Method and system for detecting quality of irregularly shaped electrical interior parts

Abstract

The present application relates to the technical field of precision measurement, and more particularly to a method and system for quality detection of irregular electrical inner parts. Through multi-view scanning combined with real-time pose feedback of a positioning tracking device, three-dimensional point cloud data of each occluded area and complex surface of the irregular electrical inner part can be completely collected without contacting the surface of the workpiece, effectively eliminating the scanning blind area under a single view. On this basis, by registering and splicing each view point cloud data based on a unified spatial pose reference to generate a complete three-dimensional point cloud, and then directly calculating the geometric parameters of the workpiece through geometric element fitting and automatically comparing with the preset tolerance standard, the whole process automation from data collection to quality determination is realized, and the precision and efficiency of the detection of irregular electrical inner parts are significantly improved.

Description

Technical Field

[0001] This invention relates to the field of precision measurement technology, and in particular to a method and system for quality inspection of irregularly shaped electrical internal components. Background Technology

[0002] Currently, the dimensional inspection of electrical internal components mainly relies on manual caliper measurement, coordinate measuring machines (CMMs), and ordinary two-dimensional vision inspection. However, manual caliper measurement is inefficient, has poor repeatability, and the contact measurement method can easily scratch the workpiece surface; although CMMs can achieve micron-level accuracy, they are expensive and have low inspection efficiency, making them unsuitable for batch online inspection needs; ordinary two-dimensional vision inspection cannot obtain the spatial three-dimensional information of the workpiece and has inherent measurement blind spots for complex structures such as deep grooves, curved surfaces, and irregular contours.

[0003] With the rapid development of 3D laser scanning and machine vision technologies, point cloud-based 3D measurement methods have provided a new technical approach for the geometric dimension inspection of irregularly shaped parts. However, the following core technical challenges remain: the surface morphology of irregularly shaped workpieces is complex, with various geometric features such as deep grooves, chamfers, inner surfaces, and curved surfaces. Single-view scanning cannot cover the entire surface of the workpiece, and occlusion areas and scanning blind spots result in incomplete point cloud data. These problems make it difficult to directly use the acquired 3D point clouds for high-precision geometric dimension calculation, and it is impossible to achieve automated extraction of the complete contour and accurate dimension calculation of complex workpieces. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method and system for quality inspection of irregularly shaped electrical internal components, so as to solve the above-mentioned technical problem.

[0005] The technical solution of this invention to solve the above-mentioned technical problems is as follows: A method for quality inspection of irregularly shaped electrical internal components, comprising: controlling a scanning device to perform multi-view scanning of the workpiece to be inspected according to a preset scanning strategy to obtain local three-dimensional point cloud data of the workpiece under multiple views, and simultaneously acquiring the spatial pose data of the scanning device in real time through a positioning and tracking device; based on the spatial pose data, registering and stitching the local three-dimensional point cloud data of the workpiece under each view to generate a complete three-dimensional point cloud; performing geometric element fitting on the complete three-dimensional point cloud to calculate the geometric parameters of the workpiece to be inspected; comparing the geometric parameters with a preset tolerance standard, and generating a quality judgment result of the workpiece to be inspected based on the comparison result.

[0006] The beneficial effects of this invention are as follows: By combining multi-view scanning with real-time pose feedback from a positioning and tracking device, it is possible to completely acquire three-dimensional point cloud data of various obstructed areas and complex curved surfaces of irregularly shaped electrical internal components without contacting the workpiece surface, effectively eliminating scanning blind spots under a single viewpoint. Based on this, by registering and stitching the point cloud data from each viewpoint according to a unified spatial pose reference to generate a complete three-dimensional point cloud, and then directly calculating the geometric parameters of the workpiece through geometric element fitting and automatically comparing them with preset tolerance standards, the entire process from data acquisition to quality judgment is automated, significantly improving the accuracy and efficiency of irregularly shaped electrical internal component inspection.

[0007] Based on the above technical solution, the present invention can be further improved as follows.

[0008] Furthermore, the step of registering and stitching the local 3D point cloud data of the workpiece from each viewpoint based on the spatial pose data to generate a complete 3D point cloud includes: converting the local 3D point cloud data of the workpiece from each viewpoint to a unified global coordinate system based on the spatial pose data, and performing initial pose alignment; performing iterative nearest-point fine registration processing on each local 3D point cloud data of the workpiece after initial pose alignment; and performing data fusion on each local 3D point cloud data of the workpiece after fine registration in the unified global coordinate system, removing redundant data in overlapping areas, and performing smooth transition processing on the stitched area to generate a complete 3D point cloud.

[0009] Furthermore, the step of comparing the geometric parameters with a preset tolerance standard and generating a quality judgment result for the workpiece to be inspected based on the comparison result includes: comparing each value in the geometric parameters with the corresponding allowable deviation range in the preset tolerance standard item by item to obtain a comparison result; when the comparison result shows that all values ​​are within the corresponding allowable deviation range, the quality judgment result of the workpiece to be inspected is determined to be qualified; when the comparison result shows that any value exceeds the corresponding allowable deviation range, the quality judgment result of the workpiece to be inspected is determined to be unqualified.

