding, resulting in unstable geometric accuracy and poor quality consistency after welding. Summary of the Invention

[0007] The purpose of this invention is to address the numerous technical problems existing in the current automotive parts welding and clamping technology. This invention proposes a method and system for clamping control of automotive parts welding fixtures, which, while ensuring production cycle time, enables regional assessment and controllable compensation of clamping deviations, improves welding and assembly accuracy and quality consistency, and reduces risks such as slippage, crushing, and vibration.

[0008] To achieve the above objectives, the present invention provides a method for clamping and controlling welding fixtures for automotive parts, which includes the following implementation process: The key features of the components to be welded are subjected to lightweight measurement to obtain deviation samples. The lightweight measurement is a low-density sampling of a limited number of measurement locations of the key features within a predetermined measurement time. By using deviation samples to predict the deviation values ​​and corresponding prediction uncertainties of unmeasured locations, the clamping error distribution of the preset evaluation area and the prediction uncertainties of each sub-region are statistically formed. When the prediction uncertainty of a sub-region exceeds a threshold, a high-precision measurement is performed on the sub-region to obtain the remeasurement deviation and update the clamping error distribution. The high-precision measurement has a higher sampling density and sampling resolution than the lightweight measurement. Based on the updated clamping error distribution, the clamping execution parameters of each fixture are collaboratively set to generate a clamping strategy; The controller drives each gripper to perform gripping actions according to the gripping strategy, and adaptively corrects the gripping execution parameters based on sensor feedback.

[0009] As a further improvement to the above technical solution, the key feature is a reference feature used to characterize the clamping and positioning state of the parts to be welded and the quality of the welding assembly. The reference feature includes at least one of the following: the center of the positioning hole or the mating feature of the positioning pin, the reference surface or the reference boundary feature, the start and end point feature of the weld, and the edge line or contour line feature of the weld neighborhood. The lightweight measurement is used to obtain deviation samples of the reference features, and the deviation samples include at least one of the following: spatial position deviation of the reference features, distance deviation between different reference features, gap deviation, or misalignment deviation.

[0010] As a further improvement to the above technical solution, the step of using deviation samples to predict the deviation prediction value and corresponding prediction uncertainty of the unmeasured position includes: constructing a probabilistic spatial correlation model based on the spatial distance and deviation difference between each measurement position in the deviation sample; performing spatial interpolation prediction on the unmeasured position in the preset evaluation area according to the probabilistic spatial correlation model to obtain the deviation prediction value of the unmeasured position, and outputting the prediction variance or confidence interval width of the unmeasured position based on the probabilistic spatial correlation model as the prediction uncertainty corresponding to the unmeasured position; The probabilistic spatial correlation model establishes a correlation characterization function with the constraint that the correlation decreases with increasing spatial distance, and determines the model parameters by fitting biased samples. The model parameters include at least one of the following: correlation length parameter, noise term parameter, and amplitude parameter.

[0011] As a further improvement to the above technical solution, the statistical formation of the prediction uncertainty of each sub-region of the preset evaluation area includes: performing statistical operations on the prediction uncertainty corresponding to the unmeasured position in each sub-region to obtain the prediction uncertainty of the sub-region; The statistical operation includes at least one of the following: taking the maximum value, mean value, or weighted mean of the prediction uncertainty within the sub-region, or calculating the proportion of unmeasured locations where the prediction uncertainty exceeds a preset uncertainty threshold.

[0012] As a further improvement to the above technical solution, the process of performing high-precision measurement on the sub-region includes: sorting the sub-regions according to the prediction uncertainty from high to low, and selecting at least one of the top-ranked retest positions to perform high-precision measurement. The retest position includes at least one of the following: the untested position with the largest prediction uncertainty, the representative position in the set of untested positions with prediction uncertainty exceeding a preset threshold, and the untested position located in the neighborhood of the weld start and end points or the edge sensitive area; and the clamping error distribution is corrected and updated using the retest deviation corresponding to the retest position.

[0013] As a further improvement to the above technical solution, the generation process of the clamping strategy includes: under the condition of satisfying the first constraint, taking the minimization of the clamping error index of the preset evaluation area or its key sub-region as the optimization objective, and at the same time suppressing the fluctuation of the clamping error index, so as to obtain a clamping strategy with uniform clamping error distribution. The clamping error index includes at least one of the maximum value, mean, or weighted mean of the predicted deviation value within the region, and the volatility includes at least one of the variance, standard deviation, or range of the predicted deviation value within the region. The first constraint includes at least one of the upper limit of the stroke, the upper limit of the clamping force, and the pressure injury risk threshold or the slippage risk threshold of the clamping execution parameters.

[0014] As a further improvement to the above technical solution, the collaborative setting of the clamping execution parameters of each fixture includes: determining the overall clamping correction amount for error suppression based on the updated clamping error distribution, and decomposing the overall clamping correction amount into the clamping execution parameter increments corresponding to each fixture under the second constraint condition. The overall clamping correction includes at least the target correction for the predicted deviation value within the region, the pose target correction for the key features within the region, and the target correction for gaps or misalignments within the region. The decomposition process is obtained by solving a weighted optimization problem, the optimization objective of which includes at least minimizing the weighted norm of the increment of each fixture clamping execution parameter and satisfying the error suppression objective. The second constraint condition includes at least one of the following: the relative displacement difference between fixtures does not exceed a threshold, the relative clamping force difference between fixtures does not exceed a threshold, and the consistency constraint of the clamping action applied to the same sub-region.

[0015] As a further improvement to the above technical solution, the adaptive correction of the clamping execution parameters based on sensor feedback includes: establishing a cascade adjustment structure of an outer adjustment ring based on displacement feedback and an inner adjustment ring based on clamping force feedback. The outer adjustment ring is used to correct the displacement-type clamping execution parameters of each fixture according to the alignment deviation of key features or the change in the clamping error distribution. The inner adjustment ring is used to correct the force-type clamping execution parameters of each fixture according to the clamping force feedback. In the adaptive correction process, a limiting constraint is set on the clamping execution parameters. The limiting constraint includes at least one of the following: stroke limiting for displacement parameters, clamping force limiting for force parameters, adjustment amplitude or adjustment rate limiting. Furthermore, when a slippage risk or vibration index is detected to exceed a preset threshold, the clamping execution parameters are suppressed or adjusted in stages.

[0016] As a further improvement to the above technical solution, the method further includes: during the welding process, performing lightweight measurement on key features according to preset welding nodes or fixed cycles to obtain welding drift deviation, and determining the drift area based on the welding drift deviation; When the welding drift deviation exceeds the preset drift threshold or shows a continuous increasing trend, the clamping error distribution is updated based on the drift area, and the clamping execution parameters of some fixtures corresponding to the drift area are adjusted. When the measurement fluctuation of welding drift deviation exceeds the preset threshold or the prediction uncertainty of the drift area exceeds the threshold, high-precision measurement is performed on the drift area to obtain the remeasured deviation, and the clamping error distribution is updated using the remeasured deviation before the clamping execution parameters are adjusted.

[0017] The present invention also provides a clamping control system for automotive parts welding fixtures, used to perform the above method, the system comprising: The deviation sample generation unit is used to perform lightweight measurement on the key features of the component to be welded to obtain deviation samples. The lightweight measurement is to perform low-density sampling on a limited number of measurement locations of the key features within a predetermined measurement time. The error distribution construction unit is used to obtain the predicted value of the deviation at the unmeasured position and the corresponding prediction uncertainty by using the deviation sample prediction, and to statistically form the clamping error distribution of the preset evaluation area and the prediction uncertainty of each sub-region. The retest deviation generation unit is used to determine when the prediction uncertainty of a sub-region exceeds a threshold, and to perform high-precision measurement on the sub-region to obtain the retest deviation. The high-precision measurement has a higher sampling density and sampling resolution than the lightweight measurement. The update unit is used to update the clamping error distribution based on the retest deviation; The clamping strategy generation unit is used to collaboratively set the clamping execution parameters of each clamp based on the updated clamping error distribution, and generate a clamping strategy. The fixture drive unit is used to control and drive each fixture to perform clamping actions according to the clamping strategy; The adaptive correction unit is used to adaptively correct the clamping execution parameters based on sensor feedback.

[0018] Compared with the prior art, the method and system provided by the present invention have at least the following beneficial effects: 1) By obtaining deviation samples through lightweight measurement and predicting the clamping error distribution for unmeasured positions, the transformation from point measurement to regional error characterization is realized, enabling clamping control to perform differentiated processing on the error characteristics of different sub-regions, thereby improving clamping accuracy and consistency.

[0019] 2) Introduce prediction uncertainty and form sub-region prediction uncertainty. Trigger high-precision measurement according to the threshold to remeasure key areas and update the error distribution, so that high-precision measurement can be intervened as needed, taking into account both measurement reliability and production cycle.

[0020] 3) Based on the updated error distribution, the multi-fixture clamping execution parameters are collaboratively set, and a clamping strategy is generated under constraint conditions and collaborative constraint conditions to reduce the risks of over-constraint, slippage and crushing caused by multi-grip coupling.

[0021] 4) By cascade adjustment of the outer displacement ring and the inner clamping force ring, as well as amplitude limiting constraints, anomaly suppression, or staged adjustment mechanisms, online adaptive correction of clamping execution parameters is achieved, thereby improving the stability and safety of the clamping process.

[0022] 5) Introducing a drift deviation monitoring and on-demand retesting and updating mechanism during the welding process can suppress the deterioration of the clamping state caused by thermal drift, and further improve the geometric stability and welding quality consistency during the welding process. Attached Figure Description

[0023] Figure 1 This is an operation flowchart of the automotive parts welding fixture clamping control method in an embodiment of the present invention; Figure 2 This is a schematic diagram of the clamping control system for automotive parts welding fixtures in an embodiment of the present invention. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be further described below with reference to the accompanying drawings. It should be understood that the embodiments described herein are only some examples of the present invention and are not intended to limit the present invention; equivalent modifications or substitutions made by those skilled in the art without departing from the concept of the present invention and without creative effort should all fall within the protection scope of the present invention.

