A method for dividing the strengthening region of an ultrathin component in ultrafast laser strengthening

Abstract

Disclosed is a method for dividing the strengthening area of an ultrathin component subjected to ultrafast laser strengthening, in which three-dimensional point cloud data of the surface of the ultrathin component to be strengthened is obtained; the three-dimensional point cloud data is preprocessed, including point cloud filtering, downsampling and normal estimation; the three-dimensional point cloud data is projected in a parameter space based on a principal component analysis method, and a parameterized representation of the component is established; the curvature characteristics of each point in the three-dimensional point cloud data are calculated, including principal curvature, Gaussian curvature and average curvature; adaptive grid division is performed based on the curvature characteristics and the amplitude plane constraint of a galvanometer machining; the divided grids are subjected to region growing and merging to generate initial strengthening areas; region boundary optimization and overlapping area setting are performed based on the continuity requirement of the initial strengthening areas; scanning path planning in each sub-area is performed for each sub-area based on the laser strengthening requirement; and the strengthening area division result and the scanning path planning scheme are output.

Description

Technical Field

[0001] This invention relates to the field of laser strengthening technology, and in particular to a method for segmenting the strengthening region in ultra-thin components using ultrafast laser strengthening. Background Technology

[0002] Ultrafast laser strengthening technology, as an advanced surface modification technique, has been widely used in aerospace, precision machinery, and other fields. This technology uses the interaction between ultrashort pulse lasers and material surfaces to form a specific strengthening layer, thereby significantly improving the material's wear resistance, corrosion resistance, and fatigue life. Especially for ultrathin components, whose thickness is typically in the millimeter range (less than 1 mm), traditional surface strengthening methods can easily lead to deformation or damage. Ultrafast laser strengthening technology, with its advantages of non-contact processing, small heat-affected zone, and high processing precision, has become an ideal choice for surface strengthening of ultrathin components.

[0003] However, in practical applications, the geometry of ultrathin components often contains complex curved surface features with varying curvature. Furthermore, galvanometer-based laser shock peening technology is limited by the processing area of ​​the galvanometer and the effective laser impact angle. Therefore, effectively segmenting the strengthening region for these components with complex curved shapes and ensuring reasonable laser scanning path planning are key issues for achieving high-quality laser strengthening. Existing region segmentation methods mainly have the following shortcomings:

[0004] (1) Traditional meshing methods use fixed-size meshes, which cannot adapt to changes in the curvature of the component surface. In areas with drastic curvature changes, over-segmentation or under-segmentation is likely to occur, affecting the uniformity of the strengthening effect.

[0005] (2) Existing methods do not fully consider the working characteristics of the laser galvanometer system. The processing area of ​​the galvanometer system varies with the laser focal length. The requirements of ultrafast laser enhancement place relatively stringent requirements on the laser spot size and laser focal length, thus limiting the processing area. If a segmented region exceeds the processing area of ​​the galvanometer, it is impossible to enhance that region in a single pose state, and further region segmentation is required.

[0006] (3) Most segmentation methods neglect the requirement of continuity between regions. In the actual strengthening process, a certain laser overlap rate needs to be maintained between adjacent laser-strengthened regions to ensure the continuity and integrity of the strengthened layer. However, how to avoid over-strengthening while ensuring the overlap rate is also a challenging problem.

[0007] To address the aforementioned issues, existing technologies in region segmentation algorithms have included curvature-based adaptive mesh generation and feature recognition-based partitioning methods. However, these methods are mostly tailored to specific types of components or specific processing requirements, lacking versatility and systematicity.

[0008] Furthermore, with the development of intelligent manufacturing technology, the requirements for automation and intelligence in the laser strengthening process are becoming increasingly stringent. How to achieve automatic and adaptive segmentation of the strengthened area has become a current research hotspot. However, existing commercial software and technical solutions still require significant manual intervention and experience-based adjustments when handling complex ultra-thin components, making it difficult to meet the demands of efficient and high-quality production.

[0009] The information disclosed in the background section is only intended to enhance the understanding of the background of the present invention, and therefore may contain information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0010] To address the shortcomings or defects of existing technologies, a visual positioning method and system for surface strengthening of ultra-thin components using ultrafast lasers are provided. By comprehensively considering the surface curvature characteristics of the component and the constraints of the galvanometer processing area, adaptive segmentation of the strengthening region is achieved, ensuring the high efficiency of the laser strengthening process and the uniformity of the strengthening quality.

[0011] The objective of this invention is achieved through the following technical solutions.

[0012] A method for segmenting the strengthened region in ultrafast laser strengthening of ultrathin components includes,

[0013] S1, acquire three-dimensional point cloud data of the surface of the ultrathin component to be strengthened;

[0014] S2, preprocessing the three-dimensional point cloud data, including point cloud filtering, downsampling and normal estimation;

[0015] S3, based on the principal component analysis method, performs parametric space projection on 3D point cloud data to establish a parametric representation of the components;

[0016] S4 calculates the curvature features of each point in the 3D point cloud data, including principal curvature, Gaussian curvature and mean curvature.