[0010] Furthermore, the step of fitting geometric elements to the complete 3D point cloud to calculate the geometric parameters of the workpiece to be inspected includes: performing denoising processing, coordinate unification calibration processing, and point cloud simplification processing on the complete 3D point cloud to obtain a preprocessed 3D point cloud; segmenting the preprocessed 3D point cloud into feature regions to extract multiple feature regions; and fitting geometric elements to the point cloud in each feature region to calculate the geometric parameters of the workpiece to be inspected. The geometric parameters include at least one of length, width, height, hole diameter, groove width, groove depth, wall thickness, parallelism, perpendicularity, coaxiality, symmetry, and contour.

[0011] Furthermore, the aperture is calculated as follows: extracting the annular point cloud of the inner wall of the hole from the point cloud in the feature region; performing spatial circle fitting on the annular point cloud of the inner wall of the hole, and determining the central spatial axis of the circular hole based on the fitting result; taking cross sections at different depths inside the hole along the direction of the central spatial axis of the circular hole; performing spatial circle fitting on the point cloud on each cross section to determine the aperture value corresponding to each cross section; calculating the average value of the aperture value corresponding to each cross section to obtain the aperture.

[0012] Furthermore, the groove width is calculated as follows: point cloud data of the two side walls of the groove are extracted from the point cloud in the feature region; plane fitting is performed on the point cloud data of the two side walls of the groove to obtain the planes of the two side walls of the groove; multiple cross-sectional positions are selected in the extension direction of the groove, and for each cross-sectional position, the vertical distance between the planes of the two side walls of the groove at the cross-sectional position is calculated as the groove width sample value at the cross-sectional position; the average value of all groove width sample values ​​is calculated to obtain the groove width.

[0013] Furthermore, the trench depth is calculated as follows: point cloud data of the trench opening edge and point cloud data of the trench bottom are extracted from the point cloud within the feature region; plane fitting is performed on the point cloud data of the trench opening edge to obtain the reference plane of the top surface of the trench opening; plane fitting is performed on the point cloud data of the trench bottom to obtain the reference plane of the bottom surface of the trench; using the reference plane of the top surface of the trench opening as a reference, the vertical sinking distance of the reference plane of the bottom surface of the trench relative to the reference plane of the top surface of the trench opening is calculated as the trench depth.

[0014] Furthermore, the contour degree is calculated as follows: extract the measured contour point cloud data of the workpiece from the point cloud in the feature region, and generate the actual contour of the workpiece based on the measured contour point cloud data of the workpiece; obtain the theoretical contour model corresponding to the workpiece to be detected; register and align the actual contour of the workpiece with the theoretical contour model; calculate the normal deviation of each contour point on the aligned actual contour of the workpiece relative to the corresponding contour point on the theoretical contour model, and obtain the normal deviation value of each contour point; calculate the contour degree based on the normal deviation value of each contour point.

[0015] To address the aforementioned technical problems, the present invention also provides a quality inspection system for irregularly shaped electrical internal components, comprising: The scanning device is used to perform multi-view scanning of the workpiece to be inspected according to a preset scanning strategy, and to obtain local three-dimensional point cloud data of the workpiece from multiple perspectives. A workpiece carrying device is used to carry the workpiece to be inspected and to drive the workpiece to be inspected to change its posture during the scanning process; A positioning and tracking device is used to acquire the spatial pose data of the scanning device in real time; A control and processing device is used to control the scanning equipment and the workpiece carrying device to work together and to execute the quality inspection method for irregularly shaped electrical internal components as described above.

[0016] Furthermore, the system also includes a loading and unloading device, and the control and processing device is also used to control the loading and unloading device to sort the workpiece to be inspected into a preset qualified area or unqualified area according to the quality judgment result. Attached Figure Description

[0017] Figure 1 This is a flowchart of the quality inspection method for irregularly shaped electrical internal components according to the present invention; Figure 2 This is a schematic diagram of the quality inspection system for irregularly shaped electrical internal components according to the present invention; Figure 3 This is a schematic diagram of the gripper in Embodiment 2; Figure 4 This is a schematic diagram of the workpiece bearing device in Embodiment 2; Figure 5 This is a schematic diagram of the installation of the positioning and tracking device in Embodiment 2. Figure 6 This is a schematic diagram of the loading trolley in Embodiment 2; Figure 7 This is a schematic diagram of the camera mount assembly in Embodiment 2.

[0018] The attached diagram lists the components represented by each number as follows: 1. Scanning equipment; 2. Workpiece carrying device; 3. Positioning and tracking equipment; 4. Measuring robot; 5. Unloading trolley; 6. Loading and unloading robot; 7. Operation screen; 8. Industrial control all-in-one computer; 9. Measuring worktable; 10. Calibration plate; 11. Loading trolley; 12. Camera frame assembly; 13. Workpiece to be inspected; 14. Gripper; 15. Gripper connecting block; 16. Gripper servo motor; 17. Robot connecting seat; 18. Support fixture; 19. Servo motor; 20. Fixture base; 21. Rotating base plate; 22. 3D camera; 23. Universal ball joint bracket; 24. Electric swing table; 25. Column; 26. Elevation frame; 27. Camera mounting bracket; 28. Positioning fixture; 29. ​​Optical tracker. Detailed Implementation

[0019] The principles and features of the present invention are described below. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.

[0020] Example 1 like Figure 1 As shown, this embodiment provides a method for quality inspection of irregularly shaped electrical internal components, including: S101. Control the scanning device 1 to perform multi-view scanning of the workpiece 13 to be inspected according to the preset scanning strategy, so as to obtain local three-dimensional point cloud data of the workpiece under multiple views, and at the same time, the spatial pose data of the scanning device 1 is obtained in real time through the positioning and tracking device 3. S102. Based on spatial pose data, register and stitch together the local three-dimensional point cloud data of the workpiece from various perspectives to generate a complete three-dimensional point cloud. S103. Perform geometric element fitting on the complete three-dimensional point cloud and calculate the geometric parameters of the workpiece 13 to be inspected. S104. Compare the geometric parameters with the preset tolerance standards, and generate the quality judgment result of the workpiece 13 to be inspected based on the comparison result.