[0025] The welding fixture clamping control method for automotive parts provided by this invention is applicable to scenarios involving clamping, positioning, and stabilizing the parts to be welded in welding stations. The parts to be welded can be brackets, connecting plates, housings, etc. The welding fixture includes multiple independently driveable clamps or grippers, collectively referred to as fixture actuators. Each fixture actuator applies a clamping action to the parts to be welded to suppress clamping deviations and maintain welding assembly quality. The fixture actuators can be driven by electric cylinders, pneumatic cylinders, or servo actuators, and can be configured with sensors to collect the clamping process status.

[0026] To facilitate implementation, a unified coordinate system is established. A fixture coordinate system is established using the fixture base or workstation datum, and a nominal coordinate system is established using the nominal model (CAD) of the component to be welded or the tooling positioning datum. The measuring device obtains the extrinsic parameter relationship between the measuring coordinate system and the fixture coordinate system through calibration, thereby uniformly transforming the measurement results to be expressed in the fixture coordinate system. Specifically, such as... Figure 1 As shown, the method includes the following implementation steps: The key features of the components to be welded are subjected to lightweight measurement to obtain deviation samples. The lightweight measurement is a low-density sampling of a limited number of measurement locations of the key features within a predetermined measurement time. By using deviation samples to predict the deviation values ​​and corresponding prediction uncertainties of unmeasured locations, the clamping error distribution of the preset evaluation area and the prediction uncertainties of each sub-region are statistically formed. When the prediction uncertainty of a sub-region exceeds a threshold, a high-precision measurement is performed on the sub-region to obtain the remeasurement deviation and update the clamping error distribution. The high-precision measurement has a higher sampling density and sampling resolution than the lightweight measurement. Based on the updated clamping error distribution, the clamping execution parameters of each fixture are collaboratively set to generate a clamping strategy; The controller drives each gripper to perform gripping actions according to the gripping strategy, and adaptively corrects the gripping execution parameters based on sensor feedback.

[0027] The key features refer to the reference geometric features used to characterize the clamping and positioning state of the parts to be welded and the quality of the welding assembly. The optional implementation methods include at least one of the following: the center of the positioning hole or the mating feature of the positioning pin, the reference surface or the reference boundary feature, the start and end point feature of the weld, and the edge line or contour line feature of the weld neighborhood.

[0028] The lightweight measurement method is used to perform low-density sampling at a limited number of measurement locations of key features within a predetermined measurement time to obtain sparse measurement results of key features and form deviation samples. The implementation of lightweight measurement is not limited to a single sensor; at least one of a vision measurement device, a laser displacement sensor, and a depth measurement device can be used. Low-density sampling is achieved by setting a smaller number of measurement locations, a larger sampling interval, or a lower number of sampling points. For a vision measurement device, an industrial camera can be used with a directional light source to acquire image data. Under the premise of meeting the predetermined measurement time, images are captured at limited locations such as positioning holes, reference boundaries, weld start and end points, or weld neighborhood contours. Pixel coordinates of key features are extracted using methods such as edge detection, circular hole fitting, corner point extraction, and line segment fitting. Coordinate transformation is then performed using camera calibration parameters to obtain the geometric quantities of the key features. For a laser displacement sensor, point measurements or a small number of line scans can be performed at limited locations on the height, gap, or misalignment of the weld neighborhood to output displacement or height difference. For depth measurement devices, structured light or binocular depth cameras can be used to output local depth data or sparse point sets in low-density sampling mode, which can be used to extract the three-dimensional coordinates or contour lines of key points, thereby completing sparse measurement in a short time.

[0029] The aforementioned deviation sample refers to the deviation data record obtained after measuring key features, which may include measurement location identifiers (point / line / region index), measured geometric quantities of key features (such as coordinates, distance, gap, misalignment), and their deviation values. The deviation value is calculated from the measured value and the nominal value. For example, spatial position deviation: the difference between the measured coordinates and the nominal coordinates of the key point forms a deviation vector; distance deviation: the difference between the measured distance and the nominal distance between two key features; gap / misalignment deviation: the difference between the relative height difference on both sides of the weld and the measured and nominal values ​​of the lap gap.

[0030] The preset evaluation area is a region related to the clamping effect, which can cover the weld neighborhood, the positioning hole neighborhood, and thin-walled easily deformable areas. The evaluation area can be discretized into several prediction locations, such as grid points or feature sampling points. A spatial correlation model is constructed based on the deviation samples to perform spatial prediction for each prediction location, thereby obtaining the deviation prediction value. The spatial correlation model can employ Kriging interpolation, Gaussian process regression, radial basis function interpolation, or distance-weighted interpolation models, etc. Their common feature is that they utilize the spatial distance and deviation variation law to achieve inference from measured points to unmeasured points.

[0031] The prediction uncertainty is used to characterize the reliability of the predicted deviation at the corresponding location. Its calculation can be directly output from a spatial correlation model or obtained from prediction statistics. For example, when using a probabilistic spatial model, the prediction variance can be output; or the uncertainty index can be calculated based on the width of the prediction confidence interval. Generally, the prediction uncertainty increases when the distance between the predicted location and the measured points increases, the distribution of the measured points becomes sparse, or local deviations change drastically.

[0032] The clamping error distribution is formed by summing the predicted deviation values ​​of each predicted position within the evaluation area. It can be recorded as an error field data structure, such as a mapping table of position coordinates and prediction deviations or a rasterized matrix, for subsequent clamping strategy generation. Furthermore, the evaluation area is divided into multiple sub-regions, and the uncertainty of the predicted position within each sub-region is statistically analyzed to obtain the sub-region prediction uncertainty. The statistical method can be the maximum value, mean, or weighted mean, etc., to reflect the overall reliability of the prediction results in that sub-region.

[0033] The high-precision measurement is used to obtain remeasurement deviations in local sub-regions with high prediction uncertainty. Compared to lightweight measurement, this method offers higher sampling density and resolution. The sampling density can be achieved by increasing the number of measurement locations, reducing the sampling interval, or increasing the number of sampling points. The sampling resolution can be achieved by improving imaging resolution, increasing depth or displacement measurement resolution, reducing the scanning step size, or using more rigorous calibration and multi-view measurement. The high-precision measurement can be implemented in at least one of the following ways: using structured light or laser scanning in a high-density sampling mode to acquire point clouds or high-resolution contour data of the sub-region; using visual measurement for higher resolution imaging and enhanced illumination control, or using more rigorous calibration and multi-view measurement to improve feature extraction accuracy; or using contact point measurement or a high-precision displacement measurement device to perform precise measurements on selected remeasurement locations. At least one remeasurement location within the sub-region is selected for high-precision measurement to obtain the remeasurement deviation. The remeasurement deviation is then added to the sample set as a new deviation sample, and spatial prediction and prediction uncertainty calculation are re-executed to update the clamping error distribution and the sub-region prediction uncertainty, thereby achieving a correction update of the error distribution in the evaluation area. The clamping execution parameters are used to describe the clamping settings of each clamping end, and may include at least clamping displacement settings or clamping force settings. To achieve coordinated settings, in one possible implementation, the influence relationship between changes in clamping execution parameters and changes in deviation in the evaluation area is established: the execution parameters of each clamp are composed into a parameter vector, and the predicted deviation values ​​in the evaluation area are composed into an error vector. An influence matrix or mapping relationship is constructed through calibration test data, tooling experience models, or simplified mechanical relationships. Under the constraints of clamping stroke and clamping force, the parameter vector that reduces the error vector is solved, thereby obtaining the coordinated setting results of each clamping end and forming a clamping strategy.

[0034] During the clamping action, sensor feedback data is collected, and the sensors include at least one of the following: a displacement sensor, used to collect the displacement of the clamping end, the position status, or the amount of displacement change, such as an encoder or a linear displacement sensor; a pressure sensor, used to collect the clamping force, clamping pressure, or the amount of change thereof, such as a tension-compression force sensor or a pressure sensor; and a visual feedback device, used to quickly re-measure key features to obtain alignment deviation or deviation change, for evaluating the geometric state after clamping.

[0035] The adaptive correction method is implemented as follows: during the clamping process, the real-time collected feedback values ​​such as displacement / clamping force are compared with the set values ​​corresponding to the clamping strategy; when the feedback value deviates from the set value or the alignment deviation of the key feature does not meet the preset requirements, the clamping execution parameters of the corresponding fixture are iteratively adjusted, and under the premise of meeting the safety boundary constraints such as fixture stroke and clamping force, the clamping state is gradually converged to the target range, thereby improving the clamping stability and consistency.

[0036] In one embodiment, the key feature is a reference feature used to characterize the clamping and positioning state of the component to be welded and the quality of the weld assembly. The reference feature can be jointly determined by the tooling positioning reference, the nominal model of the component, and the welding process requirements, and includes at least one of the following: the center of the positioning hole or the mating feature of the positioning pin, used to reflect the planar position and angular offset of the component under the constraint of the tooling positioning element; the reference surface or reference boundary feature, used to reflect the fitting state, overturning and warping of the component, and the offset of the boundary relative to the tooling; the start and end point features of the weld, used to reflect the assembly offset risk of the arc start and end positions of the weld; and the edge line or contour line feature of the weld neighborhood, used to reflect the contour changes, alignment state, and local geometric consistency of the weld neighborhood.

[0037] In this embodiment, lightweight measurement is used to obtain deviation samples of the aforementioned reference features. The deviation samples are data records formed after measuring the reference features, and at least include the measured geometric quantities of the reference features and the deviation results corresponding to the nominal values; the nominal values ​​may be derived from the nominal model of the accessory, the tooling positioning reference, or the calibrated standard part reference.