[0017] S5, adaptive mesh generation based on curvature characteristics and galvanometer processing area constraints;

[0018] S6, perform region growth and merging on the divided mesh to generate the initial reinforced region;

[0019] S7, based on the initial requirement for the continuity of the enhanced region, optimizes the region boundary and sets the overlapping region;

[0020] S8, plans the scanning path within the sub-region;

[0021] S9 outputs enhanced region segmentation results and scanning path planning scheme.

[0022] In the method described, step S2 includes the following preprocessing steps: point cloud filtering is performed using a statistical outlier removal method, by calculating the average distance from each point to its k nearest neighbors, identifying points whose distance exceeds a set threshold as noise points and removing them; point cloud downsampling is performed using a voxel grid method, dividing the point cloud space into a regular voxel grid, with one representative point retained in each voxel; and normal estimation is performed using a neighborhood covariance analysis-based method, by performing principal component analysis on the neighborhood point set of each point to obtain the normal direction.

[0023] In the method described above, step S3, parameter space projection includes: performing principal component analysis on the preprocessed 3D point cloud data, calculating the covariance matrix of the 3D point cloud data, and solving for eigenvalues ​​and eigenvectors; establishing a local coordinate system with the eigenvector corresponding to the largest eigenvalue as the first principal direction and the eigenvector corresponding to the second largest eigenvalue as the second principal direction; projecting the 3D point cloud onto a 2D parameter plane composed of the first two principal directions; and establishing a bidirectional mapping relationship between the parameter plane coordinates and the 3D spatial coordinates as the parameterized representation of the component.

[0024] In the method described above, in step S4, for each point in the three-dimensional point cloud data, the k nearest neighbors in its neighborhood are selected, and the local quadratic surface is fitted using the least squares method. The average curvature and Gaussian curvature are calculated from the coefficients of the fitted surface, and then the principal curvature is calculated.

[0025] In the method described above, step S5, adaptive mesh generation includes: determining the mesh density function based on the curvature distribution; limiting the maximum size of each mesh region to no more than 80% of the effective processing area of ​​the galvanometer based on the galvanometer processing area constraint; and adjusting the mesh distribution through iterative optimization to ensure that the size ratio of adjacent meshes does not exceed a set threshold.

[0026] In the method described above, step S6, region growth and merging includes: setting similarity criteria, including a normal angle threshold, a curvature difference threshold, and a spatial distance threshold; starting with the grid with the smallest curvature as a seed for region growth; calculating the similarity metric of adjacent grids; and merging adjacent grids into the current region when the similarity metric is greater than the set threshold to generate an initial enhanced region.

[0027] In the method described above, step S7, the region boundary optimization includes: fitting discrete boundary points with B-spline curves to generate smooth region boundaries, and setting overlapping zones between adjacent regions according to the laser spot size and the requirements of the strengthening layer thickness.

[0028] In the method described above, step S8, the scanning path planning within the sub-region includes: drawing a planar trajectory line graphic with equal spacing according to the laser spot size and overlap rate requirements, and then projecting the graphic onto the sub-region model in the galvanometer software to obtain a three-dimensional trajectory line graphic that fits the surface shape of the sub-region and satisfies the laser spot overlap rate.

[0029] In the method described, step S9 outputs enhanced region segmentation results including parameter representations of region boundaries, scan path planning within the region, and recommended process parameters for each region.

[0030] In the method described above, the recommended process parameters for each region include assigning corresponding laser process parameters to each region based on its geometric characteristics and enhancement requirements, and generating an enhancement scheme that includes regional geometric information and process parameters.

[0031] A system for implementing the aforementioned method for segmenting the strengthened region in ultrafast laser strengthening of ultrathin components includes,

[0032] The 3D scanning module is used to acquire point cloud data of the component to be reinforced;

[0033] The data processing module is used to perform enhanced region segmentation;

[0034] The laser enhancement module is used to perform laser enhancement processing based on the segmentation results;

[0035] The control module coordinates the operation of the 3D scanning module, data processing module, and laser strengthening module, and monitors the strengthening process.

[0036] Compared with existing technologies, the beneficial effects of this invention are as follows: By comprehensively considering curvature characteristics and galvanometer processing area constraints, this invention achieves adaptive segmentation of the reinforced region, improving segmentation accuracy and rationality; Principal component analysis is used for parametric space projection, simplifying the processing of complex three-dimensional surfaces and improving algorithm efficiency; Through region growth and merging strategies, the number of segmented regions is effectively reduced, lowering the complexity of path planning between different regions; The introduction of region boundary optimization and overlapping region settings ensures the continuity and uniformity of the reinforced layer; The overall method has good versatility and automation, and can adapt to ultra-thin components of different shapes and sizes.

[0037] The description provided is merely an overview of the technical solution of this invention. In order to make the technical means of this invention clearer and more understandable, so that those skilled in the art can implement it according to the contents of the specification, and to make the described and other objects, features and advantages of this invention more obvious and understandable, specific embodiments of this invention are described below. Attached Figure Description

[0038] Various other advantages and benefits of the present invention will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiments below. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. It is obvious that the drawings described below are merely some embodiments of the invention, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort. Furthermore, the same reference numerals denote the same parts throughout the drawings.