[0021] In this embodiment, multi-view 3D scanning employs a 1- to 4-round graded scanning process, with each round of scanning binding a dedicated scanning path and a table rotation angle range (adjustable from 0° to 360°). Based on the workpiece size, morphological complexity, surface features, and accuracy requirements, the number of scanning rounds is automatically determined according to preset thresholds and binding rules. Following a strict matching principle, workpiece feature tags are automatically associated with the formula library, matching 1 to 4 rounds of dedicated scanning formulas and loading the corresponding scanning parameters and paths.

[0022] After all four rounds of scanning are completed, the local point clouds from each viewpoint are registered and stitched together. Optionally, in this embodiment, based on spatial pose data, the local 3D point cloud data of the workpiece from each viewpoint is registered and stitched together to generate a complete 3D point cloud. This includes: based on spatial pose data, converting the local 3D point cloud data of the workpiece from each viewpoint to a unified global coordinate system and performing initial pose alignment; performing iterative nearest-point fine registration processing on each local 3D point cloud data of the workpiece after initial pose alignment; and fusing the finely registered local 3D point cloud data of the workpiece in a unified global coordinate system, removing redundant data in overlapping areas, and performing smooth transition processing on the stitched area to generate a complete 3D point cloud.

[0023] After generating a complete 3D point cloud, the system automatically verifies the coverage integrity, coordinate accuracy, and surface fit of the point cloud. If the verification is successful, standardized 3D point cloud data is output; if the verification fails, the system calls the corresponding formula parameters to perform rescanning for the defective areas.

[0024] Optionally, in this embodiment, geometric element fitting is performed on the complete 3D point cloud to calculate the geometric parameters of the workpiece 13 to be inspected. This includes: performing noise reduction, coordinate unification calibration, and point cloud simplification on the complete 3D point cloud to obtain a preprocessed 3D point cloud; segmenting the preprocessed 3D point cloud into feature regions to extract multiple feature regions; and performing geometric element fitting on the point cloud in each feature region to calculate the geometric parameters of the workpiece 13 to be inspected. The geometric parameters include at least one of length, width, height, hole diameter, groove width, groove depth, wall thickness, parallelism, perpendicularity, coaxiality, symmetry, and contour.

[0025] Specifically, the noise reduction process includes: based on filtering and neighborhood noise reduction, automatically removing discrete noise points, floating noise points, and redundant edge points generated during the scanning process, retaining the effective point cloud of the workpiece's true surface, and eliminating data interference caused by the acquisition environment and equipment vibration.

[0026] Point cloud simplification includes: while preserving key features such as workpiece contours, holes, grooves, and curved surfaces, adaptive simplification of high-density point clouds is performed to unify the point cloud distribution density and avoid fitting deviations caused by excessively dense or sparse local point clouds.

[0027] The coordinate unification calibration process includes: reusing the global positioning reference of the multi-round scanning tracker, performing secondary registration and calibration of the point cloud coordinates, unifying the global coordinate system of the workpiece, locking the coordinate parameters of the reference plane and reference axis, and providing a unified reference for subsequent dimension and tolerance calculations.

[0028] By employing curvature discrimination, boundary recognition, and feature clustering, the system automatically segments different feature regions of a workpiece, such as planes, cylinders, holes, grooves, and thin-walled structures. Specifically, based on the curvature, normal, and neighborhood distribution characteristics of each point in the point cloud, the system automatically segments the complete 3D point cloud using region growing or RANSAC model fitting methods, identifying and extracting different types of feature regions, including planar regions, cylindrical regions, hole regions, groove regions, and thin-walled regions. The system automatically matches the appropriate geometric element fitting method according to the geometric feature type of each region; for example, it fits a reference plane to planar regions, a reference axis to cylindrical regions, a spatial circle to hole regions, and sidewall planes and bottom surfaces to groove regions. This achieves accurate segmentation of the region to be solved and avoids calculation errors caused by cross-feature fitting.

[0029] Optionally, in an embodiment, the aperture is calculated as follows: extracting the annular point cloud of the inner wall of the hole from the point cloud in the feature region; performing spatial circle fitting on the annular point cloud of the inner wall of the hole, and determining the central spatial axis of the circular hole based on the fitting result; taking cross sections at different depths inside the hole along the direction of the central spatial axis of the circular hole; performing spatial circle fitting on the point cloud on each cross section to determine the aperture value corresponding to each cross section; calculating the average value of the aperture value corresponding to each cross section to obtain the aperture.

[0030] Specifically, the process involves extracting the annular point cloud of the inner wall of the hole from the point cloud of the segmented hole feature region, and removing outliers corresponding to the hole opening chamfer, wear, and impurities. Then, spatial circle fitting is performed on the annular point cloud of the inner wall of the hole, and the spatial axis of the circular hole's center is determined based on the fitting result. Next, cross-sections are taken at different depths within the hole along the direction of the circular hole's center spatial axis, and spatial circle fitting is performed on the point cloud of each cross-section to obtain the corresponding hole diameter value. Hole diameter values ​​of cross-sections with deviations exceeding a preset threshold are discarded. Finally, the average value of the remaining cross-section hole diameter values ​​is calculated and output as the average hole diameter parameter. This method is adaptable to different hole types, such as through holes and blind holes.

