Deviation device and method suitable for large curved surface component

Abstract

The invention discloses a calibration device and method suitable for a large curved surface component, and relates to the technical field of digital calibration, and the device comprises a movement mechanism module, a profile measurement module and a data processing module. The profile measuring module is integrated on an end effector of the movement mechanism module; the movement mechanism module is used for driving the profile measurement module to execute dynamic multi-angle scanning around the to-be-measured component and collecting full-view-field point cloud data covering the surface of the whole to-be-measured component; the data processing module is respectively communicated with the movement mechanism module and the profile measurement module; and the movement mechanism module is used for receiving the full-view-field point cloud data, performing multi-scale error analysis based on the surface curvature of the to-be-measured component and controlling the scanning process of the movement mechanism module. According to the invention, high-precision automatic detection of the processing deviation of the component can be realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of digital proofing, in particular to a proofing device and method suitable for large curved components. BACKGROUND

[0002] The existing digital proofing technology has achieved certain application results in the detection of small and medium-sized structural parts, but when facing large curved components, there are still problems such as limited measurement range, large error, low efficiency, and low automation. Especially in the context of complex curved structural parts being widely used in high-end manufacturing fields such as aerospace, energy equipment, and shipbuilding, it is difficult to complete high-precision proofing tasks relying only on manual and traditional tools, and it is difficult to meet the detection needs of modern manufacturing for intelligentization, high efficiency, and high precision. In addition, in the traditional detection of large curved components, the point cloud registration is generally used to calculate the mean square error (MSE) to quantify the processing quality. However, this method cannot reflect the local differences in error analysis, especially in key structural areas such as rib position lines, sharp angle boundaries, and high-curvature transitions, which have problems such as distortion of error evaluation and insufficient sensitivity, making it difficult to meet the actual needs of high-precision manufacturing and assembly quality control.

[0003] Therefore, based on the deficiencies of the prior art, it is necessary to provide a proofing device or method suitable for large curved components to realize high-precision and automated detection of component processing deviations. SUMMARY

[0004] The purpose of the present application is to provide a proofing device and method suitable for large curved components, which can realize high-precision and automated detection of component processing deviations.

[0005] To achieve the above-mentioned purpose, the present application provides the following solutions: In a first aspect, the present application provides a proofing device suitable for large curved components, which comprises a motion mechanism module, a shape measurement module, and a data processing module. The shape measurement module is integrated on the end effector of the motion mechanism module. The motion mechanism module is used to drive the shape measurement module to perform dynamic multi-angle scanning around the component to be measured, and to collect full-view point cloud data covering the surface of the component to be measured. The data processing module communicates with the motion mechanism module and the shape measurement module respectively; the motion mechanism module is used to receive the full-view point cloud data and perform multi-scale error analysis based on the surface curvature of the component to be measured, and control the scanning process of the motion mechanism module.

[0006] Optionally, the motion mechanism module specifically comprises a track sliding table, a collaborative robot, a servo driver, and a motion controller. The collaborative robot is mounted on a track slide and is used to control the profile measurement module to perform spatial displacement and multi-angle scanning. The track slide is arranged on both sides of the component clamping table along the feeding and discharging directions of the component to be measured and is parallel to the component to be measured. The collaborative robot and the track slide are linked and operated under the unified coordination of the servo driver and the motion controller, so that the profile measurement module completes the planning and execution of the dynamic scanning path in space. The motion controller is also used to transmit the movement distance of the profile measurement module to the data processing module; and the data processing module is used to splice the point cloud data field according to the movement distance of the profile measurement module.

[0007] Optionally, the motion controller is connected with the data processing module through an industrial bus interface.

[0008] Optionally, the profile measurement module includes a laser emitting unit, a laser receiving unit and a control and calculation unit; the control and calculation unit is connected with the data processing module through Ethernet, and is used to realize the configuration and real-time trigger control of the shooting parameters through a software interface; the shooting parameters include exposure gain level, exposure time and filtering parameters.