[0038] In one optional implementation, when performing lightweight measurement on the center of the positioning hole or the mating feature of the positioning pin, image or depth data of the neighborhood of the positioning hole or the positioning pin can be collected. By identifying the hole boundary / pin profile and performing geometric fitting, the position of the hole center or the relative mating position of the hole and the pin can be determined, thereby obtaining the measured spatial position of the reference feature.

[0039] In one optional implementation, when performing lightweight measurements on the reference surface or reference boundary features, several representative measurement positions can be selected at the reference surface or boundary to obtain local height, boundary line position or contour point information, thereby obtaining the reference surface fit state or the measured position of the reference boundary relative to the tooling.

[0040] In one optional implementation, when performing lightweight measurements on the start and end point features of the weld and the edge or contour features of the weld neighborhood, image or depth data near the planned path of the weld can be collected, the start point identifier, end point identifier, and edge or contour of the weld neighborhood can be extracted, and their measured geometric information in the fixture coordinate system can be obtained.

[0041] The deviation sample includes at least one of the following: spatial position deviation of the reference feature, distance deviation between different reference features, gap deviation, or misalignment deviation. When the reference feature is a feature with a clear spatial position meaning, such as a hole center, weld start and end point, or boundary feature point, the measured spatial position of the feature is compared with its corresponding nominal spatial position to obtain the spatial position deviation of the reference feature. This deviation can be used to reflect the overall translation, angular offset, or local offset of the component in the clamping state. When it is necessary to reflect the relative geometric relationship between multiple reference features, two reference features (e.g., the centers of two positioning holes, the hole center and the weld start point, the boundary feature point and the hole center, etc.) can be selected to calculate their actual distance and compare it with the nominal distance to obtain the distance deviation. The distance deviation can be used to reflect the overall tensile / compression trend or local geometric deformation trend of the component caused by clamping or initial positioning errors. When there is an assembly gap control requirement between the component to be welded and the lap joint, several measurement positions can be selected in the weld neighborhood or lap boundary to obtain the actual gap and compare it with the nominal gap to form the gap deviation. The gap deviation can be used to reflect the risk of insufficient lap fit, local suspension, or assembly offset. When there are requirements for height alignment or misalignment control on both sides of the weld, measurement positions can be selected in the weld neighborhood to obtain the surface height or boundary height information on both sides. The misalignment deviation is then formed based on the relative height difference between the two sides. The misalignment deviation can be used to reflect the step effect, local deformation, or misalignment risk in the weld neighborhood.

[0042] To facilitate subsequent prediction and statistics of deviation samples, deviation samples can be recorded in an itemized manner. Each sample should include at least the measurement location identifier, the corresponding reference feature type, the measured geometric quantity, the nominal value source identifier, the deviation type, and the deviation result. When there are multiple deviation types, multiple sample records can be generated at the same measurement location, or one or more of the following can be stored simultaneously in the same sample record as fields: spatial position deviation, distance deviation, gap deviation, and misalignment deviation.

[0043] After completing the lightweight measurement of the key features of the components to be welded and obtaining deviation samples, in order to obtain the deviation status of the unmeasured positions within the preset evaluation area, this embodiment further performs spatial prediction based on the deviation samples and generates corresponding prediction uncertainties for the prediction results. Specifically, the step of using deviation samples to predict the deviation prediction values ​​and corresponding prediction uncertainties of the unmeasured positions includes: constructing a probabilistic spatial correlation model based on the spatial distance and deviation differences between each measurement position in the deviation samples; performing spatial interpolation prediction on the unmeasured positions within the preset evaluation area according to the probabilistic spatial correlation model to obtain the deviation prediction values ​​of the unmeasured positions, and outputting the prediction variance or confidence interval width of the unmeasured positions as the prediction uncertainty corresponding to the unmeasured positions based on the probabilistic spatial correlation model; the probabilistic spatial correlation model establishes a correlation characterization function with the correlation decreasing with increasing spatial distance as a constraint, and determines the model parameters by fitting the deviation samples, wherein the model parameters include at least one of correlation length parameters, noise term parameters, and amplitude parameters.

[0044] In one specific implementation, spatial prediction of unmeasured locations within a preset evaluation area based on deviation samples can be achieved by constructing a probabilistic spatial correlation model. The probabilistic spatial correlation model can be a Kriging interpolation model or a Gaussian process regression model, among other probabilistic spatial models. Their common feature is that they establish correlation constraints using the spatial distance between measurement locations and the deviation variation law, enabling the model to simultaneously provide an uncertainty characterization of the prediction result while outputting the predicted deviation value for the unmeasured location.

[0045] Specifically, the deviation samples are first organized to obtain the positional description of each measurement location in the fixture coordinate system, as well as the corresponding deviation value. The deviation value can be any one of the following under the same deviation type: spatial position deviation, distance deviation, gap deviation, or misalignment deviation, and a sample set is formed with each measurement location. Then, the spatial distance between any two measurement locations is calculated within the measurement location set to characterize their proximity. For the same deviation type, the deviation difference between any two measurement locations is calculated to characterize the degree to which the deviation changes with spatial position. Based on the correspondence between spatial distance and deviation difference, a constraint is established that the correlation decreases as the spatial distance increases. A correlation characterization function is then constructed to ensure that measurement locations that are closer and have smaller deviation differences have higher correlation, while measurement locations that are farther away or have larger deviation differences have lower correlation.

[0046] Based on the above, the model parameters of the probabilistic spatial correlation model are determined by fitting the bias samples. The model parameters include at least one of the following: correlation length parameter, noise term parameter, and amplitude parameter. The correlation length parameter characterizes the scale range of correlation decay with distance; the noise term parameter characterizes the intensity of measurement error or local random disturbance; and the amplitude parameter characterizes the overall magnitude of the bias change. Fitting methods can include maximum likelihood estimation, cross-validation, or parameter optimization based on residual consistency, to ensure that the model can effectively reflect the correlation between measurement locations and the trend of bias change with spatial location.

[0047] In one specific implementation, spatial interpolation prediction of unmeasured locations within a predefined evaluation area based on a probabilistic spatial correlation model can be performed as follows: The predefined evaluation area is discretized to obtain several unmeasured locations; for any unmeasured location, based on the correlation characterization function and the determined model parameters, the correlation between the unmeasured location and multiple measurement locations is calculated, and the interpolation weight is determined by the correlation; among them, measurement locations that are spatially close to the unmeasured location and show a gradual change in deviation in the sample have relatively large interpolation weights; measurement locations that are spatially far from the unmeasured location or show drastic changes in deviation within their neighborhood have relatively small interpolation weights. Subsequently, the deviation values ​​corresponding to the measurement locations are combined according to the interpolation weights to obtain the deviation prediction value of the unmeasured location. Repeating the above process for each unmeasured location within the predefined evaluation area yields a set of deviation prediction results within the evaluation area, providing a data basis for subsequently forming a clamping error distribution.

[0048] In one specific implementation, the prediction uncertainty can be output synchronously by a probabilistic spatial correlation model when outputting the predicted deviation value. The prediction variance or confidence interval width can be used as the uncertainty index. The prediction variance reflects the dispersion of the predicted values ​​for the unmeasured location; a larger prediction variance indicates a lower reliability of the prediction for that location. The confidence interval width reflects the range of possible values ​​for the predicted value of the unmeasured location; a wider confidence interval indicates a lower reliability of the prediction for that location. Generally, the prediction uncertainty increases when the unmeasured location is far from the measured location, the measured locations are sparsely distributed, or the deviation changes significantly within the neighborhood of the unmeasured location. Conversely, the prediction uncertainty decreases when the unmeasured location is in a region where the measured locations are relatively dense and the deviation changes gradually. This achieves the synchronous output of the predicted deviation value and its prediction uncertainty for the unmeasured location.

[0049] After obtaining the predicted deviation values ​​and corresponding prediction uncertainties for unmeasured locations within the preset evaluation area, this embodiment further divides the evaluation area into sub-regions to achieve a regionalized expression of the reliability of the prediction results and subsequent retesting decisions, and statistically analyzes the prediction uncertainties within each sub-region. Specifically, the statistical analysis of the prediction uncertainties for each sub-region of the preset evaluation area includes: performing statistical operations on the prediction uncertainties corresponding to unmeasured locations within each sub-region to obtain the prediction uncertainties for that sub-region; the statistical operations include at least one of taking the maximum value, mean, or weighted mean of the prediction uncertainties within the sub-region, or calculating the proportion of unmeasured locations whose prediction uncertainties exceed a preset uncertainty threshold.

[0050] In one specific implementation, the sub-region division can be determined according to process and structural characteristics. The preset evaluation region can cover the weld neighborhood, the positioning hole neighborhood, or thin-walled, easily deformable areas, etc. To ensure that the statistical results reflect the prediction reliability of different local locations, the evaluation region can be divided into multiple sub-regions. The sub-region division method can be any of the following: regular grid segmentation, segmentation according to weld path, segmentation according to positioning feature neighborhood, or segmentation according to structural weak points. Each sub-region contains several unmeasured locations, which are candidate locations for which prediction uncertainty has been generated during the prediction process.

[0051] In one specific implementation, the statistical operation on the prediction uncertainty corresponding to the unmeasured locations in each sub-region can be completed as follows: For any sub-region, the prediction uncertainty of all unmeasured locations in the sub-region is summarized to form the uncertainty set of the sub-region; based on the uncertainty set, a preset statistical operation is performed to output the sub-region prediction uncertainty of the sub-region, which is used to characterize the reliability of the overall prediction result of the sub-region.