[0031] Optionally, in an embodiment, the slot width is calculated as follows: point cloud data of the two side walls of the slot are extracted from the point cloud in the feature region; plane fitting is performed on the point cloud data of the two side walls of the slot to obtain the planes of the two side walls of the slot; multiple cross-sectional positions are selected in the extension direction of the slot, and for each cross-sectional position, the vertical distance between the planes of the two side walls of the slot at the cross-sectional position is calculated as the slot width sample value at the cross-sectional position; the average value of all slot width sample values ​​is calculated to obtain the slot width.

[0032] Specifically, point cloud data for the two side walls of the tank are extracted from the point cloud of the segmented feature region of the tank. Then, planar fitting is performed on the point cloud data of each side wall to obtain the planes of the two side walls. Next, multiple cross-sectional locations are selected along the extension direction of the tank. For each cross-sectional location, the vertical distance between the planes of the two side walls at that cross-section is calculated as the sampled value of the tank width at that cross-section. Finally, the average value of all the sampled tank width values ​​is calculated as the tank width dimension to eliminate dimensional fluctuations caused by local deformation of the tank.

[0033] Optionally, in the embodiment, the trench depth is calculated as follows: extracting trench edge point cloud data and trench bottom point cloud data from the point cloud in the feature region; performing plane fitting on the trench edge point cloud data to obtain the trench top reference plane; performing plane fitting on the trench bottom point cloud data to obtain the trench bottom reference plane; and using the trench top reference plane as a reference, calculating the vertical sinking distance of the trench bottom reference plane relative to the trench top reference plane, which is taken as the trench depth.

[0034] Specifically, point cloud data of the trench opening edge and trench bottom are extracted from the point cloud of the segmented feature region of the trench. Plane fitting is performed on the trench opening edge point cloud data to obtain the reference plane for the top surface of the trench opening. Similarly, plane fitting is performed on the trench bottom point cloud data to obtain the reference plane for the trench bottom. Using the reference plane for the top surface of the trench opening as a reference, the vertical downward distance of the reference plane for the trench bottom relative to the reference plane for the top surface of the trench opening is calculated as the trench depth dimension.

[0035] Optionally, in the embodiment, the contour degree is calculated as follows: extract the measured contour point cloud data of the workpiece from the point cloud in the feature region, and generate the actual contour of the workpiece based on the measured contour point cloud data of the workpiece; obtain the theoretical contour model corresponding to the workpiece 13 to be detected; register and align the actual contour of the workpiece with the theoretical contour model; calculate the normal deviation of each contour point on the aligned actual contour of the workpiece relative to the corresponding contour point on the theoretical contour model, and obtain the normal deviation value of each contour point; calculate the contour degree based on the normal deviation value of each contour point.

[0036] Specifically, measured contour point cloud data of the workpiece is extracted from the point cloud of the segmented curved surface feature region or contour feature region, and the actual contour of the workpiece is generated based on the measured contour point cloud data. Then, the theoretical contour model (e.g., the standard theoretical CAD contour model of the workpiece) corresponding to the workpiece 13 to be inspected is obtained, and the actual contour of the workpiece is registered and aligned with the theoretical contour model. The normal deviation of each contour point on the aligned actual contour of the workpiece relative to the corresponding contour point on the theoretical contour model is calculated point by point to obtain the normal deviation value of each contour point. Finally, based on the normal deviation values ​​of each contour point, the maximum positive deviation value, the maximum negative deviation value, and the total contour error band are determined as the contour degree parameters of the workpiece 13 to be inspected, and the contour deviation distribution cloud map is output.

[0037] For length, width, and height dimensions, after fitting the workpiece reference plane, the extreme coordinates of the effective feature point cloud along the X, Y, and Z axes are extracted. The workpiece's six outer reference planes (top, bottom, left, right, front, and back) are fitted using the least squares method to construct a minimum outer reference cube model. The maximum and minimum coordinate differences along the X-axis are calculated for the length dimension, the Y-axis for the width dimension, and the Z-axis for the height dimension. Furthermore, during the calculation process, abnormal extreme points corresponding to edge deformation and burrs are automatically removed. Multi-region mean fitting correction is used to avoid random errors in single-section calculations, outputting high-precision, globally averaged dimensional parameters.

[0038] For thin-walled and shell-type workpieces, a spatial symmetric offset fitting method is used to calculate the wall thickness. The corresponding feature point clouds of the inner and outer surfaces of the workpiece are automatically paired, and the reference surfaces of the inner and outer surfaces are fitted. Based on the surface direction, the vertical normal distance between the inner and outer surfaces is calculated. By uniformly sampling multi-point wall thickness data across the entire domain, abnormal data caused by local unevenness and machining textures are removed, generating global wall thickness distribution data and average wall thickness parameters for the workpiece. Multi-dimensional results, including single-point wall thickness, minimum / maximum wall thickness of the region, and average wall thickness, can be output to meet the high-precision inspection requirements of thin-walled workpieces.

[0039] For geometrical tolerance parameters such as parallelism, perpendicularity, coaxiality, and symmetry, based on the core logic of datum-first fitting, tolerance zone adaptive matching, and global deviation quantification, and relying on calibrated datum elements, the position and shape deviations of the measured elements are quantified to achieve accurate calculation of tolerance parameters. Specific calculation methods can refer to conventional geometrical tolerance inspection methods in this field.

