Laser detection method for tarpaulin heat sealing joint deformation

Abstract

The application discloses a laser detection method for tarpaulin heat sealing joint deformation, which comprises the following steps: clamping a tarpaulin sample with a heat sealing joint in a loading mechanism, scanning along the joint to obtain a three-dimensional contour point cloud in a reference, loading and unloading state; completing six-degree-of-freedom rigid body correction by using a clamp reference target, fitting and deducting a slowly-varying relaxation displacement field only according to two side cloth reference belts, and then calculating a relative warping amount, a local elongation amount, a deformation gradient and a residual deformation amount according to a material station position; determining a threshold value and merging a continuous section to output the position, length and dominant abnormal type of a weak area. The line laser profilometer can also separate a thickness and a middle surface warping channel, and identify short-time recovery and continuous residual deformation. The application can realize continuous, layered and quantitative detection of the heat sealing joint along a length direction.

Description

Technical Field

[0001] This invention relates to the field of tarpaulin seam quality inspection and laser three-dimensional measurement, and particularly to a laser detection method for deformation of tarpaulin heat-sealed seams. Background Technology

[0002] The heat-sealed seam of the tarpaulin is formed by overlapping adjacent tarpaulin base fabrics and then heating and pressing them together. Its forming state is affected by factors such as heat-sealing temperature, pressing pressure, overlap width, and base fabric tension. Process fluctuations may cause local warping, shrinkage, abrupt changes in thickness, and uneven elongation under stress at the seam, which may further lead to weak points such as cracking, water seepage, or tearing.

[0003] Current quality inspection methods largely rely on manual observation, ruler and gauge measurements, or single-point tensile tests. Manual observation struggles to quantify minute deformations, while single-point tests fail to reflect the continuous state of the seam along its length. When comparing the contours before and after loading, the overall movement of the tarpaulin, changes in clamping posture, and slack in the base fabric can overlap with the actual local deformation of the seam, resulting in a lack of stable correlation between data from different locations and at different times. Unilateral height measurements also struggle to distinguish between thickness changes and mid-surface warping, and short-term viscoelastic recovery after unloading can interfere with the evaluation of residual deformation.

[0004] Therefore, a laser detection method for tarpaulin heat-sealed seam deformation is needed to address the shortcomings of existing technologies. Summary of the Invention

[0005] The purpose of this invention is to provide a laser detection method for the deformation of heat-sealed seams in tarpaulins, addressing the problems of existing detection methods that cannot continuously characterize seam deformation, cause measurement overlap due to overall movement and base fabric relaxation, and make it difficult to distinguish between thickness changes, mid-surface warping, and unloading recovery. This invention acquires a three-dimensional contour point cloud along the seam under controlled loading. Through rigid body registration with a fixture target, relaxation field compensation with double-sided base fabric reference strips, and material station correspondence, a sequence of seam deformation indicators is formed. Continuous abnormal regions are identified and merged based on the statistical fluctuations and quality limits of the base fabric. Furthermore, opposing synchronous scanning and multi-time recovery sequences are employed to evaluate thickness, mid-surface warping, short-term recoverable items, and residual items.

[0006] This invention provides a laser detection method for deformation of heat-sealed seams in tarpaulins, comprising: S1. The tarpaulin sample with heat-sealed seam is clamped in the loading mechanism with load sensor. In the reference state, at least one detection load state and the unloaded state, the line laser profilometer spanning the seam area and the base fabric reference strips on both sides is used to scan along the longitudinal direction of the seam to obtain the three-dimensional profile point cloud associated with the corresponding detection state. S2. Using the fixture reference target outside the deformation zone, obtain the six-degree-of-freedom rigid body transformation of each detection state relative to the reference state, and obtain the registration point cloud unified to the reference fixture coordinate system. S3. Divide the registration point cloud into a seam area and two base fabric reference strips on both sides. Determine the displacement of the seam area and the two base fabric reference strips on both sides based on the registration point cloud of each detection state and the reference state. Fit the base fabric relaxation displacement field that varies along the longitudinal direction of the seam only based on the displacement of the two base fabric reference strips on both sides. Interpolate the base fabric relaxation displacement field to the seam area and deduct it from the displacement of the seam area to obtain the seam compensation displacement. S4. Based on the cross-sectional profile characteristics of the reference state, establish the material station correspondence between states under longitudinal monotonic constraints. According to the material station correspondence and the joint compensation displacement, calculate the relative warping, local elongation, deformation gradient and unloading residual deformation of each joint detection segment to obtain the joint deformation index sequence. S5. Compare the seam deformation index sequence with the base fabric statistical fluctuation threshold and quality limit, merge the continuously exceeding seam detection segments, and obtain and output the start and end positions, length and dominant anomaly type of the seam exceeding the limit area.

[0007] Optionally, S1 includes: taking the leveling preload state, in which the tarpaulin sample eliminates visible slack and the load sensor reading is stable, as the reference state, and sequentially setting at least one detection load state and at least one specified evaluation time state after unloading back to the leveling preload. The load sensor collects the load values ​​for each state, and uses the load values, scan trigger time, and state identifier as data recording fields for the corresponding three-dimensional contour point cloud. Furthermore, opposing line laser profilometers are set on the same measuring frame, located on the upper and lower sides of the tarpaulin sample and triggered synchronously, to obtain the three-dimensional profile point cloud of the upper surface and the three-dimensional profile point cloud of the lower surface, respectively. The normal distance between the upper and lower surfaces at the same material station is used as the input for the thickness channel, and the median of the normal coordinates of the upper and lower surfaces is used as the input for the mid-surface warping channel. The synchronization deviation limit is determined based on the time difference distribution of repeated synchronous triggering of the opposing line laser profilometer. When the synchronous triggering deviation exceeds the synchronization deviation limit, the scan is re-scanned. If the synchronous deviation limit is still exceeded after the re-scan, the corresponding detection state is marked as dual-channel invalid. The calculation of the thickness channel and mid-surface warping channel in steps S3 to S5 for this detection state is stopped, and a dual-channel retest record containing the detection state, synchronization triggering deviation, and dual-channel retest mark is independently output.

[0008] Optionally, S2 includes: setting at least three non-collinear fixture reference targets on the non-deformable component of the loading mechanism, and determining the six-degree-of-freedom rigid body transformation based on the three-dimensional coordinates of each fixture reference target in the reference state and the current detection state; The coordinate residuals of each fixture reference target after transformation are used as the registration verification quantity. The registration error limit is determined based on the calibration uncertainty of the line laser profilometer and the distribution of coordinate residuals from repeated measurements of the fixture reference targets. When the registration verification quantity exceeds the registration error limit, the corresponding state is marked as a registration anomaly and the calculation of subsequent deformation indices for that state is stopped.

[0009] Optionally, S3 includes: extracting displacement samples relative to the reference state from the base fabric reference strips on both sides of the seam area, calculating the median of the displacement samples and the median of the absolute deviation of each displacement sample from the median, and identifying and removing displacement samples whose absolute deviation from the median exceeds a preset multiple as outliers. The preset multiple is determined based on the displacement noise distribution obtained by repeated scanning of the seamless base fabric. The retained displacement samples are fitted with displacement components that have a longitudinal wavelength not less than the preset lower limit of relaxation wavelength, and interpolation is performed according to the distance from the reference strips of the base fabric on both sides to the lateral boundary of the seam area. The preset lower limit of relaxation wavelength is determined based on the displacement power spectrum of the seamless base fabric under the same loading state, and the points in the seam area do not participate in the fitting of the displacement components. Furthermore, for detection states not marked as dual-channel invalid, fixture reference target rigid body correction and relaxation displacement field compensation of the base fabric reference strips on both sides are performed for the thickness channel and the mid-surface warping channel, respectively. The change in the distance between the upper and lower surfaces of the joint area relative to the reference state is determined as the thickness change. The normal displacement of the midpoint of the upper and lower surfaces of the joint area relative to the fitted mid-surface of the base fabric reference strips on both sides is determined as the mid-surface warping. The thickness change gradient and mid-surface warping gradient are calculated along the longitudinal direction of the joint, respectively. For detection states marked as invalid for dual channels, the rigid body correction, relaxation displacement field compensation, and calculations of the thickness change, mid-surface warping, and corresponding gradients in this step are not performed, and the dual-channel retest record is maintained.

[0010] Optionally, S4 includes: extracting cross-sectional profile features along the longitudinal direction of the reference state joint at a preset interval, and combining the cross-sectional profile features with the longitudinal position to form a material station identifier; Under the constraint of keeping the material station order unchanged, the material station correspondence is determined with the goal of minimizing the sum of the differences in cross-sectional contour features between different detection states and the changes in the distance between adjacent material stations. When there are multiple material station correspondence results with the same objective function value, the corresponding material station is marked as an invalid match. The invalid material station and the detection segment along the seam with the material station as the endpoint are marked as invalid station segments. The calculation of the relative warp, local elongation, deformation gradient and unloading residual deformation of the invalid station segments is stopped, and the material station retest flag is output. For material stations not marked as invalid matches and their adjacent valid material stations, the relative warping is determined by the normal displacement of the joint mid-surface relative to the two sides of the base fabric fitting surface at the same material station, and the local elongation is determined by the state change of the distance between adjacent valid material stations. The difference between the relative warping and the local elongation is calculated along the longitudinal direction of the joint to obtain the relative warping gradient and the local elongation gradient, and the relative warping gradient and the local elongation gradient are used as the deformation gradient. The first and last valid material stations are unilaterally differentiated. The difference between the joint compensation displacement at the specified evaluation time after unloading and the reference state is determined as the unloading residual deformation. Furthermore, for detection states not marked as dual-channel invalid, the scan is repeated at multiple preset recovery times after unloading back to the leveling preload, and recovery sequences for the thickness channel and mid-surface warping channel are formed according to the material station position. The thickness change and mid-surface warpage are accumulated between each adjacent preset recovery time after the first preset recovery time after unloading and the specified evaluation time to obtain the short-term recoverable thickness item and the short-term recoverable mid-surface warpage item. The thickness residual term is obtained by subtracting the thickness short-term recoverable term from the difference between the thickness change at the first preset recovery time after unloading and the reference state, and the mid-surface warping residual term is obtained by subtracting the mid-surface warping short-term recoverable term from the difference between the mid-surface warping at the first preset recovery time after unloading and the reference state. The coordinate residuals of the fixture reference target after six-degree-of-freedom rigid body transformation at each recovery time are used as the recovery registration verification quantity. The recovery registration error limit is determined based on the calibration uncertainty of the line laser profilometer and the distribution of coordinate residuals from repeated measurements of the fixture reference target. When the recovery registration verification quantity at any preset recovery time exceeds the recovery registration error limit, the corresponding recovery sequence is marked as an abnormal recovery sequence. The calculation of the short-term recoverable thickness item, the short-term recoverable mid-surface warpage item, the residual thickness item, and the residual mid-surface warpage item of the recovery sequence is stopped, and a retest flag is output. For detection states marked as invalid for dual channels, recovery sequences for the thickness channel and the mid-surface warping channel are not formed, and the dual-channel retest record is maintained.