[0039] In the attached diagram:

[0040] Figure 1 This is a step diagram of a method for segmenting the strengthening region in ultrafast laser strengthening of ultrathin components according to the present invention;

[0041] Figure 2 This is a schematic diagram of the principal component analysis method for the strengthening region segmentation method of ultrafast laser strengthening of ultrathin components according to the present invention;

[0042] Figure 3 This is a schematic diagram of an embodiment of the method for segmenting the strengthening region in ultrafast laser strengthening of ultrathin components according to the present invention;

[0043] Figure 4 This is a correlation analysis diagram of curvature characteristics and adaptive mesh size according to the present invention;

[0044] Figure 5 This is a similarity discrimination curve diagram for a region growth and merging process according to the present invention.

[0045] The present invention will be further explained below with reference to the accompanying drawings and embodiments. Detailed Implementation

[0046] Specific embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While specific embodiments of the invention are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the invention and to fully convey the scope of the invention to those skilled in the art.

[0047] It should be noted that certain terms are used in the specification and claims to refer to specific components. Those skilled in the art will understand that different terms may be used to refer to the same component. This specification and claims do not distinguish components based on differences in terminology, but rather on differences in function. The terms "comprising" or "including" used throughout the specification and claims are open-ended and should be interpreted as "comprising but not limited to." The following descriptions are preferred embodiments for carrying out the invention; however, these descriptions are for the purpose of understanding the general principles of the specification and are not intended to limit the scope of the invention. The scope of protection of this invention is determined by the appended claims.

[0048] To facilitate understanding of the embodiments of the present invention, the following will provide further explanation and description with reference to the accompanying drawings and several specific embodiments, and the accompanying drawings do not constitute a limitation on the embodiments of the present invention.

[0049] To better understand, such as Figures 1 to 5 As shown, the method includes,

[0050] S1, acquire three-dimensional point cloud data of the surface of the ultrathin component to be strengthened;

[0051] S2, preprocessing the three-dimensional point cloud data, including point cloud filtering, downsampling and normal estimation;

[0052] S3, based on the principal component analysis method, performs parametric space projection on 3D point cloud data to establish a parametric representation of the components;

[0053] S4 calculates the curvature features of each point in the 3D point cloud data, including principal curvature, Gaussian curvature and mean curvature.

[0054] S5, adaptive mesh generation based on curvature characteristics and galvanometer processing area constraints;

[0055] S6, perform region growth and merging on the divided mesh to generate the initial reinforced region;

[0056] S7, based on the initial requirement for the continuity of the enhanced region, optimizes the region boundary and sets the overlapping region;

[0057] S8, plans the scanning path within the sub-region;

[0058] S9 outputs enhanced region segmentation results and scanning path planning scheme.

[0059] In a preferred embodiment of the method, step S2 includes the following preprocessing steps: point cloud filtering using a statistical outlier removal method, identifying and removing points whose distance exceeds a set threshold as noise points by calculating the average distance from each point to its k nearest neighbors; point cloud downsampling using a voxel grid method, dividing the point cloud space into a regular voxel grid, retaining a representative point within each voxel; and normal estimation using a neighborhood covariance analysis-based method, obtaining the normal direction by performing principal component analysis on the neighborhood point set of each point.

[0060] In a preferred embodiment of the method, step S3, parameter space projection, includes: performing principal component analysis on the preprocessed 3D point cloud data, calculating the covariance matrix of the 3D point cloud data, and solving for eigenvalues ​​and eigenvectors; establishing a local coordinate system with the eigenvector corresponding to the largest eigenvalue as the first principal direction and the eigenvector corresponding to the second largest eigenvalue as the second principal direction; projecting the 3D point cloud onto a 2D parameter plane composed of the first two principal directions; and establishing a bidirectional mapping relationship between the parameter plane coordinates and the 3D spatial coordinates as the parameterized representation of the component.

[0061] In a preferred embodiment of the method, in step S4, for each point in the three-dimensional point cloud data, the k nearest neighbors in its neighborhood are selected, and the local quadratic surface is fitted using the least squares method. The average curvature and Gaussian curvature are calculated from the coefficients of the fitted surface, and then the principal curvature is calculated.

[0062] In a preferred embodiment of the method, step S5, adaptive mesh division includes: determining the mesh density function based on the curvature distribution; limiting the maximum size of each mesh region to no more than 80% of the effective processing area of ​​the galvanometer based on the galvanometer processing area constraint; and adjusting the mesh distribution through iterative optimization to ensure that the size ratio of adjacent meshes does not exceed a set threshold.

[0063] In a preferred embodiment of the method, step S6, region growth and merging includes: setting similarity criteria, including a normal angle threshold, a curvature difference threshold, and a spatial distance threshold; starting with the grid with the smallest curvature as a seed for region growth; calculating the similarity metric of adjacent grids; and merging adjacent grids into the current region when the similarity metric is greater than the set threshold to generate an initial enhanced region.