[0052] In one specific implementation, when using the maximum value statistical operation, the sub-region prediction uncertainty is taken as the value corresponding to the unmeasured position with the largest prediction uncertainty in the sub-region, to reflect the dominant role of the least reliable position in the sub-region in terms of risk; when using the mean statistical operation, the sub-region prediction uncertainty is taken as the average level of prediction uncertainty in the sub-region, to reflect the overall prediction reliability of the sub-region; when using the weighted mean statistical operation, the prediction uncertainties of each unmeasured position in the sub-region are weighted and summed to form the sub-region prediction uncertainty. The weights can be set according to the proximity of the unmeasured position to the start and end points of the weld, the proximity to the positioning hole / positioning pin, or the proximity to the edge sensitive area, so that the positions that have a greater impact on the welding assembly quality have a higher proportion in the statistics.

[0053] In one specific implementation, when using statistical calculations to determine the percentage of locations exceeding a threshold, an uncertainty threshold is pre-set to distinguish between unmeasured locations where the prediction uncertainty is at an acceptable level and those at a relatively high level. Then, the number of unmeasured locations within the sub-region whose prediction uncertainty exceeds the threshold is counted and compared with the total number of unmeasured locations within the sub-region to obtain the percentage of unmeasured locations exceeding the threshold. A larger percentage indicates a more concentrated concentration of locations with high prediction uncertainty within the sub-region, and a lower overall prediction reliability for the sub-region; a smaller percentage indicates a higher overall prediction reliability for the sub-region. Thus, without altering the prediction model structure, the spatial concentration of uncertainty within a sub-region can be reflected in the form of a percentage.

[0054] In one specific implementation, the above statistical operations can be fixed in advance as one of them, or one of them can be selected as the output index of the sub-region prediction uncertainty according to the process requirements, so as to be used in conjunction with the subsequent threshold determination logic, thereby realizing the quantitative characterization of the sub-region prediction reliability.

[0055] After obtaining the clamping error distribution of the preset evaluation area and statistically forming the prediction uncertainty of each sub-region, in order to improve the reliability of the prediction results in the high uncertainty region, this embodiment further triggers high-precision measurement of the sub-region and uses the remeasurement results to correct and update the clamping error distribution. Specifically, the process of performing high-precision measurement on the sub-region includes: sorting the sub-regions according to the prediction uncertainty from high to low, selecting at least one of the top-ranked remeasurement positions to perform high-precision measurement; the remeasurement position includes at least one of the following: the unmeasured position with the largest prediction uncertainty, the representative position in the set of unmeasured positions with prediction uncertainty exceeding a preset threshold, and the unmeasured position located in the neighborhood of the weld start and end points or the edge sensitive area; and using the remeasurement deviation corresponding to the remeasurement position to correct and update the clamping error distribution.

[0056] In one specific implementation, when the prediction uncertainty of a certain sub-region exceeds a threshold, that sub-region is designated as a retest sub-region. Within this sub-region, the prediction uncertainties corresponding to all untested locations are summarized to form an uncertainty ranking list. The ranking can be from highest to lowest, so that the locations with the lowest prediction reliability are given priority for inclusion in the retest candidate range.

[0057] In one specific implementation, the selection of at least one retesting position ranked first can be implemented using the following strategies: First, directly select the first-ranked untested position as the retesting position to ensure that the highest prediction uncertainty within the sub-region is prioritized for precise measurement; Second, when there are multiple untested positions within the sub-region whose prediction uncertainty exceeds a preset threshold, these untested positions are formed into an over-threshold set, and at least one representative position is selected from this set as the retesting position; the representative position can be selected according to the principle of balanced spatial distribution to avoid excessive concentration of retesting positions in local areas; or it can be selected according to the principle of prioritizing process attention, so that positions with a greater impact on welding assembly quality are prioritized for retesting; Third, untested positions in the weld start and end point neighborhood or edge-sensitive area are used as priority candidate retesting positions. The weld start and end point neighborhood can be understood as a preset range near the arc initiation and arc termination positions, and the edge-sensitive area can be understood as an area close to the free edge, weak boundary, or area prone to warping deformation of the component. By including the untested positions in the above-mentioned neighborhood or sensitive area as candidates for retesting positions, the retesting can be more closely aligned with the key risk points of welding quality and clamping stability.

[0058] In one specific implementation, the high-precision measurement can employ a measurement method with higher resolution or higher precision than the lightweight measurement to precisely measure the remeasurement position. The output data of the high-precision measurement includes at least the measured geometric quantities of the reference features corresponding to the remeasurement position, and further calculates the remeasurement deviation. The remeasurement deviation is the deviation between the measured value and the corresponding nominal value at the remeasurement position, and its deviation type can be consistent with the aforementioned deviation sample, including one or more of spatial position deviation, distance deviation, gap deviation, or misalignment deviation.

[0059] In one specific implementation, the correction and update of the clamping error distribution using the re-measurement deviation corresponding to the re-measurement position can be carried out as follows: The re-measurement deviation is added as a new deviation sample to the deviation sample set, enabling the sub-region to obtain more reliable constraint information at the deviation sample level. Subsequently, based on the updated deviation sample set, the spatial correlation model is reconstructed or its parameters are updated, and spatial interpolation prediction and prediction uncertainty calculation are re-performed for the unmeasured positions within the evaluation region, thereby obtaining the updated clamping error distribution and the updated prediction uncertainty of the sub-region. Through the above correction and update, the uncertainty of deviation prediction within the re-measurement sub-region can be reduced, making the clamping error distribution more consistent with the actual clamping state.

[0060] In one specific implementation, when multiple retest locations are selected, the multiple retest deviations can be merged into the deviation sample set before a unified update is performed to reduce redundant calculations; alternatively, they can be merged one by one according to the retest locations and updated sequentially, so as to evaluate the uncertainty reduction effect in real time during the retesting process and thus decide whether to continue adding retest locations in that sub-region. Neither of the above methods changes the basic logic of using retest deviations to correct and update the clamping error distribution.

[0061] After predicting the clamping error distribution of the preset evaluation area through deviation samples and completing the retesting and correction update, in order to effectively suppress the clamping deviation during the clamping process, this embodiment further generates a clamping strategy based on the updated clamping error distribution. Specifically, the generation process of the clamping strategy includes: under the condition of satisfying the first constraint, taking the minimization of the clamping error index of the preset evaluation area or its key sub-regions as the optimization objective, and simultaneously suppressing the fluctuation of the clamping error index to obtain a clamping strategy with a uniform clamping error distribution; the clamping error index includes at least one of the maximum value, mean, or weighted mean of the deviation prediction value in the region, the fluctuation includes at least one of the variance, standard deviation, or range of the deviation prediction value in the region, and the first constraint includes at least one of the upper limit of the stroke of the clamping execution parameters, the upper limit of the clamping force, and the pressure damage risk threshold and the slippage risk threshold.

[0062] In one specific implementation, the clamping strategy is used to provide the clamping setting results for each clamping actuator, wherein the clamping execution parameters of the clamping actuator include at least a displacement setting value or a clamping force setting value. To facilitate solution and implementation, the predicted deviation values ​​of each predicted position within the evaluation area can be used as the basis for error evaluation, and these can be summarized to form a clamping error index and a fluctuation index. The combination of clamping execution parameter settings is then determined under the premise of satisfying the first constraint condition.

[0063] In one specific implementation, the clamping error index is used to characterize the overall deviation level of the evaluation area or key sub-area. Its calculation can be performed in any of the following ways: First, take the maximum value of the predicted deviation within the area to reflect the upper limit of the deviation at the most unfavorable position within the area; Second, take the mean value of the predicted deviation within the area to reflect the average deviation level of the area as a whole; Third, take the weighted mean value of the predicted deviation within the area, where the weight can be set according to the degree of correlation between the predicted position and the weld start and end point neighborhood, the positioning hole neighborhood, the thin-walled easily deformable area, or the edge sensitive area, so that the position that has a greater impact on the welding assembly quality has a higher proportion in the error index.

[0064] When selecting a key sub-region as the optimization object, the selection criteria for the key sub-region can be determined first, such as the weld neighborhood, the positioning hole neighborhood, or the thin-walled easily deformable area, and the clamping error index can be calculated only for the deviation prediction value within the key sub-region.

[0065] In one specific implementation, the volatility is used to characterize the dispersion of predicted deviation values ​​within the evaluation area or key sub-regions, to avoid significant unevenness in the clamping effect in local areas. Volatility can be determined in either of the following ways: first, by calculating the variance or standard deviation of the predicted deviation values ​​within the region to reflect the spatial dispersion of the deviations; second, by calculating the range of the predicted deviation values ​​within the region to reflect the maximum difference in the predicted deviation values ​​within the region. By suppressing volatility during the clamping strategy generation process, the predicted deviation values ​​can be made more uniform within the region, thereby reducing the assembly risk caused by excessively large local deviations.

[0066] In one specific implementation, the first constraint condition is used to ensure the feasibility and process safety of the clamping strategy, and includes at least one of the following constraints: First, the upper limit of the stroke of the clamping execution parameter, used to limit the displacement setting value of the clamping execution end to not exceed the maximum stroke allowed by the mechanism; Second, the upper limit of the clamping force, used to limit the clamping force setting value of the clamping execution end to not exceed the maximum clamping force allowed by the clamping structure and process; Third, the pressure damage risk threshold, used to limit the clamping force or contact pressure to not exceed the risk threshold that may cause pressure damage, local dents or deformation of the accessory surface; Fourth, the slippage risk threshold, used to limit the risk of insufficient clamping stability caused by insufficient clamping force or improper displacement setting, and to avoid positioning drift or surface scratches caused by relative slippage during the clamping process.

[0067] In one implementation, the pressure damage risk threshold and slippage risk threshold can be preset based on material properties, surface treatment status, clamping contact area, friction conditions, and tooling experience data, and the candidate clamping execution parameter combinations are constrained and screened during strategy solution and output.