[0011] Optionally, S5 includes: setting statistical fluctuation thresholds for the base fabric obtained from the statistical data of the absolute values ​​of the corresponding indicators in the base fabric reference strips on both sides, and positive quality limits given by the product inspection specifications for the relative warp, local elongation, relative warp gradient, local elongation gradient and unloading residual deformation, respectively. The seam detection segments whose absolute values ​​of the same index continuously exceed the corresponding base fabric statistical fluctuation threshold and quality limit are marked as over-limit segments. Adjacent over-limit segments with an interval of no more than one seam detection segment are merged. The invalid segments of the stations are used as the dividing boundary of the continuity of over-limit segments and are not included in the merging of over-limit segments. The material station to be retested and the corresponding material station are recorded separately. Calculate the ratio of the absolute value of each out-of-limit index to the corresponding quality limit as the out-of-limit ratio and sort them in descending order of value. The anomaly type corresponding to the out-of-limit index that ranks first is determined as the dominant anomaly type. When multiple out-of-limit ratios are tied for first place, they are determined as the composite dominant anomaly type. Record the starting and ending material locations, longitudinal length, dominant anomaly type, and maximum over-limit location of the joint over-limit area; Furthermore, the statistical fluctuation threshold of the base fabric is taken as the preset empirical quantile of the distribution of the absolute values ​​of the corresponding deformation index in the reference bands of the base fabric on both sides, and the quality limit is taken as the positive absolute allowable deviation limit specified in the product inspection specification of the tarpaulin to be tested. When the statistical fluctuation threshold of the base fabric is not less than the quality limit, the statistical fluctuation threshold of the base fabric is determined as the judgment threshold; otherwise, the quality limit is determined as the judgment threshold. For detection states that are not marked as dual-channel invalid and recovery sequences that are not marked as abnormal, the ratio of the absolute values ​​of thickness change, thickness change gradient, mid-surface warping, mid-surface warping gradient, short-term recoverable thickness, short-term recoverable mid-surface warping, residual thickness, and residual mid-surface warping along the longitudinal direction of the seam to the corresponding judgment threshold is used as the over-limit ratio. Any detection segment along the seam with an over-limit ratio of not less than one is marked as a dual-channel over-limit segment. Adjacent dual-channel over-limit segments are merged, and the merged result forms the seam over-limit region or expands the existing seam over-limit region. Within each joint exceeding the limit area, the exceeding ratios are arranged in descending order of value. The anomaly type corresponding to the deformation index that ranks first is determined as the dominant anomaly type. When multiple exceeding ratios are tied for first place, the anomaly types corresponding to the tied indices are jointly determined as the composite dominant anomaly type. The joint exceeding area record containing detection status, load value, evaluation time, and restoration registration verification quantity is output. For detection states marked as dual-channel invalid, the calculation of the over-limit ratio, the merging of the dual-channel over-limit segments, and the formation or expansion of the joint over-limit area are not performed. Instead, a dual-channel retest record containing the detection state, load value, evaluation time, synchronization trigger deviation, and dual-channel retest flag is output independently. For recovery sequences that are not marked as invalid in dual channels but are marked as abnormal recovery sequences, the calculation of the overlimit ratio, the merging of the overlimit segments in dual channels, and the formation or expansion of the overlimit region of the joint are not performed. Instead, a recovery retest record containing the detection status, load value, evaluation time, recovery registration verification quantity, and retest flag is output independently.

[0012] The beneficial effects of this invention are: 1. The overall six-degree-of-freedom rigid body correction is completed by using the fixture reference target outside the deformation zone, and the joint area is limited to not participating in the solution of rigid body parameters and base fabric relaxation field parameters. This can separate the overall attitude change, the gradual relaxation of the base fabric and the local deformation of the joint into layers, reducing measurement overlap.

[0013] 2. By establishing the material station correspondence through the reference cross-sectional profile features and longitudinal monotonic constraints, warping, elongation, gradient and residual deformation can be continuously calculated along the joint length. Continuous over-limit detection segments are merged into weak areas with start and end positions, length and dominant anomaly type, improving the location and interpretability of the detection results.

[0014] 3. By employing a laser profilometer with synchronous upper and lower triggering and repeated scanning at multiple times after unloading, thickness channels and mid-surface warping channels can be formed separately, and short-term recoverable deformation and residual deformation at the specified evaluation time can be distinguished, thereby reducing misjudgments caused by vibration, synchronization deviation, short-term viscoelastic recovery and unilateral high-level overlap. Attached Figure Description

[0015] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart of the laser detection method for deformation of the heat-sealed seam of the tarpaulin according to the present invention.

[0016] Figure 2 This is a flowchart of the S3 base fabric relaxation displacement field compensation method of the present invention. Detailed Implementation

[0017] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0018] refer to Figures 1-2 A laser detection method for deformation of heat-sealed seams in tarpaulins, comprising: S1. The tarpaulin sample with heat-sealed seam is clamped in the loading mechanism with load sensor. In the reference state, at least one detection load state and the unloaded state, the line laser profilometer spanning the seam area and the base fabric reference strips on both sides is used to scan along the longitudinal direction of the seam to obtain the three-dimensional profile point cloud associated with the corresponding detection state. S2. Using the fixture reference target outside the deformation zone, obtain the six-degree-of-freedom rigid body transformation of each detection state relative to the reference state, and obtain the registration point cloud unified to the reference fixture coordinate system. S3. Divide the registration point cloud into a seam area and two base fabric reference strips on both sides. Determine the displacement of the seam area and the two base fabric reference strips on both sides based on the registration point cloud of each detection state and the reference state. Fit the base fabric relaxation displacement field that varies along the longitudinal direction of the seam only based on the displacement of the two base fabric reference strips on both sides. Interpolate the base fabric relaxation displacement field to the seam area and deduct it from the displacement of the seam area to obtain the seam compensation displacement. S4. Based on the cross-sectional profile characteristics of the reference state, establish the material station correspondence between states under longitudinal monotonic constraints. According to the material station correspondence and the joint compensation displacement, calculate the relative warping, local elongation, deformation gradient and unloading residual deformation of each joint detection segment to obtain the joint deformation index sequence. S5. Compare the seam deformation index sequence with the base fabric statistical fluctuation threshold and quality limit, merge the continuously exceeding seam detection segments, and obtain and output the start and end positions, length and dominant anomaly type of the seam exceeding the limit area.