[0064] In a preferred embodiment of the method, step S7, the region boundary optimization includes: fitting discrete boundary points with B-spline curves to generate smooth region boundaries, and setting overlapping bands between adjacent regions according to the laser spot size and the requirements of the strengthening layer thickness.

[0065] In a preferred embodiment of the method, step S8, the scanning path planning within the sub-region includes: drawing planar trajectory lines with equal spacing according to the laser spot size and overlap rate requirements, and then projecting the graphic onto the sub-region model in the galvanometer software to obtain a three-dimensional trajectory line graphic that fits the surface shape of the sub-region and satisfies the laser spot overlap rate.

[0066] In a preferred embodiment of the method, in step S9, the output enhanced region segmentation result includes parameter representation of the region boundary, scanning path planning within the region, and recommended process parameters for each region.

[0067] In a preferred embodiment of the method, the recommended process parameters for each region include assigning corresponding laser process parameters to each region based on the region's geometric characteristics and strengthening requirements, and generating a strengthening scheme that includes regional geometric information and process parameters.

[0068] A system for implementing the aforementioned method for segmenting the strengthened region in ultrafast laser strengthening of ultrathin components includes,

[0069] The 3D scanning module is used to acquire point cloud data of the component to be reinforced;

[0070] The data processing module is used to perform enhanced region segmentation;

[0071] The laser enhancement module is used to perform laser enhancement processing based on the segmentation results;

[0072] The control module coordinates the operation of the 3D scanning module, data processing module, and laser strengthening module, and monitors the strengthening process.

[0073] In one embodiment, the method first acquires surface data of the component to be strengthened using 3D point cloud technology. High-quality point cloud information is obtained through preprocessing techniques such as statistical outlier removal, voxel mesh downsampling, and neighborhood covariance analysis. Secondly, principal component analysis is used to project the point cloud into parametric space, establishing a parametric representation of the component surface and achieving the mapping transformation of complex 3D surfaces to a 2D parametric plane. Based on this, a quadratic surface fitting method is used to calculate curvature characteristic parameters such as principal curvature, Gaussian curvature, and mean curvature of each point in the point cloud, laying the foundation for subsequent adaptive mesh generation. Considering the working characteristics of the galvanometer processing area being constrained by the laser focal length, this method establishes a curvature distribution-driven mesh density function while ensuring that each mesh region does not exceed the effective processing area of ​​the galvanometer. Furthermore, by comprehensively considering the similarity criteria of normal angle, curvature difference, and spatial distance, the mesh is grown and merged to effectively reduce the number of segmented regions. Further addressing the continuity requirements between adjacent strengthening regions, B-spline curve fitting is used for boundary smoothing optimization, and an appropriate width of overlap band is set between adjacent regions according to the laser spot size and strengthening layer thickness to ensure the continuity and integrity of the strengthening layer. Finally, planar trajectory lines with equal spacing are drawn according to the laser spot size and overlap ratio requirements. These lines are then projected onto the sub-region model in the galvanometer software to obtain a three-dimensional trajectory line that fits the surface shape of the sub-region and satisfies the laser spot overlap ratio. Compared with existing technologies, this invention comprehensively considers the curvature characteristics of the component and the constraints of the galvanometer system, achieving adaptive segmentation of the strengthened region, improving algorithm efficiency and segmentation rationality. The overall method has good versatility and automation, adapting to ultra-thin components of different shapes and sizes, and providing effective technical support for intelligent laser-strengthened processing.

[0074] In one embodiment, such as Figures 1 to 5 As shown, a method for segmenting the strengthened region in ultrafast laser strengthening of ultrathin components includes the following steps:

[0075] S1. Acquire 3D point cloud data

[0076] First, the 3D vision module is used to scan the ultrathin component to obtain initial point cloud data. Optical parameters such as exposure gain are adjusted in the 3D vision module software to achieve an ideal acquisition state for the optical field of view on the ultrathin component surface. Then, the 3D vision module is controlled to scan the surface of the ultrathin component, acquiring and saving the 3D point cloud of the surface to be strengthened.

[0077] S2, Point Cloud Preprocessing

[0078] Point cloud preprocessing includes point cloud filtering, point cloud downsampling, and point cloud normal estimation.

[0079] Point cloud filtering employs a statistical outlier removal method. The number of neighboring points is set to k, and the average distance from each point to its k nearest neighbors is calculated. The average distance distribution of all points is statistically analyzed, and the mean μ and standard deviation σ are calculated. Points with a distance greater than μ + 2σ are marked as noise points and removed.

[0080] Downsampling employs a voxel mesh method, dividing the point cloud space into a regular cubic mesh, where all points within each voxel are replaced by its centroid. Downsampling reduces the amount of point cloud data while preserving the main geometric features of the components.