[0040] In this embodiment, parallelism calculation is achieved through the following steps: First, a reference plane / axis is selected from the drawing; then, the optimal reference surface of the reference feature is fitted using the minimum condition method. Next, the measured features of the target plane / axis are fitted to construct a global fitted surface for the measured feature. The vertical distance deviation of all valid feature points on the measured feature relative to the reference feature is calculated, and the maximum and minimum values ​​among all deviations are extracted; the difference between these two values ​​is the parallelism error. A parallelism tolerance zone is automatically generated, and it is determined whether the measured deviation is within the allowable tolerance range, outputting a quantified parallelism value.

[0041] Perpendicularity is calculated as follows: Using a standard reference plane or axis as a reference, an orthogonal coordinate system is constructed to determine the theoretical perpendicular direction. The spatial orientation of the measured plane or axis of the element is fitted, and the angular deviation between the measured element and the theoretical perpendicular direction is calculated. Using a spatial vector orthogonal projection algorithm, the vertical deviation of all points across the measured element is quantified. The maximum vertical deviation value across the entire area is extracted as the measured verticality error, completing the verticality parameter calculation and tolerance determination, and accurately identifying morphological deviations such as workpiece tilting and swaying.

[0042] The coaxiality is calculated as follows: First, the standard center axis of the workpiece's reference hole / reference shaft is fitted as the theoretical reference center; then, the measured center points of different sections of the rotating body are fitted layer by layer to obtain the actual center coordinates of each section. The radial offset of all measured center points relative to the reference axis is calculated, and twice the maximum radial offset value over the entire domain is calculated as the coaxiality error through spatial radial deviation synthesis. Multi-layer cross-sectional global fitting is used throughout the process to avoid single-point calculation deviations and accurately reflect the overall eccentricity and bending deformation state of the rotating body.

[0043] The symmetry is specifically calculated as follows: Using the workpiece's reference center plane and reference center axis as symmetry references, a theoretical symmetry center reference element is constructed. The global feature point cloud of the measured symmetrical structure is extracted, and the offset distance of corresponding points on both sides of the measured element relative to the symmetry reference is calculated. The deviation difference between the symmetrical points on both sides is matched. Through global symmetry deviation comparison, the maximum symmetry deviation value is selected as the symmetry measurement error, quantifying the offset and asymmetric deformation degree of the workpiece's symmetrical structure, and adapting to the detection of structures such as keyways, bosses, and symmetrical holes.

[0044] Optionally, in an embodiment, the geometric parameters are compared with a preset tolerance standard, and a quality judgment result of the workpiece 13 to be inspected is generated based on the comparison result. This includes: comparing each value in the geometric parameters with the corresponding allowable deviation range in the preset tolerance standard item by item to obtain a comparison result; when the comparison result shows that all values ​​are within the corresponding allowable deviation range, the quality judgment result of the workpiece 13 to be inspected is determined to be qualified; when the comparison result shows that any value exceeds the corresponding allowable deviation range, the quality judgment result of the workpiece 13 to be inspected is determined to be unqualified.

[0045] In this embodiment, a qualified product refers to a workpiece whose measured calculated values ​​of all controlled geometric dimensional parameters (including length, width, height, hole diameter, groove width, groove depth, and wall thickness) and controlled geometric tolerance parameters (including parallelism, perpendicularity, coaxiality, symmetry, and contour) meet the requirements of the preset tolerance standard. In addition, it is also required that the entire scanning data be complete, the calculation results stable, and that there be no judgment deviation caused by missing point clouds or abnormal fitting. A non-qualified product refers to a workpiece whose measured calculated value of any controlled geometric parameter or geometric tolerance parameter does not meet the requirements of the preset tolerance standard. For non-qualified products, the judgment criterion is the presence of an out-of-tolerance item, rather than overall structural deformation or overall positional offset. Such workpieces can usually be corrected to meet the usage standard through subsequent rework processes such as grinding, trimming, and fine-tuning.

[0046] After the quality judgment is completed, the sorting instruction is automatically triggered. Based on the judgment result, the workpiece is picked up and sent to the corresponding area: qualified products are returned to the loading trolley 11, and unqualified products are placed in the unqualified product trolley, thereby realizing the automatic sorting of qualified and unqualified products.

[0047] Upon completion of the measurement, an inspection report is automatically generated. The report includes the workpiece type, inspection time, name of each inspection item, measured value, tolerance range, judgment result, and deviation contour plot. The inspection report supports retrieval, preview, and export in multiple formats. Simultaneously, the measurement data can be synchronized to the MES system in real time via communication protocols, enabling traceable management of quality data.

[0048] Example 2 like Figure 2 As shown, this embodiment provides a quality inspection system for irregularly shaped electrical internal components, including: Scanning device 1 is used to perform multi-view scanning of the workpiece 13 to be inspected according to a preset scanning strategy, and to obtain local three-dimensional point cloud data of the workpiece under multiple views. The workpiece carrying device 2 is used to carry the workpiece 13 to be inspected and to drive the workpiece 13 to change its posture during the scanning process. Positioning and tracking device 3 is used to acquire the spatial pose data of scanning device 1 in real time; A control and processing device is used to control the coordinated operation of the scanning device 1 and the workpiece carrying device 2, and to execute the quality inspection method for irregularly shaped electrical internal components as described in Example 1.