[0019] In this specific embodiment, S1 includes: The testing system cuts heat-sealed tarpaulins from the same production batch into strip samples with a length of 600 mm and a width of 120 mm. The center line of the heat-sealed seam with a nominal width of 8 mm is parallel to the long side of the sample. The left and right slope peaks of the seam in the reference state are used as the left and right boundaries, respectively. Left and right base fabric reference strips are set within a range of 20 mm to 45 mm to the left of the left boundary and 20 mm to 45 mm to the right of the right boundary, respectively. The required transverse width of the seam area and the two reference strips is 98 mm, which is less than the 100 mm transverse field of view of the profilometer. Both ends of the sample are fixed with the same clamping length to an electric loading mechanism with a 0.1 N resolution load sensor. The loading direction is consistent with the longitudinal direction of the seam. The loading mechanism controller reads the load value at 20 Hz and establishes a test task record with the sample number, batch number, clamping length, nominal seam width, and scan batch number. The loading mechanism first applies a 50 N leveling preload for 30 seconds. When the load range is no greater than 0.5 N within 5 consecutive seconds and there is no visible relaxation on the sample surface, the current state is recorded as the baseline state. Subsequently, 100 N, 200 N, and 300 N test loads are applied sequentially. After each load reaches the set value, it is held for 20 seconds and the scan is completed. Then, the load is unloaded back to the 50 N leveling preload, and scans are triggered at three recovery times: 1 second, 10 seconds, and 60 seconds after unloading. The 60-second mark is the specified evaluation time. If the load deviation exceeds 1% of the set value during the holding period, the controller extends the holding time until the stability condition is met for 5 consecutive seconds. If the condition remains unstable after 60 seconds, the state is recorded in the load anomaly record and no scan is triggered. The record is also written to the task-level anomaly bypass table, using the task number, scan batch number, and status identifier as a joint index. The anomaly category is recorded as load anomaly, and the processing status is recorded as pending retest. The line laser profilometer is installed on a linear motion module that spans the seam area and the base fabric reference strips on both sides. The lateral field of view is 100 mm and the lateral sampling interval is 0.05 mm. The motion module moves uniformly along the longitudinal direction of the seam at a speed of 50 mm per second and outputs profilometer frames at intervals of 0.5 mm. Before scanning, the instrument completes the lateral scale, normal scale and laser plane parameters calibration using calibrated stepped standard blocks. The calibration file is saved in the detection task directory with the instrument serial number and the date as indexes. Points in a single frame with a confidence level lower than the instrument calibration threshold or whose normal coordinates exceed the range are marked as invalid points and do not participate in subsequent calculations. In this specific embodiment, S1 includes a leveling preload state as a reference state to eliminate visible slack in the tarpaulin sample and stabilize the load sensor reading. At least one detection load state and at least one specified evaluation time state after unloading back to the leveling preload are set sequentially. The state control table configures the target load, stabilization duration, number of scans and evaluation time for each state. The controller only issues a trigger pulse when the load stabilization condition and the scan preparation signal are met simultaneously, and writes the measured load value instead of the target load value into the point cloud record. To ensure traceability of scanning parameters for samples within the same batch, the testing system generates instrument parameter records using planar standard parts, stepped standard blocks, and time synchronization signal sources before the first piece is tested each day. Recorded fields include lateral scaling factor, normal scaling factor, repeatability standard deviation, confidence threshold, synchronization deviation limit, instrument serial number, and effective time. If any calibration residual exceeds the manufacturer's allowable value, the scanning task is locked, and an instrument calibration pending flag is output. Simultaneously, the instrument serial number, out-of-limit residual, task number, and unexecuted plan status are written to the task-level abnormal bypass summary table. After calibration, the task number is retained, but a new scanning batch number is created, and all statuses are re-acquired. The old record's processing status is changed to "closed," and the old and new batch numbers are written to the associated fields. The old record is not deleted. Each valid measurement point is recorded as a point cloud record. , Indicates detection status The following number is Measurement point records, This represents the baseline, detection load, or unloading recovery status index, and is a dimensionless discrete index. This represents the measurement point number and is a dimensionless positive integer. The horizontal coordinate is represented in mm. Represents the vertical axis and the unit is mm. Represents the normal coordinates in mm. This represents the dimensionless reflectance confidence level returned by the profilometer. The scan trigger time is indicated in milliseconds. The status load, status identifier, and scan batch number collected by the associated load sensor are recorded and written to the original point cloud file in ascending order of the vertical coordinate. The detection system sets up opposing line laser profilometers on the upper and lower sides of the sample on the same rigid measuring frame. The two profilometers are synchronously triggered by the same hardware pulse. The upper surface measuring points and the lower surface measuring points are written into the upper surface point cloud and the lower surface point cloud with the same frame number, respectively. The offline calibration uses the same plane standard to obtain the fixed rigid body relationship between the lower coordinate system and the upper coordinate system, so that the two point clouds adopt the transverse, longitudinal and normal axes of the fixture coordinate system. The synchronous acquisition card calculates the trigger time difference for each frame. , Indicates frame number The corresponding synchronization trigger deviation, in milliseconds (ms), This indicates the trigger time of the upper profilometer, and the unit is milliseconds (ms). This indicates the trigger time of the lower profiler, and the unit is milliseconds (ms). The common frame number is a dimensionless positive integer. The synchronization deviation limit is the maximum value of the 99.9% empirical quantile of the time difference obtained from 1000 repeated synchronization triggers under no-load conditions and 0.05 ms. The limit, its sample number, acquisition date and instrument serial number are written into the synchronization calibration record. At the same material station, the distance between the upper and lower surfaces along the common normal is used as the original input for the thickness channel, and the median of the normal coordinates of the upper and lower surfaces is used as the original input for the mid-surface warping channel. When the synchronization trigger deviation of any frame exceeds the synchronization deviation limit, it is immediately re-scanned once with the same load and motion parameters. If it still exceeds the limit after re-scanning, the corresponding detection state is marked as dual-channel invalid, and the calculation of the thickness channel and mid-surface warping channel in S3 to S5 is stopped. An independent record containing the state identifier, maximum synchronization trigger deviation, limit, number of rescans, and dual-channel retest flag is written. This independent record is simultaneously added to the task-level abnormal bypass master table, and the fact that the normal channel continues to be processed is recorded in the channel status field. After scanning, the acquisition module verifies the frame number continuity, load field integrity, and common frame number consistency of the upper and lower surfaces of the original point cloud file. It writes the verified 3D contour point cloud, detection task record, synchronous calibration record, and dual-channel valid status table into the read-only detection batch directory, and transmits the point cloud path of the valid status and the fixture reference target measurement point set to S2. The task-level anomaly bypass summary table is saved in an independent file and directly read by S5, without using the generation of point cloud, status registration package, or deformation index sequence along the seam as a transmission condition.

[0020] In this specific embodiment, S2 includes: The registration module reads the point cloud of each state output by S1 and arranges four non-coplanar spherical fixture reference targets on the non-deformable steel crossbeams on both sides of the loading mechanism. The four targets are distributed around the sample scanning area and do not deform with the sample under load. The module extracts spherical point clusters from the point cloud according to the target number, obtains the center of each target by least squares spherical fitting, and writes the coordinates of the reference state center and the coordinates of the current state center into the target coordinate table. The table also records the number of point clusters, spherical fitting residuals, detection status and scanning batch number. For detection status The registration module finds the rigid body transformation that minimizes the weighted squared distance. , Indicates detection status A six-DOF rigid body transformation to the reference fixture coordinate system, with output coordinate units in mm. This represents any three-dimensional coordinate vector in the current state, with each component in mm. This indicates that it is obtained by singular value decomposition. A dimensionless orthogonal rotation matrix with a determinant of 1. This represents a three-dimensional translation vector in mm. The index represents the detection status and is a dimensionless discrete index. The solution uses only the center of the fixture reference target and does not use the seam area or the base fabric reference strip measurement point. Registration module with Calculate the registration verification quantity. Indicates detection status The root mean square coordinate residuals of the target, in mm. This represents the number of valid targets participating in the solution, and must be a dimensionless positive integer and not less than 3. This represents the target number, which is a dimensionless positive integer. Indicates the reference state. The target center coordinate vector is given in mm. Indicates detection status The following are the center coordinate vectors of targets with the same number, and the unit is mm. In the aforementioned rigid body transformation, all three coordinate components in the Euclidean norm are represented in mm. Registration error limit adopted , This indicates the registration error limit, and the unit is mm. The expanded measurement uncertainty, expressed in mm, is obtained by calibrating a line laser profilometer using a stepped standard block. This represents the standard deviation of the coordinate residuals of the fixture reference target over 50 repeated scans, in mm. The dimensionless coverage coefficient, which is used for both the registration error limit and the recovery noise tolerance, is determined by the cumulative coverage probability verification of the repeated scanning residuals of the planar standard, so that the range of plus or minus three standard deviations covers no less than 99.7% of the effective residual samples. In this specific implementation, it is set to 3. In this specific implementation, the calibration yields an expanded measurement uncertainty of 0.04 mm and a coordinate residual standard deviation of 0.02 mm. Therefore, the registration error limit is 0.10 mm. The calibration results are saved in the registration parameter file according to the instrument serial number and the measurement frame number. If there are fewer than 3 valid targets, the rank of the geometric matrix formed after the center coordinates of the valid targets are decentered is less than 2, or the registration verification quantity exceeds the registration error limit, the registration module will mark the detection state as a registration anomaly, record the missing target number, matrix rank, registration verification quantity and error limit, stop the subsequent deformation index calculation for this state and generate a registration retest flag; the registration retest record is written into the task-level anomaly bypass table with task number, scan batch number and state identifier, and can be directly read by S5 even if no state registration package is generated; a matrix rank of not less than 2 indicates that at least three valid targets are not collinear, satisfying the geometric condition that the rigid body transformation is uniquely determined by the three-dimensional corresponding points; when the verification quantity is not greater than the limit, the rigid body transformation is applied to all valid points in this state and the point clouds of the upper and lower surfaces; Transformed measuring points according to Write the registration point cloud. This indicates the detection status unified to the reference fixture coordinate system. Next The coordinate vectors of the measurement points are given, and the unit is mm. This represents the coordinate vector of the same measuring point before transformation, with units in mm. This represents the aforementioned six-degree-of-freedom rigid body transformation. This indicates a detection state index that is a dimensionless discrete index. The measurement point number is a dimensionless positive integer. The module retains the original point number, common frame number, and state load to maintain the data association between the upper and lower surfaces and at different times. The registration module extracts the verification points of the measurement frame that were not involved in the solution to verify the transformation results. The residual of the verification point must not be greater than the registration error limit. If it exceeds the limit, it is treated as a registration abnormal branch. When the verification is passed, the rigid body transformation matrix, translation vector, target coordinate table, registration verification quantity and registration point cloud are written into the state registration package. The status registration package is stored using the scan batch number and detection status as a joint index, and a normal channel valid flag, a dual channel valid flag, and a registration pass flag are set respectively. When the registration pass flag is true and the normal channel valid flag is true, S3 receives the normal single-sided registration point cloud. It does not use the dual channel valid flag as the entry condition for the normal channel. When the dual channel valid flag is false, it only blocks the upper and lower surface synthesis, thickness channel, and mid-surface warping channel. For the state where the registration pass flag is false, it does not provide any deformation calculation input to S3, but the corresponding task-level abnormal bypass record is retained to S5 without going through S3 and S4.