[0009] Optionally, the data processing module includes an industrial processing unit, a switch and a display terminal. The industrial processing unit is connected with the switch through an industrial Ethernet, and is used to run the full-field point cloud data filtering, splicing, registration and error analysis. The switch establishes a communication link with the motion controller, and is used to issue control instructions and return state information. The display terminal is connected with the industrial processing unit, and is used to provide a human-computer interaction interface for real-time display of scanning progress, measurement results and running state.

[0010] In a second aspect, the application provides a proofing method suitable for large curved surface components, which is used to realize the proofing device suitable for large curved surface components, and the proofing method suitable for large curved surface components includes: The profile measurement module is driven by the motion mechanism module to perform dynamic multi-angle scanning around the component to be measured to collect full-field point cloud data covering the surface of the component to be measured; The data processing module is used to denoise, extract key points, perform initial registration and accurate registration on the full-field point cloud data, complete point cloud splicing, and construct a component surface point cloud model; The spliced point cloud is globally registered with the point cloud output by the component surface point cloud model to obtain a root mean square error; The multi-scale structure error proofing is performed according to the root mean square error and the surface curvature of the component to be measured.

[0011] Optionally, the data processing module is used to denoise, extract key points, initial registration and accurate registration of the full field of view point cloud data, complete point cloud stitching, and construct a component surface point cloud model, and further comprises: Converting the full field of view point cloud data into a standard format and importing the component CAD model in the standard IGS format; Discretizing the component CAD model in the standard IGS format by a three-dimensional modeling software to generate uniformly distributed point cloud data.

[0012] Optionally, the data processing module is used to denoise, extract key points, initial registration and accurate registration of the full field of view point cloud data, complete point cloud stitching, and construct a component surface point cloud model, and specifically comprises: Using a statistical outlier rejection algorithm or a radius filtering algorithm to denoise the full field of view point cloud data; Using an ISS algorithm to extract key points from the denoised point cloud; Using a principal component analysis method and an ICP algorithm to stitch the point cloud after key point extraction.

[0013] Optionally, the globally registered point cloud after stitching and the component surface point cloud model output point cloud are obtained, and specifically comprises: According to the point cloud after stitching and the component surface point cloud model output point cloud, a principal component analysis method and an ICP algorithm are used for global registration to obtain a root mean square error.

[0014] Optionally, the root mean square error and the surface curvature of the component to be measured are used to perform multi-scale structure error calibration, and specifically comprises: According to the surface curvature of the component to be measured, a scale partition is divided to obtain a plurality of structure scale levels; According to the plurality of structure scale levels, corresponding error evaluation is performed to obtain a structure hierarchical error map.

[0015] According to the specific embodiments provided by the application, the application has the following technical effects: The application provides a calibration device and method suitable for large curved surface components. The shape surface measurement module is driven by the motion mechanism module to perform dynamic multi-angle scanning around the component to be measured to collect full field of view point cloud data covering the surface of the component to be measured. Then, the data processing module is used to analyze the full field of view point cloud data to realize high-precision automatic detection of component processing deviation. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed in the embodiments. Obviously, the drawings described below only illustrate some of the embodiments of the present application, and other drawings can be obtained by those of ordinary skill in the art without any creative effort based on these drawings.

[0017] Figure 1 A structural schematic diagram of a proofing device suitable for large curved surface components in an embodiment of the present application; Figure 2 A schematic diagram of a rotation axis transformation for converting STL data to two dimensions; Figure 3 A schematic diagram of sampling and discretizing points for triangle ABC; Figure 4 A flowchart of a multi-scale structure error proofing method based on curvature division. DETAILED DESCRIPTION

[0018] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without any creative effort fall within the scope of protection of the present application.

[0019] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application will be further described in detail below with reference to the drawings and specific embodiments.

[0020] In an exemplary embodiment, as shown in Figure 1 a proofing device suitable for large curved surface components is provided, which comprises a motion mechanism module, a surface measurement module and a data processing module; The surface measurement module is integrated on the end effector of the motion mechanism module; The motion mechanism module is used to drive the surface measurement module to perform dynamic multi-angle scanning around the component to be measured, so as to collect full-view point cloud data covering the surface of the component to be measured; The data processing module communicates with the motion mechanism module and the surface measurement module respectively; the motion mechanism module is used to receive the full-view point cloud data, and perform multi-scale error analysis based on the surface curvature of the component to be measured and control the scanning process of the motion mechanism module.