[0049] In this embodiment, the positioning and tracking device 3 mainly includes an optical tracker 29. The workpiece bearing device 2 is a rotating measuring table. The scanning device 1 is a 3D scanner, generally a structured light / laser scanner, which works in conjunction with the optical tracker 29 to achieve global coordinate unification. The control and processing device uses an industrial control computer (referred to as an industrial control all-in-one computer 8) to realize functions such as point cloud data acquisition, 3D reconstruction, dimension calculation, and quality judgment.

[0050] Optionally, in an embodiment, the system further includes a loading and unloading device, and the control and processing device is also used to control the loading and unloading device to sort the workpiece 13 to be inspected into a preset qualified area or unqualified area according to the quality judgment result.

[0051] Specifically, the loading and unloading device uses a loading and unloading robot 6. This system also includes a measuring workbench 9, a camera mount assembly 12, a loading and unloading trolley assembly, a measuring robot 4, and an operation screen 7. In the pre-operation preparation stage, operators use an integrated industrial control computer 8 and the operation screen 7 to digitally configure the measurement task, including pre-setting the geometric design parameters and tolerance standards of each measured part, and supporting manual input and modification of the geometric design parameters and tolerance standards of the workpiece 13 to be inspected.

[0052] The loading and unloading trolley assembly includes a loading trolley 11 and an unloading trolley 5. The loading trolley 11 is manually pushed into the designated area (a positioning mechanism is provided between the trolley and the measuring worktable 9) and then secured. The camera mount assembly 12 includes a camera mounting bracket 27 and a 3D camera 22. The camera mounting bracket 27 has a portal frame structure, and the 3D camera 22 is mounted on the camera mounting bracket 27, such as... Figure 7 As shown. Guided by the 3D camera 22, the loading and unloading robot 6 automatically picks up the workpiece 13 to be inspected and places it on the measuring table fixture.

[0053] After receiving the type information of the workpiece 13 to be inspected and the relevant dimensional information entered earlier, the measuring robot 4 automatically selects a scanning path and, in conjunction with the equipped 3D scanner, completes the 3D data scanning of the workpiece 13. During the measurement process, the measuring table 9 slowly rotates the workpiece to ensure the integrity of the measured workpiece's dimensions and structure.

[0054] The system combines tolerance and other data to determine whether the workpiece 13 to be inspected is qualified. Based on the measurement results, the sorting command is automatically triggered, and the loading and unloading robot 6 accurately grabs the workpiece 13 to be inspected. Qualified products are returned to the loading trolley 11, and unqualified products are placed in the unqualified trolley. After the measurement is completed, the workpiece can be pushed away manually.

[0055] The measurement process and results are displayed in real time on the operation screen of the industrial control all-in-one machine 8. After the measurement is completed, multi-format reports are automatically generated according to customer requirements. The measurement data can be synchronized to the MES system in real time through the protocol.

[0056] like Figure 3 As shown, in this embodiment, the end of the loading / unloading robot 6 is connected to a gripper assembly. The gripper assembly mainly includes a gripper 14, a gripper connecting block 15, a gripper servo motor 16, and a robot connecting seat 17. The gripper 14 is used to hold the component to be inspected, the gripper connecting block 15 is used to connect the gripper 14 and the gripper servo motor 16, the gripper servo motor 16 is used to drive the gripper 14, and the robot connecting seat 17 is used to connect with the loading / unloading robot 6.

[0057] For workpieces of different sizes and weights, appropriate clamping forces, clamping strokes, and clamping speeds can be set to ensure clamping stability for different workpieces. Various types of grippers 14 are designed according to the type of workpiece being measured. When changing the type of workpiece, the corresponding matching gripper 14 needs to be replaced to meet the requirements. Silicone pads are attached to the contact surfaces between the grippers 14 and the workpiece to prevent direct contact and damage to the workpiece.

[0058] like Figure 4 As shown, the workpiece carrying device 2 mainly includes a rotating platform, a servo motor 19, a rotating base plate 21, a fixture base 20, and a supporting fixture 18. The loading / unloading robot 6 places the gripped workpiece to be tested onto the fixed position of the supporting fixture 18 for inspection. During the measurement process, the servo motor 19 drives the workpiece to rotate slowly through the rotating platform to detect the dimensions of the workpiece at various positions.

[0059] Specifically, various types of support fixtures 18 are designed according to the type of the part being tested. When the type of part being tested is changed, the corresponding matching support fixture 18 needs to be replaced to meet the positioning requirements. The fixture base 20 is designed with multiple slots for the support fixture 18. Different types of parts being tested are inserted into their corresponding slots to achieve positioning and support. The fixture supports and slots are universal and replaceable, facilitating disassembly and replacement by workers. The surfaces of the fixture supports that contact the workpiece are covered with silicone pads to prevent direct contact between the fixture and the workpiece from causing damage.

[0060] The loading trolley 11 and unloading trolley 5 are mainly used to facilitate manual handling of workpieces and to assist the loading and unloading robot 6 in loading and placing workpieces after inspection. Figure 6As shown, taking the loading trolley 11 as an example, the loading trolley 11 is equipped with a positioning fixture 28 for the workpiece 13 to be tested, so as to maintain the relative position and distance between each workpiece, so as to facilitate the orderly gripping and placement of the grippers 14. Different positioning fixtures 28 are designed for different types of workpieces to be tested. The fixtures are fixed by the bayonet on both sides of the trolley, making the replacement of the fixtures simple, quick and convenient.