[0021] In this specific embodiment, S3 includes: The compensation module reads the registered point cloud from S2 that has passed registration and is valid in the normal channel, and determines the lateral coordinates based on the peak value of the lateral slope of the heat-sealed edge of the reference state joint. The left boundary and horizontal coordinates are The right boundary of the coordinate range The left base fabric reference zone was determined, and the coordinate range was set. The right base fabric reference strip has been identified. and The left and right boundaries of the seam are represented by their transverse coordinates in mm. The region definition file saves the coordinates of the inner and outer boundaries of the seam centerline, the two boundaries, and the two reference zones. It serves as the only valid region coordinate version from S3 to S5. All detection states reuse this file. The seam area measuring points do not enter the parameter solution of the base fabric relaxation displacement field. The module performs local planar neighborhood mapping between the current state and the reference state registration point cloud on a 0.5 mm vertical grid and a 0.5 mm horizontal grid. The neighborhood radius is set to 0.8 mm. The displacement vector is obtained by the coordinate difference of corresponding points at the same spatial grid coordinates. , Representing state Lower space grid points The three-dimensional Euler displacement vector relative to the reference state and the same coordinate grid point, with units in mm. Representing state Lower grid points The registration coordinate vector is in mm. This represents the registration coordinate vector of points on the same spatial grid in the reference state, with units in mm. This indicates a detection state index that is a dimensionless discrete index. The spatial grid point number is represented by a dimensionless positive integer. Grids without corresponding points are written as missing measurements without zero-value filling. The Euler displacement is only used for estimating the gradually varying relaxation field of the left and right reference zones and is not directly used as the state difference value of the S4 material station. S4 must use the compensated absolute object in... and Reconstruct the material difference between them; For each displacement component, the module establishes a semi-overlapping longitudinal window with a fixed length of 40 mm and a fixed movement step size of 20 mm, starting from the longitudinal origin of the region definition file. The window length is taken as the minimum allowable value of the aforementioned relaxation wavelength, 40 mm, and the starting point of the window increases by 20 mm sequentially. If the end is less than 40 mm, another 40 mm window is set with the effective scan endpoint as the right boundary. Duplicates with the same start and end coordinates as the existing window are removed, so that the first and last samples belong to at least one window. For each longitudinal window, calculations are performed. as well as , Representing state Lower sample The absolute deviation from the median displacement of this window, in mm. This represents the scalar component of the displacement vector in the current processing direction, with units of mm. This represents the median displacement of the valid samples in the window, in mm. This represents the median of the absolute deviation of the window, in mm. Indicates the detection status index. The index represents the sample index within the window, both of which are dimensionless discrete indices. Missing points are not included in the median or absolute deviation statistics. When the proportion of effective samples in the window is less than 70% or there are consecutive missing points greater than 2 mm within the window, it is not feasible to directly trigger the relaxation field solution on that side without performing MAD discrimination. The abnormal rejection criteria are: , This represents the dimensionless multiple of the displacement noise distribution coverage probability calibrated by repeatedly scanning 30 times with a seamless base fabric under the same load and scanning speed, which is taken as 3.5 in this specific embodiment. This represents the minimum deviation baseline in mm that avoids zero discrete window failure, with the instrument normal resolution set at 0.01 mm. , These represent the median absolute deviation of the aforementioned absolute deviation and the absolute deviation of the current coverage window, respectively, in mm. Samples located within two semi-overlapping valid windows are only written into the anomaly table and removed if the anomaly removal criteria are met in both windows. Samples located at the beginning and end and covered by only one valid window are judged according to that window. Windows that do not meet the validity criteria do not generate normal or abnormal votes. Window length, step size, start and end coordinates, number of valid samples, missing test ratio, and voting results are all written into the fitting file. For the retained left and right reference zone displacement samples, form equally spaced sequences according to a 0.5 mm longitudinal grid. When the continuous missing measurement length on each side is no more than 2 mm, only the nearest effective sample at both ends of the gap is used as the linear filling point within the gap. When the continuous missing measurement length on any side is greater than 2 mm, no zero filling, cross-gap interpolation, or segment splicing is performed. The side and the corresponding detection state are immediately judged as infeasible for relaxation field solution and the Discrete Fourier Transform is stopped. Only for the numerical sequence where all gaps have completed the unique linear filling point and the effective sample ratio is not less than 70%, the Discrete Fourier Transform is performed on each displacement component. The frequency coefficients with longitudinal wavelengths less than the preset relaxation wavelength lower limit are set to zero and inversely transformed to obtain the slowly varying displacement sequence. The preset relaxation wavelength lower limit is taken as the maximum value of the wavelength corresponding to 40 mm when the cumulative energy of the displacement power spectrum under the same load of the seamless base fabric reaches 95%. Therefore, the non-zero frequency components entering the fitting result all satisfy that the longitudinal wavelength is not less than this lower limit. The fitting file records the load file, wavelength lower limit, retained frequency index, filling point interval, inverse transform sequence, and calibration data version. For the horizontal coordinate of the joint area The module is in accordance with Interpolation yields the relaxation displacement field of the base fabric. Representing state In coordinates The three-dimensional base fabric relaxation displacement vector at the location, with units in mm. Indicates the left reference band on the vertical axis The three-dimensional fitting displacement vector at the location is given in mm. Indicates the right reference band on the vertical axis The three-dimensional fitting displacement vector at the location is given in mm. This represents the dimensionless weight on the left side, determined by the distance from the seam point to the right boundary as a percentage of the seam width. This represents the dimensionless weight on the right side, determined by the distance from the seam point to the left boundary as a percentage of the seam width. Both weights are between 0 and 1, and their sum is 1. and These represent the horizontal and vertical coordinates, respectively, with the unit being mm. Indicates a dimensionless detection state index; Joint compensation displacement according to calculate, Representing state In the seam coordinates The displacement of the three-dimensional joint at the location is compensated, and the unit is mm. This represents the three-dimensional displacement measured in the joint area, expressed in mm. This represents the three-dimensional relaxation displacement of the base fabric obtained by the aforementioned interpolation, in mm. and These represent the horizontal and vertical coordinates, respectively, with the unit being mm. The dimensionless detection state index is represented, and the compensation results, along with the fitting residuals of the left and right reference bands, the proportion of outliers, and the region version, are written into the joint compensation displacement file. To ensure that the joint compensation displacement directly forms the subsequent coordinates, the module is configured according to... After generating the compensated seam points, Representing state Lower joint coordinates The compensated 3D point coordinate vector, in mm. This represents the coordinate vector of the registration point at the same material location in the reference state, with units in mm. This represents the aforementioned joint compensation displacement vector, with units in mm. and These represent the horizontal and vertical coordinates, respectively, with the unit being mm. Indicates the dimensionless detection state index; the compensated point only contains one relaxation field subtraction of the reference bands on both sides, and S4 must not subtract the base cloth fitting surface from its normal coordinates again; For the dual-channel active state, the module is in each 0.5 mm longitudinal coordinate. Calculate the normal spacing between the upper and lower surfaces of the joint area, the left reference zone, and the right reference zone, respectively. , and First press Save the current absolute compensation thickness channel value, then press... Form the base fabric thickness relaxation amount with the same spatial coordinates, and according to The thickness change is generated after compensation with the same spatial coordinates; Representing state In the vertical axis The thickness channel value after deducting the absolute thickness of the base fabric on both sides from the background, and the unit is mm. Representing state In the vertical axis The thickness relaxation of the base fabric is obtained by interpolation from the reference strips on both sides, and the unit is mm. This represents the thickness change of the seam area after compensation at the same spatial grid location, expressed in mm. , and Representing states respectively The thicknesses of the lower seam area, left reference zone, and right reference zone are shown in mm. The subscript 0 indicates the reference state. and Indicates the horizontal coordinate at the center of the seam. The aforementioned dimensionless interpolation weights at the given location sum to 1. This represents the dimensionless detection state index. and These represent the longitudinal coordinate and the transverse coordinate of the seam center, respectively, with units in mm; The module takes the median value of the normal coordinates of the upper and lower surfaces of the seam area, the left reference zone, and the right reference zone as the data. , and First press Save the current absolute compensation mid-surface channel value, then press... Form the relaxation amount of the base fabric in the same spatial coordinates, and according to The mid-surface warping is formed after compensation in the same spatial coordinates; Representing state In the vertical axis The mid-plane channel value after deducting the absolute mid-plane background of the base fabrics on both sides, in mm. Representing state The relaxation amount is obtained by interpolating the displacement of the reference zones on both sides, and the unit is mm. This represents the warping amount after surface compensation in the joint area at the same spatial grid location, expressed in mm. , and Representing states respectively The median values ​​of the upper and lower surfaces of the three regions are in mm. The meanings of the remaining subscripts, weights, and coordinates are the same as those in the previous section. The thickness channel and the mid-surface warpage channel respectively store the absolute compensation channel value, the spatial difference, and the valid flag. The module's effective points within the 0.5 mm vertical grid are respectively... and Calculate the thickness variation gradient and the mid-surface warping gradient in the same space. and Representing states respectively Lower space grid points Two types of dimensionless diagnostic gradients, and This represents the thickness change after compensation between adjacent spatial grid points, expressed in mm. and This represents the warping of the mid-surface after compensation for adjacent spatial grid points, expressed in mm. and Represents adjacent vertical coordinates in mm. This represents the dimensionless detection state index. This indicates the dimensionless spatial grid point index, with the first and last valid points using a one-sided difference with the same unit relationship; the same spatial gradient is only used for S3 fitting diagnosis and does not enter the S5 quality judgment. The dual-channel gradient for S5 judgment is recalculated by S4 after material station mapping. If the missing measurement ratio of any surface at a certain longitudinal position exceeds 20%, the module marks that position as a missing channel and interrupts the gradient calculation across that position; for the invalid state of dual channels, the upper and lower surface synthesis, thickness reference band relaxation compensation, mid-surface reference band relaxation compensation, and the calculation of the two types of compensation quantities and gradients are not performed, and the dual-channel remeasurement record generated by S1 is maintained, but this restriction does not change the valid mark of the ordinary channel of the ordinary single-sided point cloud in the same state; When the proportion of retained samples in the base fabric reference band on any side is less than 70%, the length of any consecutive missing measurement is greater than 2 mm, or the root mean square residual of the low-pass fitting exceeds 0.10 mm, the compensation module determines that the relaxation field solution is not feasible. Among them, the length of consecutive missing measurement is more than 10 mm and is recorded as a serious missing measurement, but missing measurements of 2 mm to 10 mm are also not allowed to enter the frequency domain fitting. The module stops generating the seam compensation displacement and dual-channel index, uses the uncompensated registration point cloud as read-only retained data and does not use it for quality judgment, and outputs a record containing the task number, scan batch number, status identifier, failure condition, sample proportion, missing measurement length, fitting residual and compensation retest flag, writes it into the task-level abnormal bypass master table, and only restores the data transmission to S4 after rescanning and passing all conditions. The compensation module ultimately outputs a joint compensation displacement file, compensated joint points, absolute compensation thickness channel value sequence, absolute compensation mid-surface channel value sequence, thickness and mid-surface difference in the same space and its diagnostic gradient, anomaly point table, and region definition file. It sets valid flags for both ordinary and dual channels according to the detection status. S4 only establishes material station correspondence for the valid state of the ordinary channel, and forms material differences in the ordinary channel using the compensated joint points rather than the joint compensation displacement in the same space. When the dual-channel valid flag is true, it forms material differences using the absolute compensation channel values ​​rather than the difference in the same space. All outputs carry the 0.5 mm longitudinal grid origin, spacing, and missing measurement flag. Statuses where the above outputs are not generated are still directly passed to S5 by the task-level anomaly bypass summary table.