[0021] As a specific embodiment, the motion mechanism module specifically comprises a track sliding table, a collaborative robot, a servo driver and a motion controller; The collaborative robot is mounted on a track sliding table, and is used to control a profile measurement module to perform spatial displacement and multi-angle scanning; wherein the collaborative robot serves as an execution unit to realize multi-degree-of-freedom attitude adjustment. The track sliding table is arranged on both sides of a component clamping table along a component feeding direction (component loading direction) and a component discharging direction (component unloading direction) of a component to be measured, and is parallel to the component to be measured; the track sliding table is composed of a servo motor, a speed reducer and a linear guide rail system, the servo motor is connected with a sliding table transmission assembly through the speed reducer to realize linear movement along the component direction. The collaborative robot and the track sliding table are linked and operated under the unified coordination of a servo driver and a motion controller, so that the profile measurement module completes high-precision, multi-angle and continuous dynamic scanning path planning and execution in space. The motion controller is also used to transmit the movement distance of the profile measurement module to a data processing module; and the data processing module is used to splice the point cloud data field of view according to the movement distance of the profile measurement module.

[0022] As a specific embodiment, the motion controller is connected with the data processing module through an industrial bus interface, and can move the profile measurement module to the next working point according to the set speed according to the control instruction.

[0023] Specifically, the proofing device of the present application supports two types of path planning methods: one is a preset scanning path, that is, a fixed trajectory is generated according to a component geometric model or a scanning strategy before the task starts; the other is adaptive path planning, that is, the trajectory is dynamically adjusted based on real-time measurement data (such as point cloud or curvature feature) during the scanning process, so as to adapt to different component topographies.

[0024] As a specific embodiment, the profile measurement module includes a laser emitting unit, a laser receiving unit and a control and calculation unit; the control and calculation unit is connected with the data processing module through Ethernet, and is used to realize configuration and real-time trigger control of shooting parameters such as exposure gain level, exposure time, filtering parameter through a software interface. The software interface communicates with the control and calculation unit by calling a dll form of sdk interface, and the sdk internally uses tcp / udp method to communicate with the control and calculation unit.

[0025] The laser emitting unit can be a stripe projection module; and the laser receiving unit can be a three-dimensional imaging camera and a color camera.

[0026] As a specific embodiment, the data processing module includes an industrial-grade processing unit, a switch and a display terminal. The industrial-grade processing unit is connected with the switch through an industrial Ethernet, and is used for running full-view field point cloud data filtering, splicing, registration and error analysis; the industrial-grade processing unit is connected with the profile measurement module through the industrial Ethernet to build an internal local area network environment; The switch is connected with the motion controller to establish a communication link, and is used for issuing control instructions and returning state information; the control instructions include reading parameters, setting IP addresses, exposure gains, exposure times, filtering parameters, triggering acquisition and reading data streams; the state information includes control instruction responses and the like; The display terminal is connected with the industrial-grade processing unit, and is used for providing a man-machine interactive interface, displaying scanning progress, measurement results (error analysis results) and running states in real time, realizing operation visualization and intuitive presentation of measurement results, and assisting manual or automatic decision-making.

[0027] The data processing module communicates with the motion mechanism module to ensure the timeliness consistency of the control instructions and the data streams.

[0028] The application further provides a sample correction method suitable for a large curved component, which is used for realizing the sample correction device suitable for the large curved component. S101, the profile measurement module is driven by the motion mechanism module to perform dynamic multi-angle scanning around the component to be measured, and full-view field point cloud data covering the surface of the component to be measured is acquired; S102, the data processing module is used for denoising, key point extraction, initial registration and accurate registration of the full-view field point cloud data, completing point cloud splicing, and constructing a component surface point cloud model; S102 further includes the following steps before S102: As shown in Figure 2 The full-view field point cloud data is converted into a standard format, and is imported into a component CAD model in a standard IGS format; The component CAD model in the standard IGS format is discretized by a three-dimensional modeling software (such as SolidWorks), and uniform distribution point cloud data is generated.