[0068] In one specific implementation, minimizing the clamping error index and suppressing volatility can be achieved through a solution process. Specifically, a search or iterative adjustment can be performed within the candidate clamping execution parameter combination space to reduce both the clamping error index and volatility simultaneously. Each time a candidate parameter combination is generated, it is checked whether it meets the first constraint condition. If a candidate parameter combination does not meet the first constraint condition, it is discarded, and a new candidate combination is generated. When a candidate parameter combination meets the first constraint condition and both the clamping error index and volatility reach the expected levels, the parameter combination is output as the clamping strategy. Through this method, a clamping strategy with a low overall deviation level and a relatively uniform deviation distribution within the region can be obtained, providing a target setting basis for subsequent clamping execution and sensor feedback correction.

[0069] In the process of generating a clamping strategy based on the updated clamping error distribution, in order to enable multiple clamping actuators to work together to achieve error suppression, this embodiment further coordinates the clamping execution parameters of the clamps. Specifically, the coordinated setting of the clamping execution parameters of each clamp includes: determining the overall clamping correction amount for error suppression based on the updated clamping error distribution; under the condition of satisfying the second constraint, decomposing the overall clamping correction amount into the clamping execution parameter increments corresponding to each clamp; the overall clamping correction amount includes at least the target correction amount for the predicted deviation value in the region, the pose target correction amount for the key features in the region, and the target correction amount for the gap or misalignment in the region; the decomposition process is obtained by solving a weighted optimization problem, the optimization objective of which includes at least minimizing the weighted norm of the clamping execution parameter increments of each clamp and satisfying the error suppression objective; the second constraint includes at least one of the following: the relative displacement difference between clamps does not exceed a threshold, the relative clamping force difference between clamps does not exceed a threshold, and the consistency constraint of the clamping action applied to the same sub-region.

[0070] In one specific implementation, the overall clamping correction amount is used to describe the deviation trend or assembly deviation target that needs to be suppressed at the evaluation area level. It can be determined jointly by the updated clamping error distribution and the deviation state of key features. Specifically, based on the spatial distribution of predicted deviation values ​​within the evaluation area, the main direction and main concentration area of ​​the overall deviation in the area can be determined, and a target correction amount for the area deviation can be formed accordingly. Simultaneously, based on the spatial position deviation or relative geometric relationship deviation of key features, a target correction amount for the pose of key features can be formed, causing the key features to converge towards their nominal state. Furthermore, when the evaluation area contains predicted deviation values ​​related to gaps or misalignments, a target correction amount for gaps or misalignments can be formed, ensuring that the assembly gap and misalignment levels in the weld neighborhood meet preset requirements. The above three types of target correction amounts can be used individually or in combination according to process focus, serving as a comprehensive error suppression target at the area level.

[0071] In one specific implementation, the decomposition of the overall clamping correction into the increments of clamping execution parameters corresponding to each fixture can be carried out as follows: First, the interaction relationship between the fixture execution end and the evaluation area is determined. This interaction relationship can be established through fixture arrangement position, clamping contact point distribution, fixture action direction, and tooling experience data, used to characterize the influence trend of the displacement setpoint and clamping force setpoint changes of each fixture execution end on the deviation state of the evaluation area. Subsequently, the increment of the clamping execution parameter of each fixture execution end is taken as the variable to be solved, and the overall clamping correction is taken as the error suppression target. The increment of the clamping execution parameter of each fixture execution end is obtained through the solution process, so that the combination of the increments of multiple fixture execution ends can jointly approximate the target area described by the overall clamping correction.

[0072] In one specific implementation, the weighted optimization problem aims to ensure that the adjustment range of each fixture's execution end is as reasonable as possible while satisfying the error suppression objective, avoiding local damage or deformation caused by excessive adjustment of any particular fixture. Specifically, different weights can be assigned to different fixture execution ends. These weights can be set based on the fixture's structural stiffness, clamping contact area, sensitivity of the clamping position, or correlation with key features. During the solution process, fixture execution ends with larger weights are given priority to have smaller adjustment ranges, making the adjustment more likely to be completed by fixture execution ends more suitable for the task. Thus, the solution achieves the effect of minimizing the overall adjustment degree of the clamping execution parameter increment of each fixture while satisfying the error suppression objective.

[0073] In one specific implementation, the second constraint condition is used to ensure that the decomposition results meet the process consistency and safety of the fixture collaborative clamping, and includes at least one of the following constraints: First, the relative displacement difference between fixtures does not exceed a threshold, which is used to limit the displacement setting difference between adjacent fixtures or collaborative fixtures, and avoid local forced deformation or stress concentration caused by excessive displacement difference; Second, the relative clamping force difference between fixtures does not exceed a threshold, which is used to limit the clamping force difference between adjacent fixtures or collaborative fixtures, and avoid slippage or local crushing caused by uneven distribution of clamping force; Third, the consistency constraint of clamping action applied to the same sub-region is used to limit the inconsistency in clamping direction, clamping strength or adjustment trend of multiple fixture execution ends that play a major role in the same sub-region, so as to avoid mutual cancellation or mutual resistance of clamping actions in the sub-region, thereby ensuring that the clamping effect in the sub-region is consistent and repeatable.

[0074] In solving the weighted optimization problem, the incremental parameter of the candidate clamping must simultaneously satisfy the second constraint condition mentioned above. If it is not satisfied, the candidate solution is adjusted or discarded until a decomposition result that satisfies the constraint and the error suppression objective is obtained.

[0075] Through the above process, an interpretable overall clamping correction amount can be formed at the regional level, and the clamping execution parameter increments of each clamping execution end can be obtained under cooperative constraints, so that the clamping execution parameter settings have cooperativeness, executableness and consistency, thereby providing a stable parameter basis for subsequent clamping action execution and sensor feedback correction.

[0076] During the clamping process, each fixture completes the clamping action setting according to the clamping strategy and begins to execute the clamping. In order to compensate for the influence of factors such as fixture execution error, friction state change and local elastic deformation of workpiece on the clamping effect, this embodiment further adaptively corrects the clamping execution parameters based on sensor feedback so that the alignment state of key features and the distribution of clamping error continuously meet the preset requirements. Specifically, the adaptive correction of clamping execution parameters based on sensor feedback includes: establishing a cascaded adjustment structure of an outer adjustment ring based on displacement feedback and an inner adjustment ring based on clamping force feedback. The outer adjustment ring is used to correct the displacement-type clamping execution parameters of each fixture based on the alignment deviation of key features or the change in the distribution of clamping errors. The inner adjustment ring is used to correct the force-type clamping execution parameters of each fixture based on clamping force feedback. During the adaptive correction process, a limiting constraint is set on the clamping execution parameters. The limiting constraint includes at least one of the following: stroke limiting for displacement-type parameters, clamping force limiting for force-type parameters, adjustment amplitude or adjustment rate limiting. When a slippage risk or vibration index is detected to exceed a preset threshold, suppression adjustment or phased adjustment of the clamping execution parameters is triggered.

[0077] In one specific implementation, the displacement feedback is used to characterize the displacement position of the fixture's actuator and the change in the alignment state of key features during clamping. The displacement feedback can be obtained from an encoder, linear displacement sensor, or position switch of the fixture actuator, and is used to output the actual displacement value, displacement change, or position signal of the fixture's actuator. Alternatively, the alignment deviation of the key features can be obtained by rapidly re-measuring the key features, and is used to characterize the change in the relative nominal position of the key features before and after clamping. The clamping force feedback is used to characterize the clamping force level applied to the workpiece by the fixture's actuator, and can be obtained from a tension / compression force sensor, pressure sensor, or the force estimation module of the fixture driver, and is used to output the clamping force value or the change in clamping force.

[0078] In one specific implementation, the cascade adjustment structure is established as follows: the correction of displacement-type clamping execution parameters is used as the output of the outer adjustment loop, and the correction of force-type clamping execution parameters is used as the output of the inner adjustment loop. The inner adjustment loop provides rapid and stable force feedback support to the correction process of the outer adjustment loop. Specifically, the outer adjustment loop uses the alignment deviation of key features or the change in the clamping error distribution as the adjustment basis: when the alignment deviation of key features is detected as not converging to the target range, or the clamping error distribution shows an increasing trend in the key sub-region, the outer adjustment loop generates a correction amount for the displacement-type clamping execution parameters of the corresponding clamping end, so as to cause the position setting of the clamping end to adjust in the direction of error suppression. The inner adjustment loop uses clamping force feedback as the adjustment basis: when the clamping force does not reach the set range, the clamping force fluctuates too much, or abnormal changes in the clamping force occur, the inner adjustment loop generates a correction amount for the force-type clamping execution parameters of the corresponding clamping end, so as to maintain the clamping force within the range that satisfies clamping stability and does not cause damage. By cascading the outer and inner layers, both geometric alignment and clamping force stability can be achieved during the clamping process.

[0079] In one specific implementation, the input to the outer adjustment loop can be any of the following: First, the alignment deviation of key features, which can be obtained by rapid retesting of key features. Rapid retesting can be performed using a visual measurement device or a depth measurement device in low-density sampling mode to obtain the current geometric position of the key features and compare it with the nominal position to obtain the alignment deviation; Second, the change in the clamping error distribution, which can be obtained by re-predicting some key positions or by rapidly retesting to form an updated error distribution, and comparing it with the error distribution at the previous moment. Based on the above inputs, the outer adjustment loop determines the range of the clamping execution end that needs adjustment and outputs corrections to displacement-type clamping execution parameters, causing the clamping state to move towards reducing error.