[0022] In this specific embodiment, S4 includes: The index calculation module uses the S3 baseline state joint area contour as the material station reference. It extracts a cross section every 2 mm along the longitudinal direction of the joint. In each cross section, it records the joint peak height, the position of the left and right slope peaks of the joint, the cross section area in square millimeters, the left and right boundary spacing and the contour sampling vector. Stations with a cross section effective point ratio of less than 80% are directly marked as needing to be retested and do not enter the state matching. First press the module Determine the effective discrete scale, and then use Transformation; Indicates the first The effective standard deviation of the class feature, with the same unit as the original feature of that class and greater than 0. This represents the first of all valid stations in the reference state. The standard deviation of the class features and the units are the same as those of the original features. This represents the result of 50 repeated scans of a planar standard part. The repeatability standard deviation of the class feature is greater than 0 and has the same unit as the original feature. Indicates the first The first station Dimensionless standardized values ​​of class features This represents the corresponding original feature value. This represents the corresponding benchmark mean. Indicates a dimensionless feature type index. This indicates a dimensionless material station index; if the repeatability standard deviation record is missing, non-positive, or the units are inconsistent, this type of feature is discontinued; if there are fewer than 3 types of remaining valid features, the station is marked as to be remeasured; the cross-sectional area components are transformed to the same unit using the mean square millimeter and the effective standard deviation square millimeter, and then the effective cross-sectional profile features and longitudinal position are combined to form the material station identifier. For the benchmark station ,state Candidate positions and the previous actual matching station in the dynamic programming path Module definition ; This represents the dimensionless matching cost. and Representing the reference station positions With state Candidate positions dimensionless eigenvectors This represents the calculation of the Euclidean norm 2 of the dimensionless eigenvalue difference vector within the double vertical lines. and This represents the longitudinal coordinates of adjacent reference stations, in mm. and This represents the vertical coordinates of the current candidate station and the previous actual matching station on the path, in mm. This indicates a 2 mm station spacing reference, with the unit being mm. This represents the weight of dimensionless feature differences. The dimensionless spacing variation weights are 0.7 and 0.3, determined by a grid search that minimizes the mismatch rate on 30 sets of repeatedly loaded seamless base fabric data, and satisfying the following conditions. , , , and All represent dimensionless discrete indices; when the first station has no predecessor, only the feature difference term is calculated and the candidate range is constrained by a position window of ±6 mm. The matching module uses dynamic programming to search for the path with the minimum total cost in the candidate station sequence. The path must satisfy the following conditions: The longitudinal monotonic constraint limits the longitudinal offset of a single station to within ±6 mm of the reference coordinates. Indicates relative to the reference station position The corresponding candidate station index for the current state is a dimensionless positive integer. This represents the candidate station index corresponding to the next reference station and is a dimensionless positive integer. The reference station index is a dimensionless positive integer; each state transition includes the predecessor. Substituting the preceding interval term, the dynamic programming table stores the cumulative cost, predecessor index, validity flag, and current state vertical coordinate row by row. And form from each reference material station to state Candidate positions A unique mapping table; prohibiting the use of candidate sequences that are not selected by the path. It is also prohibited to replace the actual forerunner with a non-corresponding position outside the joint area to replace the missing station; The matching threshold is set to When there are multiple material stations with the same objective function value, i.e., the difference between the minimum total cost path and the second minimum total cost path is not greater than the threshold, or there are two candidate results with matching cost differences not greater than the threshold for the same reference station, the material station is marked as invalid matching, and the station and the detection segment along the seam with the station as the endpoint are marked as invalid station segments. The calculation of relative warping, local elongation, deformation gradient and unloading residual deformation of these segments is stopped, and the record containing the reference station, candidate station, two costs and the material station to be retested is output. For effective material stations, the module is configured according to... Calculate the relative warpage. Representing state Downmapped to the The relative warpage of each reference material station, in mm. This represents the dimensionless unit normal vector of the base fabric fitting surface on both sides of the reference state. This represents the transpose of the unit normal vector. This indicates the horizontal coordinate of S3 at the center of the joint. Coordinates corresponding to the current state The output or interpolated result is the compensated absolute point coordinate vector, with units in mm. This represents the absolute point coordinate vector of the same material station under the reference condition, with the unit being mm. Representing state China and benchmark material station The corresponding longitudinal coordinates of the candidate stations, in mm. This indicates the reference longitudinal coordinate of the material's location, in mm. This represents the candidate station index. This represents the dimensionless detection state index. This indicates the index of the dimensionless reference material station. If the current coordinate does not fall on an S3 grid point, linear interpolation is only allowed between the nearest valid points within 0.5 mm to the left and right of the longitudinal direction. If there are no valid points on either side, the total interpolation span exceeds 1 mm, or the corresponding path is broken at that point, the ordinary station will be marked as invalid. Local elongation according to calculate, Representing state Next Dimensionless local elongation of the detection segment along the seam Representing state Next adjacent effective material station and The arc length along the centerline of the joint, in mm. The arc length of the same material segment under the reference condition is expressed in mm and must be greater than 0. This represents the dimensionless detection state index. This represents the dimensionless index of the detection segment along the seam; a positive value indicates elongation, and a negative value indicates contraction. The 0.5 mm dual-channel mesh in S3 uses the same longitudinal origin as the 2 mm reference material station in this step, but S4 does not directly read the thickness variation or mid-surface warpage formed by S3 according to the same spatial coordinates; for each reference material station... The module first uses the current state coordinates given in the corresponding table. Read the absolute compensation thickness channel value and absolute compensation surface channel value Then press separately and This creates a dual-channel differential measurement at the same material station location; Representing state Mapped to reference material station The amount of thickness change after compensation, in mm. This indicates the amount of warping of the mid-surface after compensation at the same material location, expressed in mm. and These represent the absolute compensation thickness and mid-surface channel value at the current state mapped coordinates, respectively, in mm. and These represent the absolute compensation channel values ​​for the same material station under the baseline condition, with units in mm. and This indicates a dimensionless discrete index. When the current coordinate does not fall on a grid point, linear interpolation is only allowed using the two nearest valid grid points within 0.5 mm on either side of the vertical axis. If there are no valid points on either side, the total interpolation span exceeds 1 mm, or the corresponding path is broken, the material station will be marked as an invalid dual-channel station. For internal valid stations, the module is configured as follows: Calculate the relative warp gradient, and first... Determine the first The center coordinates of the detection section along the seam, then according to Calculate the local elongation gradient; for effective dual-channel stations, the module will be configured separately. and Calculate the thickness variation gradient and mid-surface warping gradient at the material station location; This represents a dimensionless relative warping gradient. This represents the local elongation gradient, expressed in millimeters to the power of negative one. and These represent the dimensionless material station gradients formed by the difference between the mapped two channels, with superscripts indicating their respective properties. This indicates the material station location mapping caliber, not the detection status index. and These represent the center coordinates of two adjacent detection segments, with the unit being mm. The meanings of the remaining station coordinates, status index, and material station index are the same as those of the previous segment. The first and last consecutive valid intervals use the same coordinate reference for single-sided difference. For ordinary stations or dual-channel stations, the difference is not calculated across the breakpoints on both sides of the invalid breakpoint. S5 only reads the material station gradient of this segment and does not read the same spatial diagnostic gradient of S3. The evaluation status is recorded 60 seconds after uninstallation. The module uses its own material station location correspondence table based on the evaluation status, according to... Determine the three-dimensional residual vector mapped to the same reference material station, and according to... This results in the residual deformation of the unloading process in the ordinary channel, which is then used for judgment by S5. Indicates the first The three-dimensional unloading residual vector of each reference material station, in mm. Indicates the evaluation status In its corresponding coordinates The absolute point coordinate vector after compensation is read from the reference point or interpolated according to the aforementioned 1 mm span limit, and the unit is mm. This represents the absolute point coordinate vector of the same material station under the reference state after compensation, with the unit being mm. Indicates the evaluation status relative to the baseline station. The corresponding vertical coordinate, in mm. This indicates a dimensionless index for the evaluation state. This represents the Euclidean norm of the unloading residual vector, expressed in mm. This represents finding the Euclidean norm of a three-dimensional vector within the double vertical lines. This indicates the dimensionless reference material station index; if the corresponding measurement fails or the interpolation exceeds the limit, the residual quantity will be invalidated and written into the material station remeasurement record. For the dual-channel effective state, the module calculates the corresponding material station positions at 1 s, 10 s, and 60 s after unloading, and then calculates the corresponding coordinates at each recovery time. Read or interpolate the absolute compensation thickness channel value and the absolute compensation mid-surface channel value according to the aforementioned 1 mm span limit, and subtract the value of the same material station under the reference state. The corresponding absolute compensation channel value at that location is obtained from the value defined in the previous section. and Subsequently, a recovery sequence is formed based on the same 2 mm reference material station; if any corresponding failure, path breakage, or interpolation exceeding the limit occurs at any time, the recovery sequence for that station is invalid; the recovery noise tolerance is taken as... , This represents the allowable positive noise tolerance for adjacent amplitudes in the recovered sequence, expressed in mm. In this specific embodiment, it is 0.10 mm. This indicates the expanded measurement uncertainty of the line laser profilometer used in S2, with the unit being mm. This represents the standard deviation of the target repeat measurement coordinate residuals used in S2, in mm. This indicates that S2 represents the same dimensionless coverage coefficient that has been defined and determined by the residual cumulative coverage probability check; in this specific implementation, it is taken as 3. If the absolute value of the later value at adjacent time points increases by more than [a certain percentage] compared to the absolute value of the earlier value... Or both absolute values ​​exceed If a sign reversal occurs, the recovery sequence of that material station is marked as abnormal; For thickness sequences not marked as recovery sequence anomalies, the module uses and Form a monotonic amplitude sequence, and then... The cumulative thickness is a short-term recoverable item; Indicates the first The recovery time is the first The non-negative thickness variation amplitude at each station after noise tolerance constraints, in mm. This indicates the thickness change after compensation at the corresponding time point, with the unit being mm. This indicates a recoverable item with a short thickness, and the unit is mm. This represents the total number of recovery times, which is a dimensionless positive integer and is set to 3. Indicates the dimensionless recovery time sequence number. This indicates the dimensionless material station index; increments less than or equal to the tolerance are pushed to the previous value, while increments exceeding the tolerance are entered into the previous abnormal branch and do not participate in the calculation. For mid-surface sequences not marked as recovery sequence anomalies, the module uses and Form a monotonic amplitude sequence, and then... The cumulative mid-surface warping can be recovered in a short time. Indicates the first The recovery time is the first The non-negative midplane warpage amplitude of each station after noise tolerance constraints, in mm. This indicates the amount of mid-surface warping after compensation at the corresponding time point, expressed in mm. This indicates that the mid-surface warping can be recovered in a short time, and the unit is mm. , and The meaning is the same as the previous paragraph; Thickness residual terms Calculate the residual term of mid-surface warping according to calculate; and They represent the first The non-negative thickness residuals and mid-surface warpage residuals at each station at the specified evaluation time, in mm. and They represent the first The thickness and mid-surface monotonic amplitude at each recovery time, in mm. and These represent the thickness and mid-surface amplitude at the first recovery moment, respectively, in mm. and This indicates a short-term recoverable item, and the unit is mm. and All represent dimensionless indices; the above equations ensure that the amplitude at the first time step is strictly equal to the sum of the short-time recoverable term and the residual term at the specified evaluation time; At each recovery time, the recovery registration verification quantity is calculated using the S2 method. If the verification quantity exceeds the 0.10mm limit at any recovery time, or if any material station experiences a retracement exceeding the recovery noise tolerance, a sign reversal exceeding the tolerance, or if the material station fails to correspond or the interpolation exceeds the limit, the corresponding recovery sequence is marked as an abnormal recovery sequence. The calculation of the short-term recoverable thickness item, short-term recoverable mid-surface warpage item, residual thickness item, and residual mid-surface warpage item for that sequence is stopped. The output includes a record containing the task number, scan batch number, status identifier, trigger time, retracement increment, sign status, recovery registration verification quantity, mapping status, and recovery retest flag, and is synchronously written to the task-level abnormal bypass master table. For dual-channel invalid states, no recovery sequence is established, and dual-channel retest records are maintained. To construct the standard channel base fabric index samples required for S5, the module divides each 25 mm wide reference band into five 5 mm sub-bands along the transverse direction. Material stations are established on the center line of each sub-band with the same 2 mm longitudinal origin and station spacing as the seam. The same rigid body correction, longitudinal monotonic correspondence, and seam compensation displacement calculation methods are used for each sub-band. However, when calculating the gradually changing background of the current sub-band, only the other nine sub-bands are used. Then, the relative warping, local elongation, two types of gradients, and 60 s residual vector Euclidean norm are calculated according to the same definition as the seam. Each sample saves the index type, left and right sides, sub-band number, material station, detection status, unit, and valid mark. Samples with a missing measurement ratio of more than 20%, invalid matching, or crossing invalid breakpoints are removed. For the dual-channel base fabric corresponding index samples, the module forms thickness sequences and mid-surface sequences in the above ten sub-bands based on the upper and lower surface spacing and the median of the upper and lower surfaces, respectively. The low-pass sequences of the other sub-bands are used as the reference relaxation field of the sub-band. The absolute compensation thickness channel value and absolute compensation mid-surface channel value of each state are formed in the same way as in S3. The 0.5 mm absolute compensation sequence of each state is read from the current longitudinal coordinate given in the corresponding table of the state or interpolated according to the aforementioned 1 mm span limit. Then, the absolute compensation channel value of the same 2 mm material station in the reference state is subtracted. The compensation dual-channel difference, material station gradient, two types of short-time recoverable terms and residual terms are calculated according to the material station difference, recovery time, noise tolerance and closure formula of the same step. The effective samples of the left and right reference bands are only merged into a sample set of the same index type without mean replacement, forming a base fabric corresponding index sample table for S5 to calculate empirical quantiles. The index calculation module writes the relative warpage, local elongation, relative warpage gradient, local elongation gradient, scalar unloading residual deformation, and material station thickness change after compensation, material station thickness change gradient, material station mid-surface warpage after compensation, material station mid-surface warpage gradient, short-term recoverable items, and residual items into the seam deformation index sequence based on the 2 mm reference material station. At the same time, it writes the valid flags of ordinary stations, dual-channel stations, recovery sequence, load values, evaluation time, and the version of the corresponding table for each state. It then transmits this sequence, the base fabric corresponding index sample table, and all records to be retested to S5. The dual-channel difference and diagnostic gradient formed by S3 according to the same spatial coordinates are not written into this judgment sequence. The task-level abnormal bypass summary table is still read independently by S5 to cover the states that failed to generate the sequence.