[0029] S102 specifically includes the following steps: S21, a statistical outlier removal algorithm (Statistical Outlier Removal, SOR) or a radius outlier removal algorithm (Radius Outlier Removal, ROR) is used for denoising the full-view field point cloud data; noise points and isolated points in the original point cloud are removed, and data stability and subsequent processing effect are improved; S22, an ISS algorithm is used for extracting key points from the denoised point cloud; S23, using principal component analysis (PCA) and ICP algorithm for key point extraction after point cloud splicing.

[0030] The initial registration is performed by using the principal component analysis method to obtain the approximate attitude. On the basis of the initial registration, the ICP algorithm is used for accurate registration to calculate the transformation matrix and splice all the sub-point clouds to obtain a complete curved surface point cloud. Then, the resampling processing is performed on the spliced point cloud to unify the point spacing, eliminate the joint differences, and improve the overall model continuity and smoothness.

[0031] S103, globally registering the spliced point cloud and the point cloud output by the component surface point cloud model to obtain a root mean square error; the root mean square error is used to measure the registration accuracy of the point cloud and the component model in the overall scale, and provides a global benchmark for subsequent error analysis; Specifically, the principal component analysis method and the ICP algorithm are used for global registration based on the spliced point cloud and the point cloud output by the component surface point cloud model to obtain a root mean square error.

[0032] S104, multi-scale structure error calibration is performed according to the root mean square error and the surface curvature of the component to be measured; and then the global error is further decomposed into different structure levels to realize differentiated local error evaluation; For large curved surface features, scale division is performed based on surface curvature and structure details, and combined with local and global error evaluation methods, the precision detection and structured error output of the large curved surface component are completed.

[0033] Firstly, based on the curvature distribution of the component surface point cloud model, the component surface is divided into different scales. Specifically, by analyzing the curvature variation trend, a suitable threshold is set to automatically divide the surface of the component to be measured into multiple structure scale levels, including the overall flat area (large scale smooth area), the transition area with medium curvature (medium curvature area), and the detail feature area with significant local curvature variation (local high curvature area). The detail feature area includes: ribs, hole edges, folded edges, etc. This scale division not only reflects the hierarchical structure of the component in terms of geometric appearance, but also provides a basis for subsequent differentiated error measurement.

[0034] On this basis, different error evaluation strategies are selected for different scale areas: (1) For the large-scale smooth area, a global error measurement method such as root mean square error (RMSE) is used to measure the overall forming deviation and contour distortion; (2) For the medium curvature area, a weighted local error mapping is used in combination with error gradient variation, and the transition smoothness is used as the evaluation standard; (3) For the local high curvature area, a local geometric neighborhood is constructed and a reference surface is fitted, and the distance from the point to the local fitted surface is used for high-resolution error projection and fine geometric deviation analysis.

[0035] Finally, based on the spatial distribution of error values and the structure classification information, the error atlas of the structure hierarchy is output, providing high-credibility data support for precision detection, defect tracing and manufacturing process optimization of large curved surface components.

[0036] The following is described by specific examples, and the specific operation steps are as follows: (1) The software first calculates the optimal shooting pose sequence according to the size parameters of the component to be measured (which can be obtained by external sensors, CAD model import or manual input), using the built-in spatial coverage algorithm. Based on the geometric characteristics of the component and the camera imaging model, the imaging field of view, the workspace limit, the angle of view obstruction and the point cloud overlap rate are considered to generate shooting sites and corresponding camera poses that cover the entire component surface, thereby ensuring sufficient spatial coverage and redundant matching areas in the scanning process.

[0037] (2) The collaborative robot carries the shape measurement module and moves to each shooting pose point according to the planned path. When it arrives at a point, the system triggers a three-dimensional imaging after stabilizing the camera pose, completing single-frame point cloud acquisition. To improve the splicing accuracy, there is a partial overlap area between adjacent frames to ensure the robustness and continuity of subsequent data registration.

[0038] (3) During the scanning process, the spatial coverage and local density distribution of the collected point cloud are continuously evaluated. If there are blocked dead angles, insufficient sampling in areas with sudden curvature changes or angle of view blind spots, the software will automatically determine the uncovered areas and re-plan the shooting pose to guide the robot to perform local retake operations, improving the overall point cloud integrity and detail fidelity.