[0081] Normal estimation employs covariance analysis. For each point pi, its q nearest neighbors are searched, forming a neighborhood point set N(pi). The covariance matrix of the neighborhood point set is then calculated. .in Let the centroid of the neighborhood point set be . Let be a point within the neighborhood, and q be the number of nearest neighbors during the normal estimation process. Eigenvalue decomposition is performed on the covariance matrix; the eigenvector corresponding to the smallest eigenvalue is the normal direction at that point. To ensure consistency in the normal direction, the minimum spanning tree method is used for normal direction adjustment.

[0082] S3, Parametric Space Projection

[0083] Principal component analysis is performed on the preprocessed point cloud. The covariance matrix of all points is calculated, and its eigenvalues ​​and eigenvectors are obtained. The eigenvector corresponding to the largest eigenvalue λ1 is... The principal direction of the point cloud is represented by the eigenvector corresponding to the second largest eigenvalue λ2. Indicating the secondary direction, the eigenvector corresponding to the smallest eigenvalue λ3. Indicates the principal direction of the normal. With the centroid of the point cloud as the origin, and... , , Establish a local coordinate system for the coordinate axes. Project all points onto the local coordinate system. - From a plane, we obtain a two-dimensional parameter representation. Establish the mapping relationship between parametric coordinates and 3D coordinates as follows: ,in As a reference point on the projection plane, The height is a function obtained from the point cloud data through interpolation.

[0084] S4. Calculate the curvature features of the point cloud.

[0085] Local quadratic surface fitting is performed on the parameterized point cloud. For each point, its n nearest neighbors are selected, and the quadratic surface is fitted using the least squares method: Where n is the number of nearest neighbors in the point cloud curvature feature calculation process. The values ​​of coefficients a, b, c, d, e, and f are obtained through equation solving. The principal curvature is calculated from the quadratic surface coefficients. and : ,in For the mean curvature, It represents Gaussian curvature.

[0086] S5, Adaptive Mesh Generation

[0087] The mesh density function is determined based on curvature characteristics. The baseline mesh size is set to m millimeters, and the mesh size at curvature k is s(k). ,in Based on the standard dimensions, This is an adjustment coefficient. During mesh generation, the maximum size of each mesh region is ensured to not exceed the galvanometer's width, with a certain margin reserved for boundary treatment. The mesh distribution is adjusted through iterative optimization. After the initial mesh is generated according to the density function, the shape factor (aspect ratio) of each mesh is calculated. Mesh with a shape factor greater than a certain threshold is subdivided, and adjacent meshes are smoothly transitioned.

[0088] S6, Grid Region Growth and Merging

[0089] The similarity criteria parameters for region growth are defined: a normal angle threshold, a curvature difference threshold, and a spatial distance threshold. The region starts as a seed grid with the minimum curvature and grows incrementally. For each seed grid, its neighboring grids are searched, and a similarity metric is calculated. ,in The angle between the normals, Due to curvature differences, For spatial distance, , , For weighting coefficients. When When the value exceeds a certain threshold, adjacent grids will be merged into the current region.

[0090] S7. Region Boundary Optimization and Overlapping Region Setting

[0091] B-spline curves are fitted to the boundary points of each region. Control point locations are determined using the least squares method to minimize the sum of squared distances between the fitted curve and the original boundary points. The width of the overlapping region is set, and overlapping bands are generated at the boundaries of adjacent regions by offsetting the boundary curves.

[0092] S8, Scanning path planning within sub-regions

[0093] Based on the laser spot size and overlap ratio requirements, draw equally spaced planar trajectory lines. If the spot diameter is Φ and the required overlap ratio is R0%, then the required line spacing of the designed planar trajectory lines is... The graphic is then projected onto the sub-region model in the galvanometer software to obtain a three-dimensional trajectory line graphic that fits the surface shape of the sub-region and satisfies the laser spot overlap rate.

[0094] S9. Output the segmentation result

[0095] The final output includes the geometric information of the divided reinforcement regions, including: parameter representation of the region boundaries, scan path planning within the regions, and recommended process parameters for each region.

[0096] 1. By comprehensively considering curvature characteristics and galvanometer processing area constraints, adaptive segmentation of the enhanced region was achieved, improving segmentation accuracy and rationality;

[0097] 2. Principal component analysis is used for parametric space projection, which simplifies the processing of complex three-dimensional surfaces and improves algorithm efficiency;

[0098] 3. By employing regional growth and merging strategies, the number of segmented regions was effectively reduced, thereby lowering the complexity of path planning between different regions;

[0099] 4. By introducing regional boundary optimization and overlapping area settings, the continuity and uniformity of the reinforcement layer are ensured;

[0100] 5. The overall method has good versatility and automation, and can adapt to ultra-thin components of different shapes and sizes. Example

[0101] This embodiment uses an aero-engine blade, a typical ultra-thin component, as an example to illustrate the implementation process of the method of the present invention in detail. The blade is made of titanium alloy, with a thickness of approximately 2 millimeters, and its surface contains complex curved features.