[0023] In this specific embodiment, S5 includes: The judgment module reads the seam deformation index sequence of S4 and the corresponding index sample table of the base fabric. It filters samples according to index type, detection status, ordinary channel or dual-channel set identifier, and unit. After removing samples with missing values, invalid matches, cross-invalid breakpoints, and abnormal recovery sequences, it first counts the number of valid samples on the left and right sides and the proportion of each valid sample to the candidate samples on that side. Only when each side has at least 50 valid samples, the proportion of valid samples on each side is at least 70%, and the units on both sides are consistent, are the samples from both sides merged and sorted according to... Determine the statistical fluctuation threshold of the base fabric; Indicates indicator type The baseband statistical fluctuation threshold is set, and the unit is the same as the corresponding indicator. The operation represents the 99th percentile of the empirical distribution of the effective sample. This indicates that S4 is constructed according to the same definition as the seam. No. Each base fabric measurement sample has the same unit as the corresponding index. This indicates an index for dimensionless index types. This indicates the dimensionless base fabric sample index; left and right sides, sub-band number, number of samples on each side, and effective proportion of each side are all retained in the sample record; Each indicator type Set a threshold validity flag separately; the threshold validity flag is only set to true when the left and right sides respectively meet the requirements of no less than 50 valid samples and no less than 70% valid proportion in the front segment, and no less than 100 valid samples in total on both sides, with all sample units being consistent; if either side does not meet the requirements, the flag is set to false, and the threshold calculation, out-of-limit segment marking, region formation, and dominant anomaly ranking for that indicator are stopped. A record containing the detection status, indicator type, number of valid samples on the left side, number of valid samples on the right side, valid proportion on both sides, missing side, unit inspection results, and threshold sample insufficiency flag is output; other threshold validity indicators continue to be judged independently, but the result integrity flag for that detection status is set to incomplete. Quality Limits The samples were retrieved from the inspection specifications of the tarpaulin products to be tested, using a combination of product model, base fabric weight, seam process, and index type. Indicates indicator type The positive absolute allowable deviation limit and the unit is the same as the corresponding index. This indicates a dimensionless index type index. Each index is set with a valid standard flag. The flag is true only when a unique valid record is found, the value is greater than 0, and the unit is consistent with the index. If there is no match, multiple matches, the value is not positive, or the unit is inconsistent, the flag is set to false, stopping the over-limit segment marking, region formation, and dominant anomaly sorting of the index. The output includes the standard version, failure reason, related index, and standard confirmation flag. At the same time, the status result integrity flag is set to incomplete. For indicators where both the threshold valid flag and the standard valid flag are true, the threshold is determined according to... Sure, Indicates indicator type The final positive value threshold is determined, and the unit is the same as the corresponding indicator. This indicates the statistical fluctuation threshold of the baseband and has the same unit as the corresponding indicator. This indicates the quality limit of the product inspection specification, and the unit is the same as the corresponding indicator. This indicates a dimensionless index type index; the module sets the validity flag of this index to true, and if any preceding flag is false, the validity flag is false and no index is generated. This prevents invalid thresholds from being included in subsequent calculations; The module calculates the relative warpage, local elongation, relative warpage gradient, local elongation gradient, and scalar unloading residual deformation of the ordinary channel that are deemed valid. The dimensionless exceedance ratio, as a relative quality limit, is calculated simultaneously. As a unified ranking score across channels; Indicates indicator type In the detection section The percentage of out-of-limit relative to the quality limit This represents the dimensionless unified ranking score of the same measured value relative to the final judgment threshold. This represents the measured value and the unit is the same as the threshold. This indicates a positive quality limit. This indicates the final threshold for positive values. This indicates an index for dimensionless index types. Indicates a dimensionless detection segment index; only when and The normal channel is marked as exceeding the limit segment when the value is equal to any threshold segment; For segments of the same index that exceed the limit in vertical order and Module calculation , This represents the number of valid detection segments between two out-of-limit segments and is a dimensionless, non-negative integer. This represents the index of the detection segment at the end of the previous over-limit segment, and is a dimensionless positive integer. This represents the index of the starting detection segment of the next out-of-limit segment, and is a dimensionless positive integer. and This indicates the adjacent over-limit segment number and is a dimensionless discrete index. When the number of intervals does not exceed 1 and there are no invalid station segments in the interval, the merging is performed; otherwise, the current region ends and a new region is established from the next over-limit segment. The module scans the over-limit sections in ascending order of material stations. It merges the over-limit sections with the same index that are adjacent or separated by only one valid detection section without exceeding the limit into the index section. Interval sections are merged into the section but their original index values ​​are not changed. Invalid sections of ordinary stations or invalid sections of dual-channel stations are used as the continuous forced division boundary according to the affected channels and do not participate in the merging. The material station retesting mark and its coordinates are written separately into the retesting table. Each index section saves the index of the first and last detection sections, the first and last material stations and the index type. After merging all indicators, the module arranges all ordinary channel indicator segments in ascending order of starting coordinates. When the current segment and the next indicator segment overlap, connect, or are separated by only one valid detection segment that does not exceed the limit at the material station and there are no invalid stations between them, the two segments are merged to form a unified joint over-limit area. Otherwise, the current area is terminated and a new area is created. Interval inclusion, partial overlap, and chain overlap are all updated according to the minimum starting point and maximum ending point after the merge. Cross-indicator merging does not change the original indicator values ​​and segment numbers within the indicator, so that the subsequent dominant anomaly sorting has a unique regional carrier. The longitudinal length of each merged region is as follows calculate, Indicates the region The longitudinal length, in mm. Indicates the region The longitudinal coordinates of the end material station, in mm. Indicates the region The longitudinal coordinates of the starting material station, in mm. The number of the joint exceeding the limit is a dimensionless positive integer. If the area contains only one inspection segment, the start and end coordinates are the material station coordinates of the two endpoints of the segment, and the length is not recorded as zero. Within the joint exceeding the limit area containing only ordinary channel indicators, the module is first judged according to the ratio of exceeding the quality limit. The maximum values ​​are sorted in descending order to determine the dominant anomaly type in the normal channel; when a region contains dual-channel indicators or is extended by dual-channel segments, all normal and dual-channel indicators in the region are sorted using a unified ranking score relative to their respective final judgment thresholds. Sort in descending order, and write the warping anomaly, elongation anomaly, gradient anomaly, thickness anomaly, or residual deformation anomaly corresponding to the first index as the dominant anomaly type; the difference between two or more sorted values ​​should not exceed [value missing]. When the maximum exceedance location is recorded as a composite dominant anomaly type, the material station where the first value is located is taken as the maximum exceedance location. When there are more than one parallel location, the station with the smallest vertical coordinate is taken as the record location point and all parallel lists are retained. Regional severity by Sure, Indicates the area where the seam exceeds the limit. The dimensionless maximum unified ranking score relative to the final judgment threshold. Indicates whether it is a normal channel or a dual channel indicator. In the detection section The unified sorted score, This indicates an index for dimensionless index types. Indicates the index of the dimensionless detection segment. Indicates the dimensionless region number; since all All values ​​use the corresponding final judgment threshold as the denominator. Cross-channel values ​​have the same meaning of "relative judgment boundary multiple". Regions with the same severity are then output in ascending order of vertical starting coordinates. For states where both channels are valid, both channel positions are valid, the recovered sequence is valid, and the determination is valid, the module calculates the compensated thickness change, thickness change gradient, compensated mid-surface warpage, mid-surface warpage gradient, two types of short-term recoverable terms, and two types of residual terms. and order ; Indicates the type of dual-channel indicator In the detection section The dimensionless over-limit ratio relative to the final judgment threshold, This indicates a uniform sorting score for the same value used in cross-channel sorting. This indicates a dual-channel indicator measurement value with the same unit as the judgment threshold. This indicates the final threshold for determining the positive value of the dual-channel indicator, and the unit is the same as the measured value. (Superscript) Indicates a dimensionless dual-channel set identifier. This indicates an index for dimensionless index types. Indicates the index of the dimensionless detection segment; When any dual-channel over-limit ratio is not less than 1, the corresponding joint detection segment is marked as a dual-channel over-limit segment; when it equals 1, it enters the dual-channel over-limit branch. The module first merges adjacent dual-channel over-limit segments according to the index, and then calculates the union of the segments within the dual-channel index and the unified area of ​​the ordinary channel according to the same rules of material station overlap, connection, or only one valid segment. Any invalid segment of the ordinary station or the invalid segment of the dual-channel station is used as the forced segmentation boundary of its corresponding channel. After the region is formed or expanded, the start and end positions, longitudinal length, maximum over-limit position, and dominant anomaly type are recalculated. Cross-channel sorting only uses the unified sorting score. or ; For states where the dual-channel valid flag is false, the module does not perform dual-channel over-limit ratio calculation, dual-channel over-limit segment merging, or region formation or expansion. Instead, it independently outputs records including detection status, load value, evaluation time, maximum synchronization trigger deviation, and dual-channel retest flag. For states where the dual-channel is valid but the recovery sequence is abnormal, the module does not calculate the recovery-related dual-channel over-limit ratio and does not consider its formation or expansion region. Instead, it independently outputs records including abnormal material location, recovery increment, symbol status, recovery registration verification quantity, and recovery retest flag. For any indicator where the threshold valid, specification valid, or judgment valid flag is false, both the ordinary channel and dual-channel exclude the indicator and it cannot be ranked based on its formation region or dominant type. The final record of the seam exceeding the limit area uses the sample number, scanning batch number, and area number as a joint index. The fields include detection status, load value, evaluation time, start and end material station positions, start and end longitudinal coordinates, longitudinal length, dominant anomaly type, list of composite types, maximum exceeding limit location, exceeding ratio of each indicator's quality limit, unified ranking score of each indicator, registration verification quantity, synchronous trigger deviation, and recovery registration verification quantity. The detection report data package simultaneously reads the task-level anomaly bypass summary table and writes the load anomaly record, instrument calibration record, registration retest record, compensation retest record, threshold sample insufficiency record, specification confirmation record, material station retest record, dual-channel retest record, and recovery retest record item by item. For each record, the task number, old scanning batch number, new scanning batch number, association status, indicator type, and processing status are saved. The output module summarizes the completeness of the status results based on the plan status table, the validity flags of each indicator, and the task-level abnormal bypass summary table: all plan statuses are marked as complete when the required channel data has been generated and all required indicators have been judged; if at least one plan status fails to generate an indicator sequence due to load, calibration, registration, or compensation anomalies, or if at least one indicator is excluded due to threshold, specification, station location, synchronization, or recovery anomalies, it is marked as incomplete or pending confirmation, and the reasons are written item by item; when re-acquisition is successful, the old abnormal records are not deleted, but the processing status of the old records is changed to closed and associated with the new scan batch number. Only after the corresponding status of the new batch has completed all judgments is it counted as complete; The output module generates an index curve and a list of out-of-limit areas along the longitudinal direction of the seam based on the detection report data package. The curve retains the invalid segments of ordinary stations, invalid segments of dual-channel stations, and the breakpoints of dual-channel invalid states and abnormal states of recovery sequences without interpolation. Even if a certain planned state is truncated by S1, S2, or S3 and no index curve exists, the report still generates the corresponding undetermined state entry from the task-level anomaly bypass summary table. Even if there are no valid out-of-limit areas, the integrity status and all causes are written into the data package. This allows each weak area and each incomplete result to be traced back to the original point cloud, load record, calibration version, material station corresponding path, base fabric sample structure, threshold source, and the correlation between new and old scanning batches, forming the terminal detection result of a laser detection method for deformation of tarpaulin heat-sealed seams, and the detection report data package is used as the terminal output of this method.