[0061] After the loading / unloading robot 6 fixes the workpiece to be tested at the measurement station, the measuring robotic arm receives the type information of the workpiece 13 to be tested and the relevant dimensional information entered in the early stage. It automatically plans the scanning path and, in conjunction with the equipped 3D scanner, completes the 3D data scanning of the workpiece 13 to be tested, thus completing the 3D modeling of the workpiece 13. The system further calculates the key dimensions of the workpiece 13 to be tested and, in conjunction with tolerance and other data, determines whether the workpiece 13 to be tested is qualified.

[0062] like Figure 5 As shown, the optical tracker 29 is fixed to the ground by a column 25 and a raised frame 26, at a certain distance from the worktable. The optical tracker 29 is equipped with an electric swing table 24 and a universal ball joint bracket 23. When calibration is required using a calibration rod, the electric swing table 24 rotates to a suitable position for manual calibration. The calibration plate 10 is fixed to the worktable and can be automatically calibrated by the movement of the measuring robot 4.

[0063] The loading / unloading robot 6 is a USTUN ER20-1200-M six-axis industrial robot with a reach of 1200mm, a payload of 20kg, and a repeatability of ±0.03mm. The measuring robot 4 is a USTUN ER7-910-M six-axis industrial robot with a reach of 910mm, a payload of 7kg, and a repeatability of ±0.03mm.

[0064] The positioning camera uses the Mech-Eye LSR L industrial-grade 3D camera 22, with a working distance of 1200mm to 3000mm, a depth map resolution of 2048×1536, an RGB resolution of 4000×3000, and a typical acquisition time of 0.5 to 0.9 seconds.

[0065] Multi-view 3D scanning employs a 1-4 round graded scanning process, with each round bound to a dedicated scanning path and a table rotation angle range (adjustable from 0° to 360°). Based on workpiece size, morphological complexity, surface features, and accuracy requirements, the system automatically determines the number of scanning rounds according to preset thresholds and binding rules. Following a strict matching principle, it automatically associates workpiece feature tags with a formula library, matching 1-4 rounds of dedicated scanning formulas and loading corresponding scanning parameters and paths.

[0066] In actual use, the specific scanning process is as follows: The workpiece to be measured is fixed at the center position of the high-precision rotary table, and the tooling positioning calibration is completed. The 3D scanner, optical tracker 29, and table control system are connected, and the system self-test is initiated. The self-test covers the stability of the scanner light source, the sensitivity of the point cloud acquisition module, the positioning accuracy of the tracker, and the calibration of the table rotation angle. After ensuring that all parameters of the equipment meet the scanning reference requirements and there are no deviations or faults, the system enters the scanning state.

[0067] After identifying the workpiece attributes, the system automatically calls the corresponding level of scanning recipe and simultaneously loads the dedicated scanning path trajectory, worktable rotation angle range, point cloud acquisition parameters, and tracking and positioning parameters for this round. After completing the one-click parameter configuration, the scanning task for this round is generated.

[0068] The control system drives the worktable to rotate to the specified starting angle of the current formula, completing precise angle positioning and locking to eliminate rotational deviations. The scanner moves at a constant speed along a preset path, simultaneously activating laser scanning to acquire 3D point cloud data of the workpiece surface frame by frame, maintaining a stable acquisition frame rate and point cloud density throughout the process. The optical tracker captures the scanner's spatial position and attitude information in real time throughout the process, updating the positioning coordinates frequently and calibrating the scanner's motion deviations to ensure that each frame of point cloud data has a precise spatial coordinate reference.

[0069] After all four rounds of scanning are completed, the local point clouds from each viewpoint are registered and stitched together. The system automatically verifies the coverage integrity, coordinate accuracy, and surface fit of the complete point cloud. If the verification is successful, standardized 3D point cloud data is output; if the verification fails, the corresponding formula parameters are called to rescan the defective areas.

[0070] Actual testing has verified that the system's dimensional calculation accuracy can reach 0.01mm, meeting the requirements for precision inspection; the inspection time for a single piece is about 4 minutes, and the inspection efficiency is 5 to 10 times that of manual inspection.

[0071] The above description is merely a preferred embodiment of the present invention and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of disclosure in this invention is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-disclosed concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this invention.

[0072] It should be noted that the terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and represent a limitation on a specific order or sequence. Where appropriate, the order of use for similar objects can be interchanged so that the embodiments of this application described herein can be implemented in an order other than that shown or described.

[0073] Those skilled in the art will recognize that this invention can be implemented as a system, method, or computer program product. Therefore, this invention can be specifically implemented in the following forms: it can be entirely hardware, entirely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software, generally referred to herein as a "circuit," "module," or "system." Furthermore, in some embodiments, this invention can also be implemented as a computer program product contained in one or more computer-readable media, which includes computer-readable program code.

[0074] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A method for quality inspection of irregularly shaped electrical internal components, characterized in that, include: The scanning device is controlled to perform multi-view scanning of the workpiece to be inspected according to a preset scanning strategy in order to obtain local three-dimensional point cloud data of the workpiece under multiple views. At the same time, the spatial pose data of the scanning device is obtained in real time through the positioning and tracking device. Based on the spatial pose data, the local three-dimensional point cloud data of the workpiece from each viewpoint are registered and stitched together to generate a complete three-dimensional point cloud. Geometric element fitting is performed on the complete 3D point cloud to calculate the geometric parameters of the workpiece to be inspected; The geometric parameters are compared with preset tolerance standards, and the quality judgment result of the workpiece to be inspected is generated based on the comparison result.