[0102] S1. Acquire 3D point cloud data

[0103] First, a 3D vision module is used to scan the ultra-thin component to obtain the initial scan point cloud of the blade, such as... Figure 3 As shown, there is some missing point cloud data in the middle part of the blade because there is a triangular protrusion at that location on the blade itself. The protrusion is parallel to the scanning direction, so no point cloud is generated at this location during scanning. In this embodiment, only the already scanned blades are processed, and this location can be ignored. The scanning accuracy is set to 0.05 mm, and approximately 180,000 raw point cloud data points are obtained.

[0104] S2, Point Cloud Preprocessing

[0105] Point cloud preprocessing includes point cloud filtering, point cloud downsampling, and point cloud normal estimation.

[0106] Point cloud filtering employs a statistical outlier removal method. With the neighborhood point count k=50, approximately 5000 noise points are removed after filtering. Figure 3 As shown.

[0107] Downsampling was performed using a voxel grid method with a voxel size of 0.2 mm. After downsampling, the point cloud data size was reduced to 3006 points while preserving the main geometric features of the blade, such as... Figure 3 As shown.

[0108] Normal estimation employs covariance analysis. The number of nearest neighbors, q, is set to 30, forming a neighborhood point set N(pi). The covariance matrix of this neighborhood point set is calculated, and eigenvalue decomposition is performed on the covariance matrix. The eigenvector corresponding to the smallest eigenvalue is the normal direction of that point. To ensure consistency in the normal direction, a minimum spanning tree method is used for adjustment.

[0109] S3, Parametric Space Projection

[0110] Principal component analysis was performed on the preprocessed point cloud. The covariance matrix of all points was calculated, and its eigenvalues ​​and eigenvectors were obtained. The results show that the largest eigenvalue λ1 is 45.32, and the corresponding eigenvector is the first principal direction. Along the blade length direction; the second largest eigenvalue λ2 is 8.76, and the corresponding eigenvector is the second principal direction. Along the blade width direction; the minimum eigenvalue λ3 is 0.28, and the corresponding eigenvector is the principal normal direction. Along the thickness direction of the blade. The centroid coordinates of the point cloud are (125.43mm, 68.57mm, 15.32mm). Taking the centroid of the point cloud as the origin, and... , , Establish a local coordinate system for the coordinate axes. Project all points onto the local coordinate system. - From the plane, we obtain a two-dimensional parameter representation of each point. Establish the mapping relationship between parametric coordinates and three-dimensional coordinates as follows: .

[0111] S4. Calculate the curvature features of the point cloud.

[0112] Perform local quadratic surface fitting on the parameterized point cloud. Set the number of nearest neighbors in this step to n=30. For each point, select its 30 nearest neighbors and fit the quadratic surface using the least squares method: Taking a typical point in the middle of the blade as an example, the quadratic surface coefficients obtained by solving the least squares method at this point are: a = -0.0045 mm^(-1), b = 0.0012 mm^(-1), c = -0.0082 mm^(-1), d = 0.0015 mm, e = -0.0008 mm, f = 0.0 mm. The principal curvature is then calculated from these quadratic surface coefficients. and The value is: k1 = 0.0089 mm -1 k2 = -0.0163 mm -1 ,in , is the mean curvature. , where is the Gaussian curvature.

[0113] S5, Adaptive Mesh Generation

[0114] The mesh density function is determined based on curvature characteristics. The baseline mesh size is set to 1 mm, and the mesh size s(k) at curvature k is... In this embodiment, we take =10. This results in a mesh size of approximately 0.1-0.2 mm in high curvature regions and 0.8-1 mm in low curvature regions. Considering the working range constraint of the galvanometer, it is assumed that the effective working range of the galvanometer system used in this embodiment is 50×50 mm. During mesh generation, the maximum size of each mesh region is ensured not to exceed 40 mm, with a 10 mm margin reserved for boundary processing. The mesh distribution is adjusted through iterative optimization. After the initial mesh is generated according to the density function, the shape factor (aspect ratio) of each mesh is calculated. Mesh with a shape factor greater than 3 is subdivided, and a smooth transition is performed for adjacent meshes whose size difference is greater than 2 times. After three iterations of optimization, 389 mesh elements are obtained. Figure 4 This shows the distribution of the size of these 389 grid cells as the local curvature changes.

[0115] observe Figure 4 As can be seen, the scatter plot strictly follows the density function law of s(k). Specifically, in the gently curving region corresponding to the blade body (curvature |k| < 0.2 mm^(-1)), the grid distribution is relatively dense and the size is maintained between 0.8 mm and 1.0 mm; while in the high curvature region corresponding to the blade edge (curvature |k| > 0.6 mm^(-1)), the grid size adaptively shrinks to around 0.1 mm. The slight deviation between the scatter plot and the theoretical curve in the figure reflects that the algorithm performs secondary fine-tuning of the grid size during the iteration process to meet the constraints of the grid shape factor not exceeding 3 and the smooth transition of adjacent grid sizes. This distribution characteristic proves that the proposed method can ensure the accuracy of capturing complex curvature details while also taking into account the overall efficiency of processing area division.