[0024] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

[0025] This invention combines six-DOF rigid body correction of the fixture reference target, relaxation displacement field compensation of the base fabric reference strips on both sides, material station correspondence, deformation index calculation along the seam, and merging judgment of continuous over-limit sections to form a layered data processing flow for the load-bearing deformation of tarpaulin heat-sealed seams. Specifically, six-DOF rigid body correction eliminates the influence of loading mechanism attitude changes, scanning coordinate drift, and overall sample movement on the measurement results; base fabric relaxation displacement field compensation is used to peel away the gradually changing common displacement generated by the base fabric under load from the seam area displacement; and material station correspondence is applied to ensure that deformation comparisons under different load and unloading states occur at the same material location or corresponding material segment. Through the above algorithm combination, local warping, elongation, gradient abrupt changes, and residual deformation of the seam can be stably extracted from the overall motion and base fabric relaxation background, thereby improving the continuity, positioning accuracy, and quantitative reliability of the detection results along the seam length direction, and enabling the detection results to be directly converted into records of weak seam areas with start and end positions, length, and dominant anomaly types.

[0026] The data processing structure of this invention addresses several technical issues, including the high flexibility of tarpaulin heat-sealed joints, the susceptibility to overall relaxation under load, the easy overlap of local deformation and base fabric background deformation, the difficulty in distinguishing thickness changes and mid-surface warping in single-sided measurements, and the short-term recovery after unloading. Specifically, this invention employs a progressive algorithm structure consisting of a rigid body registration layer, a base fabric compensation layer, a material station correspondence layer, a deformation index layer, and an over-limit judgment layer. Each layer incorporates mechanisms such as outlier removal, long-wavelength relaxation component fitting, longitudinal monotonic constraint matching, station invalid segment segmentation, synchronous deviation verification, and recovery registration verification. This ensures that error sources are identified, removed, or marked before subsequent index calculations. Simultaneously, a thickness channel and a mid-surface warping channel are formed using a laser profilometer, and combined with a multi-time recovery sequence after unloading, short-term recoverable deformation and continuous residual deformation are distinguished and output. Therefore, the algorithm structure of this invention can more effectively adapt to the material characteristics and testing conditions of tarpaulin heat-sealed seams, reduce misjudgments caused by overall displacement, base fabric relaxation, synchronization errors and viscoelastic recovery, and further improve the accuracy and interpretability of seam quality evaluation.

Claims

1. A laser detection method for deformation of heat-sealed seams in tarpaulins, characterized in that, include: S1. The tarpaulin sample with heat-sealed seam is clamped in the loading mechanism with load sensor. In the reference state, at least one detection load state and the unloaded state, the line laser profilometer spanning the seam area and the base fabric reference strips on both sides is used to scan along the longitudinal direction of the seam to obtain the three-dimensional profile point cloud associated with the corresponding detection state. S2. Using the fixture reference target outside the deformation zone, obtain the six-degree-of-freedom rigid body transformation of each detection state relative to the reference state, and obtain the registration point cloud unified to the reference fixture coordinate system. S3. Divide the registration point cloud into the seam area and the base fabric reference strips on both sides. Determine the displacement of the seam area and the base fabric reference strips on both sides based on the registration point cloud of each detection state and the reference state. Fit the base fabric relaxation displacement field that varies along the longitudinal direction of the seam only based on the displacement of the base fabric reference strips on both sides. Interpolate the base fabric relaxation displacement field to the seam area and subtract it from the displacement of the seam area to obtain the seam compensation displacement. S4. Based on the cross-sectional profile characteristics of the reference state, establish the material station correspondence between states under longitudinal monotonic constraints. Based on the material station correspondence and joint compensation displacement, calculate the relative warping, local elongation, deformation gradient and unloading residual deformation of each joint detection segment to obtain the joint deformation index sequence. S5. Compare the deformation index sequence along the seam with the statistical fluctuation threshold and quality limit of the base fabric, merge the continuously exceeding detection segments along the seam, and obtain and output the start and end positions, length and dominant anomaly type of the seam exceeding the limit area.

2. The laser detection method for deformation of tarpaulin heat-sealed seams according to claim 1, characterized in that, S1 includes: Using the leveling preload state where the tarpaulin sample eliminates visible slack and the load sensor reading is stable as the baseline state, at least one detection load state and at least one specified evaluation time state after unloading back to the leveling preload are set sequentially; the load value of each state is collected by the load sensor, and the load value, scan trigger time and state identifier are used as data recording fields of the corresponding three-dimensional contour point cloud.