[0080] In one specific implementation, the limiting constraints are used to ensure that parameter adjustments during the adaptive correction process meet the mechanism's capabilities and process safety boundaries. The stroke limit for displacement parameters restricts the displacement setpoint or correction amount from exceeding the maximum allowable stroke range of the fixture actuator; the clamping force limit for force parameters restricts the clamping force setpoint or correction amount from exceeding the upper limit of clamping force allowed by the fixture structure and workpiece material; the adjustment amplitude limit restricts the change in a single correction from exceeding a preset upper limit to avoid workpiece slippage or localized stress impact caused by abrupt adjustments; and the adjustment rate limit restricts the adjustment speed per unit time from exceeding a preset upper limit to avoid increased vibration or clamping instability due to excessively rapid adjustments. These limiting constraints can be verified before each output correction amount. When the intended output correction amount does not meet the limiting constraints, the correction amount is truncated or reduced, ensuring that the correction process remains within a safe and controllable range.

[0081] In one specific implementation, the slippage risk and vibration index are used to characterize the stability state of the clamping process. When the stability state is abnormal, suppression adjustment or phased adjustment is triggered. Slippage risk can be determined based on the following information: inconsistency between clamping force changes and displacement changes, sudden drop in clamping force at the clamping contact point, sudden increase in alignment deviation of key features in a short period of time, or detection of small displacement drift of the workpiece relative to the tooling through visual re-measurement. Vibration index can be obtained by an accelerometer, vibration sensor, or driver vibration monitoring module, or it can be characterized by high-frequency fluctuation characteristics of displacement feedback or high-frequency fluctuation characteristics of clamping force feedback. When the slippage risk or vibration index is detected to exceed a preset threshold, suppression adjustment or phased adjustment of the clamping execution parameters is triggered: suppression adjustment can be manifested as reducing the displacement correction amount, reducing the clamping force correction rate, or suspending the adjustment output of the outer adjustment ring; phased adjustment can be manifested as breaking down a large adjustment into multiple small adjustments, and performing feedback detection after each stage adjustment to confirm that stability has been restored before entering the next stage.

[0082] Through the aforementioned cascade adjustment structure, amplitude limiting constraint, and stability triggering mechanism, adaptive correction of displacement and force clamping execution parameters can be achieved based on sensor feedback during clamping execution, keeping the alignment state of key features and the distribution of clamping errors within a controllable range, and reducing the risk of clamping instability caused by slippage and vibration.

[0083] After the clamping is completed according to the clamping strategy and the welding operation begins at the fixture execution end, considering that heat input, welding sequence, structural stiffness differences, and changes in local constraint conditions may cause geometric drift of the parts to be welded during the welding process, this embodiment further monitors key features during the welding process and dynamically updates the clamping error distribution and clamping execution parameters when trigger conditions are met. Specifically, during the welding process, lightweight measurements are performed on key features according to preset welding nodes or fixed cycles to obtain welding drift deviations, and drift regions are determined based on the welding drift deviations; when the welding drift deviation exceeds a preset drift threshold or shows a continuous increasing trend, the clamping error distribution is updated based on the drift region, and the clamping execution parameters of some fixtures corresponding to the drift region are adjusted; when the measurement fluctuation of the welding drift deviation exceeds a preset threshold or the prediction uncertainty of the drift region exceeds a threshold, high-precision measurements are performed on the drift region to obtain remeasurement deviations, and the clamping error distribution is updated using the remeasurement deviations before the clamping execution parameters are adjusted.

[0084] In one specific implementation, the preset welding nodes can be set according to the welding process, such as before arc initiation, after completing a certain weld segment, after completing a corner segment or arc termination segment, or after completing several welding points; the fixed cycle can be set according to the production rhythm, such as triggering a measurement at a fixed time interval or at a fixed welding length interval. By triggering measurements at nodes or cycles during the welding process, geometric changes caused by thermal deformation or stress release can be captured in a timely manner.

[0085] In one specific implementation, the lightweight measurement of key features during the welding process to obtain welding drift deviation can employ the same measurement device and processing flow as the aforementioned lightweight measurement. However, the measurement location can preferentially select key features sensitive to welding drift, including the weld start and end point neighborhood, weld neighborhood edge lines or contour lines, locating hole centers or locating pin mating features, and reference surfaces or reference boundary features near free edges. The output of the lightweight measurement is the current geometric state of the key features, which is further compared with the geometric state before welding or the previous measurement node to obtain the welding drift deviation. The welding drift deviation can be expressed as a deviation amount such as key point position change, key boundary line offset, gap or misalignment change, used to characterize the degree of drift of the component relative to the fixture reference during the welding process.

[0086] In one specific implementation, determining the drift region based on welding drift deviation can be achieved as follows: The welding drift deviation is associated with its spatial location within the evaluation region, and a localized area where the drift deviation is significantly concentrated is identified as the drift region. The drift region can be one or more sub-regions within the evaluation region, or it can be a local neighborhood surrounding the weld path. When the drift deviation mainly occurs near the weld start and end points, the corresponding neighborhood can be identified as the drift region; when the drift deviation mainly occurs near the thin-walled free edge, the area corresponding to the edge-sensitive region can be identified as the drift region. By determining the drift region, subsequent updates and adjustments can be focused on the affected area, reducing unnecessary global adjustments.

[0087] In one specific implementation, an update and adjustment are triggered when the welding drift deviation exceeds a preset drift threshold or shows a continuous increasing trend. The preset drift threshold can be set based on assembly tolerance requirements, welding quality requirements, or tooling experience data to limit the acceptable drift range; a continuous increasing trend can be determined by the consistent direction of drift deviation changes and the cumulative increase in amplitude across multiple consecutive measurement nodes. After triggering, the clamping error distribution can be updated based on the drift region in the following ways: the drift deviation measured during welding is added as a new deviation sample to the deviation sample set, so that the clamping error distribution can reflect the dynamic deviation during welding; or spatial prediction and prediction uncertainty calculation are re-performed only within the drift region to obtain a locally updated error distribution for the drift region, and this locally updated result is merged into the overall clamping error distribution. Through the above updates, the clamping error distribution is dynamically corrected as the welding process changes.

[0088] In one specific implementation, the adjustment of the clamping execution parameters of the fixtures corresponding to the drift region can be determined according to the range of action of the fixtures: First, based on the fixture arrangement and contact point distribution, the set of fixture execution ends that exert the main constraint on the drift region is determined; then, only the clamping execution parameters of this set of fixture execution ends are adjusted, without needing to simultaneously adjust the fixture execution ends unrelated to the drift region. The adjustment may include correction of displacement-type clamping execution parameters and correction of force-type clamping execution parameters. The adjustment basis may be the error change direction of the drift region and the changing trend of the alignment deviation of key features, so that the geometry of the drift region converges to an acceptable range.

[0089] In one specific implementation, when the measurement fluctuation of welding drift deviation exceeds a preset threshold or the prediction uncertainty of the drift region exceeds a threshold, to avoid clamping instability caused by direct adjustments based on unreliable measurements or predictions, this embodiment further performs high-precision measurements on the drift region. Measurement fluctuation can be understood as a large change in drift deviation obtained from repeated measurements of the same key feature within a short period of time, or an unstable jump in the measurement output; prediction uncertainty exceeding the threshold indicates insufficient reliability of deviation prediction for unmeasured positions within the drift region. At this time, performing high-precision measurements on the drift region can employ higher resolution imaging, denser point cloud sampling, or more rigorous calibration measurement methods to obtain highly reliable measured results of key features within the drift region, and thereby obtain the remeasurement deviation.

[0090] In one specific implementation, the step of updating the clamping error distribution using the remeasured deviation before adjusting the clamping execution parameters can be carried out as follows: the remeasured deviation is incorporated into the deviation sample set, and the spatial prediction and prediction uncertainty calculation for the drift region are re-executed to obtain the clamping error distribution after remeasurement correction; after obtaining the corrected clamping error distribution, the clamping execution parameter adjustment amount is output to the clamping execution end corresponding to the drift region, so that the adjustment is based on a more reliable data basis, thereby reducing the risk of misadjustment and improving the clamping stability of the welding process.

[0091] Through the above-mentioned welding process drift monitoring, drift area determination, error distribution update and local fixture adjustment mechanism, the geometric drift caused by thermal deformation can be controlled during the welding process, so that the clamping state is dynamically maintained within a controllable range during the welding process, thereby improving the consistency of welding assembly quality.

[0092] In one embodiment, the present invention provides a clamping control system for a welding station, suitable for applications requiring comprehensive control over the clamping and positioning, clamping stability, and drift suppression during the welding process of the workpiece to be welded. The workpiece to be welded can be a bracket, connecting plate, shell-type component, etc. The welding station includes a clamp base, multiple independently driveable clamping actuators, and welding equipment; the clamping actuators can be driven by electric cylinders, pneumatic cylinders, or servo actuators to apply clamping action to the workpiece to be welded, thereby suppressing clamping deviations and maintaining welding assembly quality.

[0093] To ensure consistent expression of measurement results and fixture control, this embodiment establishes a unified coordinate system: a fixture coordinate system is established based on the fixture base or station reference, and a nominal coordinate system is established based on the nominal model of the part to be welded or the tooling positioning reference. The measuring device obtains the external parameter relationship between the measuring coordinate system and the fixture coordinate system through calibration, enabling the measurement results to be converted to the fixture coordinate system for setting and adjusting the parameters of the fixture actuator.

[0094] This system can be configured with the following hardware and software support objects: a measuring device for lightweight or high-precision measurement of key features, which may include at least one of an industrial camera and light source, a depth measuring device, a laser displacement sensor, or a laser scanning device; a fixture status sensor for collecting status information such as displacement and clamping force of the fixture actuator, which may include at least one of an encoder, a linear displacement sensor, a force sensor, or a pressure sensor; a control and computing device for performing spatial prediction, uncertainty assessment, clamping strategy solution, and adaptive correction logic, which may be an industrial control computer, a programmable controller, or an edge computing unit; an actuator and driver for driving the fixture actuator to move according to control commands; and a communication interface for transmitting data and control commands between the measuring device, the control and computing device, the driver, and the sensor.