[0116] S6, Grid Region Growth and Merging

[0117] The similarity criteria parameters for region growing are set as follows: normal angle threshold of 15 degrees, curvature difference threshold of 0.05 mm^(-1), and spatial distance threshold of 2 mm. Similarity measurement formula. middle, , , The three weighting coefficients are set to 0.4, 0.3, and 0.3 respectively. The threshold for S is set to 0.7. When S is greater than 0.7, adjacent grids are merged into the current region. When S is less than 0.7, adjacent grids are classified into a new region.

[0118] During the region growth process, the similarity metric S between the seed grid and its neighboring grids is calculated in real time to determine whether to merge them. Figure 5 This shows the similarity score distribution of the 389 initial grid cells during the merging process. From... Figure 5 As can be observed, the curve exhibits a clear alternation of peaks and troughs. During growth within the same reinforcement region, adjacent meshes have small normal angles and low curvature differences, resulting in a stable similarity score S between 0.72 and 0.95 (above the set threshold of 0.7), thus being continuously merged into a single sub-region. When the growth search reaches a point of geometric abrupt change, the sharp change in normal and curvature weights causes the S value to rapidly drop below the threshold of 0.7. At this point, the algorithm automatically stops growing the current region and starts growing the next region. Ultimately, the seven significant threshold drops in the curve effectively divide the 389 initial meshes into eight initial reinforcement regions. This dynamic discrimination mechanism based on similarity scores ensures the applicability of process parameters within a single reinforcement region while minimizing the total number of segmented regions, providing a concise input for subsequent path planning.

[0119] S7. Region Boundary Optimization and Overlapping Region Setting

[0120] For each region's boundary points, perform B-spline curve fitting. Select a 3rd-order B-spline, with the number of control points being 1 / 5 of the number of boundary points. Determine the control point locations using the least squares method to minimize the sum of squared distances between the fitted curve and the original boundary points.

[0121] The laser spot diameter Φ = 20 μm, and the width of the overlapping region is set to 30 micrometers, approximately 1.5 times the laser spot diameter. An overlapping band is generated at the boundary of adjacent regions by offsetting the boundary curve.

[0122] S8, Scanning path planning within sub-regions

[0123] For each sub-region, a scanning path is planned within that sub-region. Based on the laser spot size and overlap ratio requirements, equally spaced planar trajectory lines are drawn. Given a spot diameter Φ = 20 μm and an overlap ratio R0% = 50%, the required line spacing for the designed planar trajectory lines is... Then, the divided sub-region files are imported sequentially into the galvanometer software, followed by the designed planar trajectory line graphic file. The graphic is then projected onto each sub-region model in the software to obtain a three-dimensional trajectory line graphic that fits the surface shape of the sub-region and satisfies the laser spot overlap rate.

[0124] S9. Output the segmentation result

[0125] The final output is the geometric information of the 8 enhanced regions after division, such as... Figure 3 As shown, Figure 3 This embodiment demonstrates the implementation process of region segmentation using the method of this invention on a typical ultra-thin component, an aero-engine blade. The process includes filtering the initial scanned point cloud of the blade, downsampling the filtered point cloud, and then segmenting the downsampled point cloud into regions to obtain the region segmentation result. Besides... Figure 3 The final output also includes: parameter representation of the region boundaries, scan path planning within the region, and recommended process parameters for each region.

[0126] Furthermore, in the curvature perception stage, this invention accurately extracts the principal curvature, Gaussian curvature, and average curvature of each surface point by performing high-precision normal estimation and local quadratic surface fitting on the point cloud, providing a geometric basis for subsequent adaptive meshing. Secondly, in the area-constrained adaptive meshing, a dynamic mesh density function with the inverse of curvature as the weight is introduced, and the size of a single mesh is strictly limited to not exceeding the effective processing area of ​​the galvanometer (e.g., 50×50). The process involves several key steps. First, a multi-dimensional similarity metric is employed in the region fusion stage. This metric considers the consistency of the fusion normal direction, curvature similarity, and spatial proximity. Through region growth and merging strategies, the initial fine-grained mesh is aggregated into semantically coherent and technologically consistent reinforced sub-regions, significantly reducing the number of regions and path jump frequency, thus improving processing efficiency. Second, in the boundary optimization stage, B-spline curves are used to smoothly fit the region boundaries, eliminating jagged edges. The overlap band width is set based on the laser spot diameter (e.g., 1.5 times the spot diameter), creating a controllable overlap area between adjacent regions. This ensures continuous coverage of the reinforced layer while avoiding thermal damage caused by excessive energy accumulation. Finally, a planar trajectory line is designed based on the spot diameter and overlap ratio requirements. This planar trajectory line is then projected onto each sub-region model in the galvanometer software to obtain a 3D trajectory line, completing the scanning path planning within the sub-region. The aforementioned technical methods deeply integrate geometric modeling, equipment physical constraints, and process knowledge, forming an intelligent and adaptive region segmentation method for ultrafast laser strengthening scenarios. This not only improves the uniformity of strengthening and the robustness of the process, but also significantly enhances the system's universal adaptability to ultrathin components with different shapes, sizes, and curvature distributions, providing key technical support for achieving high-quality, high-efficiency, and fully automated laser surface strengthening.