[0039] (4) After scanning is completed and the device is reset, the motion controller automatically resets the shape measurement module to the initial position; the component is removed from the measurement area by the transfer device, preparing for the measurement process of the next component.

[0040] After obtaining the point cloud data of the large curved surface component, the next step is to convert the point cloud format of the model data, preprocess the point cloud data, and splice the point cloud.

[0041] The conversion steps of the point cloud format of the model data are as follows: Step 1, export the IGS model data of the component to be measured to STL format data; further, discretize the three-dimensional model set data stored in STL format to point cloud data. Specifically, project the spatial triangular facets to a two-dimensional plane, taking the projection to the xoy plane as an example.

[0042] Step 1.1, obtain the coordinates of the three vertices of each triangular facet in the STL file and calculate the normal vector.

[0043] Step 1.2, as shown in Figure 2 the normal vector is transformed to the xoz plane by rotating the angle of OA around the z axis by α; Step 1.3, the normal vector is rotated to the z axis by rotating the angle of OA around the y axis clockwise by β; Step 1.4, map to the xoy plane using the translation vector formula Step 1.5, as shown in Figure 3 ABC is a triangular patch projected to the xoy plane by the space transformation, take the maximum and minimum coordinates in the three vertices, set as , , , ; Step 1.6, according to the maximum and minimum coordinates, establish a rectangular range as shown in Figure 3 ; Step 1.7, take a sampling point every in the horizontal direction of the rectangle, and take a sampling point every in the vertical direction; Step 1.8, for each two-dimensional sampling point , use the barycentric coordinate method to judge whether it is located inside the triangle. The specific judgment method is as follows: let the triangle vertices be , , , calculate the barycentric coordinates: ; ; ; If , then the point is located inside the triangle; Step 1.9, after sampling and judging, map the sampling points that meet the requirements back to the three-dimensional space according to their barycentric coordinates; Step 1.10, repeat the above operation for all triangular patches, and finally get a set of discrete points uniformly distributed on the surface of the three-dimensional model, which constitutes the target point cloud data; Step 2, the point cloud preprocessing is performed by using the statistical outlier rejection algorithm and the ISS algorithm for denoising and key point extraction respectively. First, the statistical outlier rejection algorithm calculates the average distance of each point from its neighboring points, counts the distance distribution of all points, and sets the threshold as the mean value plus several times the standard deviation, and removes those points whose average distance is much larger than the normal range, thereby effectively removing the noise and outliers in the point cloud. The ISS algorithm first calculates the covariance matrix in the neighborhood of each point and extracts the eigenvalue, selects the points with local geometric significance as candidate key points, and then through non-maximum suppression, retains the points with the largest response value as the final key points.

[0044] Step 3, the point cloud is used to splice the overall surface of the measured component. Specifically, principal component analysis is used for initial registration to obtain the approximate pose. On this basis, ICP algorithm is used for accurate registration to calculate the transformation matrix, and the complete curved surface point cloud is obtained after splicing. The specific implementation steps of principal component analysis and ICP are as follows: Step 3.1, calculate the centroid of the point cloud, set the point cloud as , each point , calculate the centroid ; Step 3.2, move the point cloud to the coordinate system with the centroid as the origin to obtain the centralized point cloud , calculate the covariance matrix ; Step 3.3, perform eigenvalue decomposition on the covariance matrix to obtain the eigenvectors corresponding to the eigenvalues. The eigenvalues are sorted in descending order, and the eigenvectors corresponding to the three largest eigenvalues are selected as the principal directions of the point cloud; Step 3.4, set the principal direction matrix of the source point cloud as , the principal direction matrix of the target point cloud as , and calculate the rotation matrix ; Step 3.5, the translation vector of the centroid is , and the rigid transformation is obtained ; Step 4, use the point-to-plane ICP algorithm for precise registration.