3. The laser detection method for deformation of tarpaulin heat-sealed seams according to claim 1, characterized in that, S2 includes: At least three non-collinear fixture reference targets are set on the non-deformable component of the loading mechanism, and the six-degree-of-freedom rigid body transformation is determined based on the three-dimensional coordinates of each fixture reference target in the reference state and the current detection state. The coordinate residuals of each fixture reference target after transformation are used as the registration verification quantity. The registration error limit is determined based on the calibration uncertainty of the line laser profilometer and the distribution of coordinate residuals from repeated measurements of the fixture reference targets. When the registration verification quantity exceeds the registration error limit, the corresponding state is marked as a registration anomaly and the calculation of subsequent deformation indices for that state is stopped.

4. The laser detection method for deformation of tarpaulin heat-sealed seams according to claim 1, characterized in that, S3 includes: extracting displacement samples relative to the baseline state from the base fabric reference strips on both sides of the seam area; calculating the median of the displacement samples and the median of the absolute deviation of each displacement sample from the median; identifying and removing displacement samples whose absolute deviation from the median exceeds a preset multiple as outliers, the preset multiple being determined based on the displacement noise distribution obtained from repeated scanning of the seamless base fabric; fitting displacement components containing only longitudinal wavelengths not less than a preset relaxation wavelength lower limit to the retained displacement samples, and interpolating according to the distance from the base fabric reference strips on both sides to the lateral boundary of the seam area, the preset relaxation wavelength lower limit being determined based on the displacement power spectrum of the seamless base fabric under the same loading state, and the points in the seam area not participating in the fitting of the displacement components.

5. The laser detection method for deformation of tarpaulin heat-sealed seams according to claim 1, characterized in that, S4 includes: extracting cross-sectional profile features along the longitudinal direction of the joint in the reference state at preset intervals, and combining the cross-sectional profile features with the longitudinal position to form a material station identifier; under the constraint of maintaining the material station order without reversing, determining the material station correspondence with the objective of minimizing the sum of the differences in cross-sectional profile features between different detection states and the changes in the distance between adjacent material stations; when there are multiple material station correspondence results with the same objective function value, marking the corresponding material station as an invalid match, marking the invalid material station and the detection segment along the joint with the material station as the endpoint as an invalid station segment, and stopping the relative warp, local elongation, deformation gradient, and unloading residual deformation of the invalid station segment. The deformation is calculated, and the material station remeasurement mark is output. For material stations not marked as invalid matches and their adjacent valid material stations, the relative warping is determined by the normal displacement of the joint mid-surface relative to the two sides of the base fabric fitting surface at the same material station. The local elongation is determined by the state change of the distance between adjacent valid material stations. The difference between the relative warping and the local elongation is calculated along the joint longitudinal direction to obtain the relative warping gradient and the local elongation gradient. The relative warping gradient and the local elongation gradient are used as the deformation gradient, wherein the first and last valid material stations adopt the single-sided difference. The difference between the joint compensation displacement at the specified evaluation time after unloading and the reference state is determined as the unloading residual deformation.

6. The laser detection method for deformation of tarpaulin heat-sealed seams according to claim 5, characterized in that, S5 include: The statistical fluctuation thresholds for the base fabric, obtained from the statistical data of the absolute values ​​of the corresponding indicators in the reference strips of the base fabric on both sides, and the positive quality limits given by the product inspection specifications are set for the relative warp, local elongation, relative warp gradient, local elongation gradient, and unloading residual deformation. Sections along the seam where the absolute value of the same indicator continuously exceeds the corresponding statistical fluctuation threshold and quality limit are marked as over-limit sections. Adjacent over-limit sections with an interval of no more than one section along the seam are merged. Invalid sections at the specified stations serve as the dividing boundary for the continuity of over-limit sections and are not included in the merging. The material station retesting markers and corresponding material stations are recorded separately. The ratio of the absolute value of each over-limit indicator to the corresponding quality limit is calculated as the over-limit ratio and arranged in descending order of value. The anomaly type corresponding to the top-ranked over-limit indicator is determined as the dominant anomaly type. When multiple over-limit ratios are tied for first place, they are determined as a composite dominant anomaly type. The starting and ending material stations, longitudinal length, dominant anomaly type, and maximum over-limit location of the seam over-limit area are recorded.

7. The laser detection method for deformation of tarpaulin heat-sealed seams according to claim 2, characterized in that, In step S1, opposing line laser profilometers are set on the same measuring frame, located on the upper and lower sides of the tarpaulin sample and triggered synchronously, to obtain the three-dimensional profile point cloud of the upper surface and the three-dimensional profile point cloud of the lower surface, respectively. The normal distance between the upper and lower surfaces at the same material station is used as the input of the thickness channel, and the median of the normal coordinates of the upper and lower surfaces is used as the input of the mid-surface warping channel. The synchronization deviation limit is determined according to the time difference distribution of repeated synchronous triggering of the opposing line laser profilometer. When the synchronous triggering deviation exceeds the synchronization deviation limit, the scan is repeated. If the synchronous deviation limit is still exceeded after the rescan, the corresponding detection state is marked as dual-channel invalid, and the calculation of the thickness channel and mid-surface warping channel in steps S3 to S5 is stopped. A dual-channel retest record containing the detection state, synchronous triggering deviation, and dual-channel retest mark is output independently.

8. The laser detection method for deformation of tarpaulin heat-sealed seams according to claim 7, characterized in that, In step S3, for detection states not marked as invalid dual-channel, rigid body correction of the fixture reference target and relaxation displacement field compensation of the base fabric reference strips on both sides are performed for the thickness channel and the mid-surface warping channel, respectively; the change in the distance between the upper and lower surfaces of the seam area relative to the reference state is determined as the thickness change, and the normal displacement of the median of the upper and lower surfaces of the seam area relative to the fitted mid-surface of the base fabric reference strips on both sides is determined as the mid-surface warping, and the thickness change gradient and mid-surface warping gradient are calculated along the longitudinal direction of the seam, respectively; for detection states marked as invalid dual-channel, the rigid body correction, relaxation displacement field compensation, and calculation of the thickness change, mid-surface warping, and corresponding gradients in this step are not performed, and the dual-channel retest record is maintained.

9. The laser detection method for deformation of tarpaulin heat-sealed seams according to claim 8, characterized in that, In step S4, for detection states not marked as dual-channel invalid, the scan is repeated at multiple preset recovery times after unloading back to the leveling preload, and recovery sequences for the thickness channel and mid-surface warping channel are formed according to the material station. The thickness change and mid-surface warpage attenuation between each adjacent preset recovery time from the first preset recovery time after unloading to the specified evaluation time are accumulated to obtain short-term recoverable thickness and short-term recoverable mid-surface warpage items. The difference between the thickness change at the first preset recovery time after unloading and the reference state is subtracted from the short-term recoverable thickness item to obtain the residual thickness item. The difference between the mid-surface warpage at the first preset recovery time after unloading and the reference state is subtracted from the short-term recoverable mid-surface warpage item to obtain the residual mid-surface warpage item. The coordinate residual of the fixture reference target at each recovery time after a six-degree-of-freedom rigid body transformation is used as the recovery registration verification quantity. Based on the calibration uncertainty of the line laser profilometer and the coordinate residual distribution of repeated measurements of the fixture reference target, the recovery registration error limit is determined. When the recovery registration verification quantity exceeds the recovery registration error limit at any preset recovery time, the corresponding recovery sequence is marked as an abnormal recovery sequence. The calculation of the short-term recoverable thickness item, the short-term recoverable mid-surface warpage item, the residual thickness item, and the residual mid-surface warpage item of the recovery sequence is stopped, and a retest flag is output. For detection states marked as dual-channel invalid, recovery sequences for the thickness channel and the mid-surface warpage channel are not formed, and the dual-channel retest record is maintained.

10. The laser detection method for deformation of tarpaulin heat-sealed seams according to claim 9, characterized in that, In step S5, the statistical fluctuation threshold of the base fabric is taken as the preset empirical quantile of the absolute value distribution of the corresponding deformation index in the reference strips of the base fabric on both sides, and the quality limit is taken as the positive absolute allowable deviation limit specified in the inspection specification of the tarpaulin to be tested; when the statistical fluctuation threshold of the base fabric is not less than the quality limit, the statistical fluctuation threshold of the base fabric is determined as the judgment threshold, otherwise the quality limit is determined as the judgment threshold; for the detection state not marked as dual-channel invalid and the recovery sequence not marked as recovery sequence abnormal, the ratio of the absolute value of the thickness change, thickness change gradient, mid-surface warpage, mid-surface warpage gradient, short-term recoverable thickness item, short-term recoverable mid-surface warpage item, residual thickness item, and residual mid-surface warpage item to the corresponding judgment threshold is calculated along the longitudinal direction of the seam as the over-limit ratio, and any detection segment along the seam with an over-limit ratio not less than one is marked as a dual-channel over-limit segment, adjacent dual-channel over-limit segments are merged, and the merged result forms the seam over-limit area or expands the existing seam over-limit area; in each Within the joint exceeding the limit area, the exceeding ratios are arranged in descending order of value. The anomaly type corresponding to the deformation index at the top of the ranking is determined as the dominant anomaly type. When multiple exceeding ratios are tied for first place, the anomaly types corresponding to the tied indices are jointly determined as the composite dominant anomaly type. A joint exceeding the limit area record containing detection status, load value, evaluation time, and recovery registration verification quantity is output. For detection status marked as dual-channel invalid, the exceeding ratio calculation, dual-channel exceeding segment merging, and the formation or expansion of the joint exceeding the limit area are not performed. A dual-channel retest record containing detection status, load value, evaluation time, synchronization trigger deviation, and dual-channel retest flag is output independently. For recovery sequences not marked as dual-channel invalid but marked as recovery sequence anomalies, the exceeding ratio calculation, dual-channel exceeding segment merging, and the formation or expansion of the joint exceeding the limit area are not performed. A recovery retest record containing detection status, load value, evaluation time, recovery registration verification quantity, and retest flag is output independently.

Images