2. The method for quality inspection of irregularly shaped electrical internal components according to claim 1, characterized in that, The step of registering and stitching local 3D point cloud data of the workpiece from various viewpoints based on the spatial pose data to generate a complete 3D point cloud includes: Based on the spatial pose data, the local three-dimensional point cloud data of the workpiece from each viewpoint is converted to a unified global coordinate system for initial pose alignment. Iterative nearest-point fine registration processing is performed on the local 3D point cloud data of each workpiece after initial pose alignment; The local 3D point cloud data of each workpiece after fine registration are fused under the unified global coordinate system. Redundant data in overlapping areas are removed, and the spliced ​​areas are smoothed to generate a complete 3D point cloud.

3. The method for quality inspection of irregularly shaped electrical internal components according to claim 1, characterized in that, The step of comparing the geometric parameters with a preset tolerance standard and generating a quality judgment result for the workpiece to be inspected based on the comparison result includes: Each value in the geometric parameters is compared with the corresponding allowable deviation range in the preset tolerance standard to obtain the comparison result; When the comparison result shows that all values ​​are within the corresponding allowable deviation range, the quality judgment result of the workpiece to be inspected is determined to be qualified. If any value in the comparison results exceeds the corresponding allowable deviation range, the quality judgment result of the workpiece to be inspected is determined to be unqualified.

4. The method for quality inspection of irregularly shaped electrical internal components according to claim 1, characterized in that, The step of fitting geometric elements to the complete 3D point cloud and calculating the geometric parameters of the workpiece to be inspected includes: The complete 3D point cloud is subjected to denoising, coordinate unification calibration and point cloud simplification respectively to obtain the preprocessed 3D point cloud; The preprocessed 3D point cloud is segmented into feature regions, and multiple feature regions are extracted. Geometric element fitting is performed on the point cloud in each feature region to calculate the geometric parameters of the workpiece to be inspected. The geometric parameters include at least one of length, width, height, hole diameter, groove width, groove depth, wall thickness, parallelism, perpendicularity, coaxiality, symmetry, and contour.

5. The method for quality inspection of irregularly shaped electrical internal components according to claim 4, characterized in that, The aperture is calculated in the following way: Extract the annular point cloud of the inner wall of the hole from the point cloud within the feature region; Spatial circle fitting is performed on the annular point cloud inside the hole, and the spatial axis of the center of the circular hole is determined based on the fitting result. Cross sections are taken at different depths within the hole along the direction of the central spatial axis of the circular hole; Spatial circle fitting is performed on the point cloud of each cross section to determine the aperture value corresponding to each cross section; The aperture is obtained by calculating the average value of the aperture value corresponding to each cross section.

6. The method for quality inspection of irregularly shaped electrical internal components according to claim 4, characterized in that, The slot width is calculated in the following way: Extract point cloud data of the two side walls of the tank from the point cloud within the feature region; Plane fitting is performed on the point cloud data of the two side walls of the tank to obtain the planes of the two side walls of the tank; Multiple cross-sectional positions are selected along the extension direction of the tank. For each cross-sectional position, the vertical distance between the two side walls of the tank at the cross-sectional position is calculated and used as the sampled value of the tank width at the cross-sectional position. The average value of all slot width samples is calculated to obtain the slot width.

7. The method for quality inspection of irregularly shaped electrical internal components according to claim 4, characterized in that, The trench depth is calculated in the following way: Extract the point cloud data of the groove edge and the groove bottom from the point cloud within the feature region, respectively; Plane fitting is performed on the point cloud data of the slot edge to obtain the reference plane of the top surface of the slot; The point cloud data at the bottom of the trench is fitted with a plane to obtain the reference plane at the bottom of the trench. Using the reference plane at the top of the slot as a reference, the vertical downward distance of the reference plane at the bottom of the slot relative to the reference plane at the top of the slot is calculated and used as the slot depth.

8. The method for quality inspection of irregularly shaped electrical internal components according to claim 4, characterized in that, The contour ratio is calculated in the following way: Extract the workpiece measured contour point cloud data from the point cloud in the feature region, and generate the actual contour of the workpiece based on the workpiece measured contour point cloud data. Obtain the theoretical contour model corresponding to the workpiece to be inspected; The actual contour of the workpiece is registered and aligned with the theoretical contour model; The normal deviation of each contour point on the actual contour of the aligned workpiece is calculated point by point relative to the corresponding contour point on the theoretical contour model, and the normal deviation value of each contour point is obtained. The profile degree is calculated based on the normal deviation value of each profile point.

9. A quality inspection system for irregularly shaped electrical internal components, characterized in that, include: The scanning device is used to perform multi-view scanning of the workpiece to be inspected according to a preset scanning strategy, and to obtain local three-dimensional point cloud data of the workpiece from multiple perspectives. A workpiece carrying device is used to carry the workpiece to be inspected and to drive the workpiece to be inspected to change its posture during the scanning process; A positioning and tracking device is used to acquire the spatial pose data of the scanning device in real time; A control and processing device is used to control the scanning device and the workpiece carrying device to work together and to perform the quality inspection method for irregularly shaped electrical internal components as described in any one of claims 1 to 8.

10. The quality inspection system for irregularly shaped electrical internal components according to claim 9, characterized in that, It also includes a loading and unloading device, and the control and processing device is further used to control the loading and unloading device to sort the workpiece to be inspected into a preset qualified area or unqualified area according to the quality judgment result.

Images