[0045] Step 4.1, transform the source point cloud obtained by step 5.5 into the source point cloud ; ; Step 4.2, establish a kd-tree data structure for the target point cloud Q; Step 4.3, for each target point , obtain its normal vector information (if unknown, need to use principal component analysis to obtain) ; Step 4.4, for each transformed source point , find its nearest point from the target point cloud , then calculate the point-to-plane distance between the point and the surface of the target point cloud. The point-to-plane error is: ; Step 4.5, use singular value decomposition to obtain the optimal rotation matrix and translation vector ; ; ; ; ; ; Step 4.6, update the source point cloud by the obtained rotation matrix and translation vector ; ; Step 4.7, judge whether the ICP algorithm has converged by calculating the error between the source point cloud and the target point cloud. If the error change is less than the preset threshold or the maximum iteration number is reached, the algorithm stops.

[0046] ; Step 5, use principal component analysis and ICP algorithm to register the obtained overall surface point cloud and the surface point cloud extracted from the model, and obtain the mean square error, which is the same as steps 3 and 4.

[0047] Step 6, as shown in Figure 4 , perform multi-scale structure error calibration.

[0048] Step 6.1, use kd-tree to search for nearby points within a certain range; Step 6.2, calculate the covariance matrix, which is the same as steps 3.1 and 3.2; Step 6.3, perform eigenvalue decomposition, denoted as ; for point , its curvature is: ; Further, the is the point The eigenvalues of the neighborhood covariance matrix reflect the dispersion of the neighborhood point cloud in the three principal directions. Among them, The normal direction corresponds to the deviation of the point cloud along the normal direction, which is used to represent the bending strength of the local surface; The distribution in the tangent plane reflects the extension characteristics of the neighborhood point cloud in the plane.

[0049] Step 6.4, using data-driven and statistical analysis to analyze the curvature data, observing its distribution pattern, calculating its mean, standard deviation, and quartile; Step 6.5, based on the statistical results, select the appropriate threshold division strategy (can refer to experience or expert knowledge value). If the curvature data distribution is close to the normal distribution, use the mean and standard deviation to divide into low curvature area, medium curvature area and high curvature area.

[0050] Step 6.6, for the low curvature area, calculate its mean square error, same as step 3.7; Step 6.7, for the medium curvature area, perform weighted error calculation.

[0051] ; Among them, is the point in the source point cloud after registration, is the corresponding point in the target point cloud; is the curvature change rate, is a variable parameter for adjusting the weight; is the weighted coefficient of point , used to enhance the error contribution of the area with significant curvature change.

[0052] Step 6.8, for the high curvature area, calculate the distance from the point to the fitting plane. First, get the centroid and normal vector of the fitting plane, same as steps 3.1 to 3.3. Second, get the fitting plane equation. Finally, calculate the distance from the point to the plane, and get the total deviation between the source point and the target plane.

[0053] ; ; Among them, is the fitting plane equation coefficient, is the plane normal vector, is the plane constant term; is the three-dimensional coordinates of the point to be tested ; is the vertical distance from point to the fitting plane, which quantifies the local deviation from the target plane.

[0054] Step 7, output the hierarchical error map of the structure.

[0055] In this application, all actions of obtaining signals, information or data are performed under the premise of complying with the corresponding data protection regulations and policies of the country where the device is located, and obtaining the authorization given by the owner of the corresponding device.

[0056] The technical features of the above embodiments can be combined in any manner. In order to make the description simple, all possible combinations of the technical features in the above embodiments are not described, however, as long as the combinations of the technical features do not exist contradictions, they should be considered as the scope of the present disclosure.

[0057] The principles and implementation manners of the present application are described by using specific examples herein, and the above descriptions of the embodiments are only used to help understand the method and the core idea of the present application; meanwhile, for those skilled in the art, the specific implementation manners and application ranges can be changed according to the idea of the present application. In conclusion, the content of the present description should not be understood as a limitation of the present application.

Claims

1. A prototyping device suitable for large curved surface components, characterized in that, The prototyping device for large curved surface components includes: a motion mechanism module, a surface measurement module, and a data processing module; The surface measurement module is integrated into the end effector of the motion mechanism module; The motion mechanism module is used to drive the surface measurement module to perform dynamic multi-angle scanning around the component under test, and to collect full-field point cloud data covering the entire surface of the component under test. The data processing module communicates with the motion mechanism module and the surface measurement module respectively; the motion mechanism module is used to receive full field-of-view point cloud data, perform multi-scale error analysis based on the surface curvature of the component under test, and control the scanning process of the motion mechanism module.