[0095] In the above-described implementation environment, the clamping control system of this embodiment is specifically a clamping control system for automotive parts welding fixtures, such as... Figure 2 As shown, the system includes: The deviation sample generation unit is used to perform lightweight measurement on the key features of the component to be welded to obtain deviation samples. The lightweight measurement is to perform low-density sampling on a limited number of measurement locations of the key features within a predetermined measurement time. The error distribution construction unit is used to obtain the predicted value of the deviation at the unmeasured position and the corresponding prediction uncertainty by using the deviation sample prediction, and to statistically form the clamping error distribution of the preset evaluation area and the prediction uncertainty of each sub-region. The retest deviation generation unit is used to determine when the prediction uncertainty of a sub-region exceeds a threshold, and to perform high-precision measurement on the sub-region to obtain the retest deviation. The high-precision measurement has a higher sampling density and sampling resolution than the lightweight measurement. The update unit is used to update the clamping error distribution based on the retest deviation; The clamping strategy generation unit is used to collaboratively set the clamping execution parameters of each clamp based on the updated clamping error distribution, and generate a clamping strategy. The fixture drive unit is used to control and drive each fixture to perform clamping actions according to the clamping strategy; The adaptive correction unit is used to adaptively correct the clamping execution parameters based on sensor feedback.

[0096] In one specific implementation, the deviation sample generation unit is communicatively connected to the measuring device and is used to perform lightweight measurement on key features before clamping begins or at a predetermined measurement time. Key features are reference geometric features used to characterize the clamping and positioning state and welding assembly quality of the parts to be welded, including at least one of the following: positioning hole center or positioning pin mating features, reference surface or reference boundary features, weld start and end point features, and edge line or contour line features of the weld neighborhood. Lightweight measurement outputs sparse measurement results in a short time: when using a vision measuring device, key feature images are acquired and corresponding geometric feature information is extracted; when using a laser displacement sensor, displacement data related to weld neighborhood height, gap, or misalignment are acquired; when using a depth measuring device, the three-dimensional coordinates or local contour information of key points are output in a low-density sampling mode. The deviation sample includes at least a measurement position identifier, measured geometric quantities of the key features, and a deviation value calculated from the measured value and the nominal value. The deviation value can be one or more of spatial position deviation, distance deviation, gap deviation, or misalignment deviation.

[0097] In one specific implementation, the error distribution construction unit is used to discretize a preset evaluation area and perform spatial prediction on the discretized unmeasured locations based on deviation samples, outputting the deviation prediction value and the corresponding prediction uncertainty. The preset evaluation area may cover the weld neighborhood, the positioning hole neighborhood, or the thin-walled easily deformable area, etc., and the unmeasured locations may be grid points or feature sampling points. The error distribution construction unit can establish a spatial correlation model based on the spatial distance between measurement locations and the deviation difference, and perform spatial interpolation prediction on the unmeasured locations according to the spatial correlation model to obtain the deviation prediction value of the unmeasured locations; at the same time, the prediction uncertainty is output or statistically obtained based on the spatial correlation model to characterize the reliability of the prediction result. The error distribution construction unit further divides the evaluation area into sub-regions and performs statistical operations on the prediction uncertainty of the unmeasured locations in the sub-regions to obtain the prediction uncertainty of each sub-region, so as to reflect the overall prediction reliability of the sub-region.

[0098] In one specific implementation, the retest deviation generation unit is used to determine the prediction uncertainty and threshold of each sub-region. When the prediction uncertainty of a sub-region exceeds the threshold, the sub-region is identified as a retest sub-region, and the measuring device is triggered to enter the high-precision measurement mode. High-precision measurement has higher precision or higher resolution than lightweight measurement, and retest data can be obtained through higher resolution imaging, denser point cloud sampling, or more rigorous calibration measurement methods. Within the retest sub-region, the retest deviation generation unit can sort the unmeasured positions from high to low prediction uncertainty, select at least one of the top-ranked retest positions to perform high-precision measurement, and output the retest deviation corresponding to the retest position. The retest position includes at least one of the following: the unmeasured position with the largest prediction uncertainty, a representative position in the set of unmeasured positions with prediction uncertainty exceeding a preset threshold, or an unmeasured position located in the vicinity of the weld start and end points or in the edge sensitive area.

[0099] In one specific implementation, the updating unit incorporates the remeasured deviation into the deviation sample set and triggers the error distribution construction unit to re-execute spatial prediction and uncertainty calculation based on the updated samples, thereby updating the clamping error distribution and the predicted uncertainty of each sub-region. The update can be performed on the entire evaluation area, or it can be performed preferentially in sub-regions where drift or uncertainty is mainly concentrated, and the local update results are merged into the overall clamping error distribution, making the error distribution closer to the actual clamping state.

[0100] In one specific implementation, the clamping strategy generation unit generates a clamping strategy based on the updated clamping error distribution and outputs the clamping execution parameter settings for each clamping actuator. The clamping execution parameters include at least displacement-type or force-type setting values. Under certain constraints, the clamping strategy generation unit aims to reduce the clamping error index of a preset evaluation area or its key sub-regions and suppresses the fluctuation of the clamping error index to make the clamping error distribution more spatially uniform. The constraints include at least one of the following: clamp travel limit, clamping force limit, crushing risk threshold, and slippage risk threshold. Further, the clamping strategy generation unit determines the overall clamping correction amount for error suppression based on the updated clamping error distribution, and under certain clamping coordination constraints, decomposes the overall clamping correction amount into the clamping execution parameter increments corresponding to each clamp, enabling multiple clamping actuators to work together to achieve the error suppression target. The coordination constraints include at least one of the following: relative displacement difference between clamps does not exceed a threshold, relative clamping force difference between clamps does not exceed a threshold, and consistency constraints on the clamping action applied to the same sub-region.

[0101] In one specific implementation, the fixture driving unit is communicatively connected to the driver of each fixture execution end. It sends the clamping execution parameters corresponding to the clamping strategy to each fixture execution end and controls each fixture execution end to perform clamping actions according to the set parameters, thereby completing the clamping, positioning, and stabilizing of the parts to be welded. The fixture driving unit can support staged clamping control to ensure a smooth clamping process and reduce slippage and vibration caused by impact.

[0102] In one specific implementation, the adaptive correction unit is used to perform online correction of the clamping execution parameters based on sensor feedback during the clamping process. The sensor feedback includes at least displacement feedback and clamping force feedback: displacement feedback can come from an encoder or a linear displacement sensor, reflecting the actual displacement state of the clamping end; clamping force feedback can come from a force sensor or a pressure sensor, reflecting the clamping force level and its fluctuations. The adaptive correction unit can establish a cascaded adjustment structure of an outer layer adjustment based on displacement feedback and an inner layer adjustment based on clamping force feedback: the outer layer adjustment corrects displacement-related clamping execution parameters based on the alignment deviation of key features or the change in the distribution of clamping errors; the inner layer adjustment corrects force-related clamping execution parameters based on clamping force feedback; and during the correction process, limiting constraints are set on the clamping execution parameters, including at least one of the following: travel limit for displacement-related parameters, clamping force limit for force-related parameters, adjustment amplitude limit, or adjustment rate limit; when a slippage risk or vibration index exceeds a preset threshold, suppression adjustment or staged adjustment of the clamping execution parameters is triggered to reduce the risk of clamping instability and improve clamping consistency.

[0103] Through the above system implementation method, this system can realize a closed-loop process in the welding station, which includes lightweight measurement, spatial prediction and uncertainty assessment, local high-precision re-measurement and correction, collaborative generation of clamping strategy, clamping drive execution, and adaptive correction based on sensor feedback. This allows the clamping error distribution to be continuously updated and used to guide the clamping execution end to cooperate in clamping, thereby improving the stability of the clamping and positioning of the parts to be welded and the consistency of the welding assembly quality.

[0104] This embodiment also provides a computer-readable storage medium storing a computer program. When the computer program is run on an electronic device, it causes the electronic device to execute the aforementioned automotive parts welding fixture clamping control method. The computer-readable storage medium may be a read-only memory, a random access memory, a magnetic storage medium, an optical storage medium, or any combination thereof, and is not limited thereto.

[0105] This embodiment also provides a computer program product, which includes computer program code. When the computer program code runs on an electronic device, it causes the electronic device to execute the aforementioned automotive parts welding fixture clamping control method. The computer program code can be loaded into control and computing devices such as industrial control computers, programmable logic controllers, or edge computing units to achieve clamping control of the welding station fixture execution end.

[0106] This embodiment also provides a chip, which includes a processor for calling and running a computer program from a memory, causing an electronic device equipped with the chip to execute the aforementioned automotive parts welding fixture clamping control method. In one possible implementation, the chip may be integrated into a workstation controller, a fixture drive controller, or a measurement and control integrated device to realize at least some of the functions such as deviation sample generation, error distribution construction, retest triggering and updating, clamping strategy generation, and adaptive correction.

[0107] Based on the above description of the embodiments, those skilled in the art will understand that, for ease of description and explanation, the division of functional units in the foregoing system embodiments is only a logical functional division. In specific implementation, the above functions can be allocated to different functional modules according to the needs of hardware resources, workstation layout, and control architecture. In other words, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0108] It should be understood that the device structure and process disclosed in the above embodiments are not limited to the described form, and can be implemented in other ways without departing from the inventive concept. For example, the division of modules or units is only an exemplary functional division, and other division methods can be adopted in actual implementation. Multiple modules or units can be merged into one module or unit, or some modules or units can be integrated into another device. In addition, some non-essential functions can be selectively ignored or not executed according to actual working conditions. The connection relationship between modules or units is not limited to direct connection or direct communication. Indirect coupling or communication connection between devices or units can be achieved through interfaces. The connection form can be electrical connection, mechanical connection or other forms.

[0109] The modules described as functional units may or may not be physically separate; the components shown as modules may be a single physical unit or multiple physical units, and may be centrally located in the same control device or distributed across different devices or locations, such as in measuring devices, workstation controllers, fixture drivers, and sensor acquisition modules. Some or all of the functional units can be selected to achieve the purpose of this invention according to actual needs.