[0127] The basic principles of this application have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in this application are merely examples and not limitations, and should not be considered as essential features of each embodiment of this application. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not limit the application to the necessity of employing the aforementioned specific details for implementation.

[0128] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of this application to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations thereof.

Claims

1. A method for segmenting the strengthened region in ultrafast laser strengthening of ultrathin components, characterized in that, It includes, S1, acquire three-dimensional point cloud data of the surface of the ultrathin component to be strengthened; S2, preprocessing the three-dimensional point cloud data, including point cloud filtering, downsampling and normal estimation; S3, based on the principal component analysis method, performs parametric space projection on 3D point cloud data to establish a parametric representation of the components; S4 calculates the curvature features of each point in the 3D point cloud data, including principal curvature, Gaussian curvature and mean curvature. S5, adaptive mesh generation based on curvature characteristics and galvanometer processing area constraints; S6, perform region growth and merging on the divided mesh to generate the initial reinforced region; S7, based on the initial requirement for the continuity of the enhanced region, optimizes the region boundary and sets the overlapping region; S8, plans the scanning path within the sub-region; S9 outputs enhanced region segmentation results and scanning path planning scheme.

2. The method as described in claim 1, characterized in that, Preferably, in step S2, the preprocessing includes: using a statistical outlier removal method to filter the point cloud, by calculating the average distance from each point to its k nearest neighbors, identifying points whose distance exceeds a set threshold as noise points and removing them; using a voxel grid method to downsample the point cloud, dividing the point cloud space into a regular voxel grid, retaining a representative point in each voxel; and using a neighborhood covariance analysis-based method to estimate the normal, by performing principal component analysis on the neighborhood point set of each point to obtain the normal direction.

3. The method as described in claim 1, characterized in that, In step S3, the parameter space projection includes: performing principal component analysis on the preprocessed 3D point cloud data, calculating the covariance matrix of the 3D point cloud data and solving for eigenvalues ​​and eigenvectors; establishing a local coordinate system with the eigenvector corresponding to the largest eigenvalue as the first principal direction and the eigenvector corresponding to the second largest eigenvalue as the second principal direction; projecting the 3D point cloud onto a 2D parameter plane composed of the first two principal directions; and establishing a bidirectional mapping relationship between the parameter plane coordinates and the 3D spatial coordinates as the parameterized representation of the component.

4. The method as described in claim 1, characterized in that, In step S4, for each point in the 3D point cloud data, select its k nearest neighbors, fit a local quadratic surface using the least squares method, calculate the average curvature and Gaussian curvature from the coefficients of the fitted surface, and then calculate the principal curvature.

5. The method as described in claim 1, characterized in that, In step S5, adaptive mesh generation includes: determining the mesh density function based on the curvature distribution; limiting the maximum size of each mesh region to no more than 80% of the effective processing area of ​​the galvanometer based on the galvanometer processing area constraint; and adjusting the mesh distribution through iterative optimization to ensure that the size ratio of adjacent meshes does not exceed a set threshold.

6. The method as described in claim 1, characterized in that, In step S6, region growth and merging include: setting similarity criteria, including normal angle threshold, curvature difference threshold and spatial distance threshold; starting with the grid with the smallest curvature as a seed for region growth; calculating the similarity metric of adjacent grids; and merging adjacent grids into the current region when the similarity metric is greater than the set threshold to generate an initial enhanced region.

7. The method as described in claim 1, characterized in that, In step S7, the region boundary optimization includes: fitting discrete boundary points with B-spline curves to generate smooth region boundaries, and setting overlapping zones between adjacent regions according to the laser spot size and the requirements of the strengthening layer thickness.

8. The method as described in claim 1, characterized in that, In step S8, the scanning path planning within the sub-region includes: drawing planar trajectory lines with equal spacing according to the laser spot size and overlap rate requirements, and then projecting the graphic onto the sub-region model in the galvanometer software to obtain a three-dimensional trajectory line graphic that fits the surface shape of the sub-region and satisfies the laser spot overlap rate.

9. The method as described in claim 1, characterized in that, In step S9, the output enhancement region segmentation result includes the parameter representation of the region boundary, the scanning path planning within the region, and the recommended process parameters for each region. The recommended process parameters for each region include the allocation of corresponding laser process parameters to each region based on the region's geometric features and enhancement requirements, generating an enhancement scheme containing region geometric information and process parameters.

10. A system for implementing the method for segmenting the strengthened region in ultrafast laser strengthening of ultrathin components as described in any one of claims 1-9, characterized in that, It includes, The 3D scanning module is used to acquire point cloud data of the component to be reinforced; The data processing module is used to perform enhanced region segmentation; The laser enhancement module is used to perform laser enhancement processing based on the segmentation results; The control module coordinates the operation of the 3D scanning module, data processing module, and laser strengthening module, and monitors the strengthening process.

Images