2. The prototyping device for large curved surface components according to claim 1, characterized in that, The motion mechanism module specifically includes: a track slide, a collaborative robot, a servo driver, and a motion controller; The collaborative robot is mounted on a track slide and is used to control the surface measurement module to perform spatial displacement and multi-angle scanning. The track slide is arranged on both sides of the component clamping platform along the feeding and unloading directions of the component to be tested, and is parallel to the component to be tested; The collaborative robot and the track slide operate in unison under the unified coordination of the servo driver and motion controller, enabling the shape measurement module to plan and execute the dynamic scanning path in space. The motion controller is also used to transmit the moving distance of the shape measurement module to the data processing module; the data processing module is used to stitch the point cloud data field of view according to the moving distance of the shape measurement module.

3. The prototyping device for large curved surface components according to claim 2, characterized in that, The motion controller is connected to the data processing module via an industrial bus interface.

4. The prototyping device for large curved surface components according to claim 1, characterized in that, The surface measurement module includes a laser emitting unit, a laser receiving unit, and a control and calculation unit. The control and calculation unit is connected to the data processing module via Ethernet and is used to configure and control the shooting parameters in real time through a software interface. The shooting parameters include exposure gain level, exposure time, and filtering parameters.

5. The prototyping device for large curved surface components according to claim 1, characterized in that, The data processing module includes: an industrial-grade processing unit, a switch, and a display terminal; The industrial-grade processing unit is connected to the switch via an industrial Ethernet network and is used to perform full-field point cloud data filtering, stitching, registration, and error analysis. The switch establishes a communication link with the motion controller for issuing control commands and transmitting status information. The display terminal is connected to the industrial-grade processing unit and is used to provide a human-machine interface for real-time display of scanning progress, measurement results and operating status.

6. A method for prototyping large curved surface components, used to implement the prototyping device for large curved surface components as described in any one of claims 1-5, characterized in that, The prototyping method applicable to large curved surface components includes: The motion mechanism module drives the surface measurement module to perform dynamic multi-angle scanning around the component under test, and collect full-field point cloud data covering the entire surface of the component under test; The data processing module is used to denoise, extract key points, perform initial registration and precise registration on the full field of view point cloud data, complete point cloud stitching, and construct a point cloud model of the component surface; The stitched point cloud is globally registered with the point cloud output from the component surface point cloud model to obtain the root mean square error. Multi-scale structural error calibration is performed based on the root mean square error and the surface curvature of the component under test.

7. The prototyping method for large curved surface components according to claim 6, characterized in that, The process of using a data processing module to denoise, extract key points, perform initial registration and precise registration on the full-field point cloud data, complete point cloud stitching, and construct a point cloud model of the component surface, previously included: Convert the full field-of-view point cloud data into a standard format and import it into the standard IGS format component CAD model; The surface of the component CAD model in standard IGS format is discretized using 3D modeling software to generate uniformly distributed point cloud data.

8. The prototyping method for large curved surface components according to claim 6, characterized in that, The process of using a data processing module to denoise, extract key points, perform initial registration and precise registration on the full field-of-view point cloud data, complete point cloud stitching, and construct a point cloud model of the component surface specifically includes: Statistical outlier removal algorithm or radius filtering algorithm are used to denoise the full field-of-view point cloud data; The ISS algorithm is used to extract key points from the denoised point cloud. Principal component analysis and ICP algorithm were used to stitch together the point cloud after key point extraction.

9. The prototyping method for large curved surface components according to claim 6, characterized in that, The step of globally registering the stitched point cloud with the point cloud output from the component surface point cloud model to obtain the root mean square error specifically includes: Based on the stitched point cloud and the point cloud output from the component surface point cloud model, global registration is performed using principal component analysis and ICP algorithm to obtain the root mean square error.

10. The prototyping method for large curved surface components according to claim 6, characterized in that, The multi-scale structural error calibration based on the root mean square error and the surface curvature of the component under test specifically includes: Based on the surface curvature of the component under test, multiple structural scale levels are obtained by dividing the scale into zones. Error assessments are performed at multiple structural scale levels to obtain a hierarchical error map.

Images