[0110] Furthermore, the aforementioned functional units can be integrated into one processing unit, or implemented separately in multiple processing units, or partially integrated while partially independent. The integrated unit can be implemented in hardware or as a software functional unit; when implemented as a software functional unit and used or sold as an independent product, it can be stored in a computer-readable storage medium.

[0111] Based on the above understanding, all or part of the technical solution of the present invention can be implemented in the form of a software product. The software product is stored in a computer-readable storage medium and contains several instructions for causing an electronic device or processor to execute all or part of the steps of the method embodiments of the present invention. The storage medium includes, but is not limited to, various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory, random access memory, magnetic disks, or optical disks.

[0112] It should be noted that all or part of the technical features of the various embodiments provided by the present invention can be combined or used in combination without contradiction.

[0113] The above description is merely an example of specific embodiments of the present invention and is not intended to limit the scope of protection of the present invention. Any equivalent changes, substitutions, or improvements made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the concept of the present invention, should be covered within the scope of protection of the present invention. The scope of protection of the present invention should be determined by the scope defined in the claims.

Claims

1. A method for clamping and controlling a welding fixture for automotive parts, characterized in that, The method includes: The key features of the components to be welded are subjected to lightweight measurement to obtain deviation samples. The lightweight measurement is a low-density sampling of a limited number of measurement locations of the key features within a predetermined measurement time. By using deviation samples to predict the deviation values ​​and corresponding prediction uncertainties of unmeasured locations, the clamping error distribution of the preset evaluation area and the prediction uncertainties of each sub-region are statistically formed. When the prediction uncertainty of a sub-region exceeds a threshold, a high-precision measurement is performed on the sub-region to obtain the remeasurement deviation and update the clamping error distribution. The high-precision measurement has a higher sampling density and sampling resolution than the lightweight measurement. Based on the updated clamping error distribution, the clamping execution parameters of each fixture are collaboratively set to generate a clamping strategy; The control drives each fixture to perform clamping actions according to the clamping strategy, and adaptively corrects the clamping execution parameters based on sensor feedback. This includes: establishing a cascade adjustment structure of an outer adjustment ring based on displacement feedback and an inner adjustment ring based on clamping force feedback. The outer adjustment ring is used to correct the displacement-type clamping execution parameters of each fixture based on the alignment deviation of key features or the change in the clamping error distribution. The inner adjustment ring is used to correct the force-type clamping execution parameters of each fixture based on clamping force feedback. In the adaptive correction process, a limiting constraint is set on the clamping execution parameters. The limiting constraint includes at least one of the following: stroke limiting for displacement parameters, clamping force limiting for force parameters, adjustment amplitude or adjustment rate limiting. Furthermore, when a slippage risk or vibration index is detected to exceed a preset threshold, the clamping execution parameters are suppressed or adjusted in stages.

2. The method for clamping and controlling automotive parts welding fixtures according to claim 1, characterized in that, The key features are reference features used to characterize the clamping and positioning state of the parts to be welded and the quality of the welding assembly. The reference features include at least one of the following: the center of the positioning hole or the mating feature of the positioning pin, the reference surface or the reference boundary feature, the start and end point feature of the weld, and the edge line or contour line feature of the weld neighborhood. The lightweight measurement is used to obtain deviation samples of the reference features, and the deviation samples include at least one of the following: spatial position deviation of the reference features, distance deviation between different reference features, gap deviation, or misalignment deviation.

3. The method for clamping and controlling automotive parts welding fixtures according to claim 1 or 2, characterized in that, The method of using deviation samples to predict the deviation prediction value and corresponding prediction uncertainty of unmeasured locations includes: constructing a probabilistic spatial correlation model based on the spatial distance and deviation difference between each measurement location in the deviation sample; performing spatial interpolation prediction on unmeasured locations within a preset evaluation area based on the probabilistic spatial correlation model to obtain the deviation prediction value of the unmeasured locations; and outputting the prediction variance or confidence interval width of the unmeasured locations based on the probabilistic spatial correlation model as the prediction uncertainty corresponding to the unmeasured locations. The probabilistic spatial correlation model establishes a correlation characterization function with the constraint that the correlation decreases with increasing spatial distance, and determines the model parameters by fitting biased samples. The model parameters include at least one of the following: correlation length parameter, noise term parameter, and amplitude parameter.

4. The method for clamping and controlling automotive parts welding fixtures according to claim 1, characterized in that, The statistical formation of the prediction uncertainty of each sub-region of the preset evaluation area includes: performing statistical operations on the prediction uncertainty corresponding to the unmeasured position in each sub-region to obtain the prediction uncertainty of the sub-region; The statistical operation includes at least one of the following: taking the maximum value, mean value, or weighted mean of the prediction uncertainty within the sub-region, or calculating the proportion of unmeasured locations where the prediction uncertainty exceeds a preset uncertainty threshold.

5. The method for clamping and controlling automotive parts welding fixtures according to claim 1, characterized in that, The process of performing high-precision measurement on the sub-region includes: sorting the sub-regions according to the prediction uncertainty from high to low, and selecting at least one of the top-ranked retest locations to perform high-precision measurement; The retest position includes at least one of the following: the untested position with the largest prediction uncertainty, the representative position in the set of untested positions with prediction uncertainty exceeding a preset threshold, and the untested position located in the neighborhood of the weld start and end points or the edge sensitive area; and the clamping error distribution is corrected and updated using the retest deviation corresponding to the retest position.

6. The method for clamping and controlling automotive parts welding fixtures according to claim 1, characterized in that, The generation process of the clamping strategy includes: under the condition of satisfying the first constraint, taking the minimization of the clamping error index of the preset evaluation area or its key sub-region as the optimization objective, and at the same time suppressing the fluctuation of the clamping error index, so as to obtain a clamping strategy with uniform clamping error distribution. The clamping error index includes at least one of the maximum value, mean, or weighted mean of the predicted deviation value within the region, and the volatility includes at least one of the variance, standard deviation, or range of the predicted deviation value within the region. The first constraint includes at least one of the upper limit of the stroke, the upper limit of the clamping force, and the pressure injury risk threshold or the slippage risk threshold of the clamping execution parameters.

7. The method for clamping and controlling automotive parts welding fixtures according to claim 1, characterized in that, The collaborative setting of the clamping execution parameters of each fixture includes: determining the overall clamping correction amount for error suppression based on the updated clamping error distribution, and decomposing the overall clamping correction amount into the clamping execution parameter increments corresponding to each fixture under the second constraint condition; The overall clamping correction includes at least the target correction for the predicted deviation value within the region, the pose target correction for the key features within the region, and the target correction for gaps or misalignments within the region. The decomposition process is obtained by solving a weighted optimization problem, the optimization objective of which includes at least minimizing the weighted norm of the increment of each fixture clamping execution parameter and satisfying the error suppression objective. The second constraint condition includes at least one of the following: the relative displacement difference between fixtures does not exceed a threshold, the relative clamping force difference between fixtures does not exceed a threshold, and the consistency constraint of the clamping action applied to the same sub-region.

8. The method for clamping and controlling automotive parts welding fixtures according to claim 1, characterized in that, Also includes: During the welding process, lightweight measurements are performed on key features according to preset welding nodes or fixed cycles to obtain welding drift deviation, and the drift area is determined based on the welding drift deviation; When the welding drift deviation exceeds the preset drift threshold or shows a continuous increasing trend, the clamping error distribution is updated based on the drift area, and the clamping execution parameters of some fixtures corresponding to the drift area are adjusted. When the measurement fluctuation of welding drift deviation exceeds the preset threshold or the prediction uncertainty of the drift area exceeds the threshold, high-precision measurement is performed on the drift area to obtain the remeasured deviation, and the clamping error distribution is updated using the remeasured deviation before the clamping execution parameters are adjusted.

9. A clamping control system for welding fixtures for automotive parts, characterized in that, The system includes: The deviation sample generation unit is used to perform lightweight measurement on the key features of the component to be welded to obtain deviation samples. The lightweight measurement is to perform low-density sampling on a limited number of measurement locations of the key features within a predetermined measurement time. The error distribution construction unit is used to obtain the predicted value of the deviation at the unmeasured position and the corresponding prediction uncertainty by using the deviation sample prediction, and to statistically form the clamping error distribution of the preset evaluation area and the prediction uncertainty of each sub-region. The retest deviation generation unit is used to determine when the prediction uncertainty of a sub-region exceeds a threshold, and to perform high-precision measurement on the sub-region to obtain the retest deviation. The high-precision measurement has a higher sampling density and sampling resolution than the lightweight measurement. The update unit is used to update the clamping error distribution based on the retest deviation; The clamping strategy generation unit is used to collaboratively set the clamping execution parameters of each clamp based on the updated clamping error distribution, and generate a clamping strategy. The fixture drive unit is used to control and drive each fixture to perform clamping actions according to the clamping strategy; An adaptive correction unit is used to adaptively correct the clamping execution parameters based on sensor feedback. This includes: establishing a cascaded adjustment structure of an outer adjustment ring based on displacement feedback and an inner adjustment ring based on clamping force feedback. The outer adjustment ring is used to correct the displacement-type clamping execution parameters of each fixture based on the alignment deviation of key features or the change in the distribution of clamping error. The inner adjustment ring is used to correct the force-type clamping execution parameters of each fixture based on the clamping force feedback. In the adaptive correction process, a limiting constraint is set on the clamping execution parameters. The limiting constraint includes at least one of the following: stroke limiting for displacement parameters, clamping force limiting for force parameters, adjustment amplitude or adjustment rate limiting. Furthermore, when a slippage risk or vibration index is detected to exceed a preset threshold, the clamping execution parameters are suppressed or adjusted in stages.

Images