A method for multi-component positioning and cross-section extraction in electric arc additive manufacturing based on point cloud

Abstract

This invention discloses a point cloud-based method for multi-component positioning and cross-section extraction in arc additive manufacturing, relating to the fields of additive manufacturing and digital inspection. The method includes: acquiring an original 3D point cloud containing a substrate and multiple additive components; identifying the substrate's principal plane to construct a global spatial reference, resulting in a globally aligned point cloud; separating the substrate point cloud and component point clouds to obtain a multi-component point cloud set; performing spatial connectivity analysis and segmentation on the multi-component point cloud set to obtain multiple independent component point clouds corresponding to a single component; performing principal direction identification and attitude standardization processing to obtain component point clouds under standard attitudes; and based on the component point clouds under standard attitudes, determining the cross-section position and extracting the point set within the cross-section's neighborhood to generate cross-section point cloud data characterizing the component's cross-sectional morphology. This method automates the processing from the original point cloud to standardized cross-sectional features, improving the accuracy and consistency of multi-component identification and cross-section extraction in arc additive manufacturing.

Description

Technical Field

[0001] This invention relates to the fields of additive manufacturing and digital inspection, and more specifically to a method for multi-component positioning and cross-section extraction in arc additive manufacturing based on point cloud. Background Technology

[0002] Wire Arc Additive Manufacturing (WAAM) has been widely used in aerospace, energy equipment, and large structural component manufacturing due to its advantages such as high forming efficiency, high material utilization, and suitability for manufacturing medium and large metal components. In actual production, in order to improve forming efficiency and equipment utilization, multiple components are often formed simultaneously on the same substrate. These formed components need to be inspected using three-dimensional digital inspection technology to obtain their size, morphology, and cross-sectional characteristics, so as to provide a basis for subsequent quality evaluation, process optimization, and structural performance analysis.

[0003] However, in multi-component arc additive manufacturing scenarios, existing detection technologies first face the problem of overall tilt in the raw point cloud data. Due to factors such as sensor installation tilt angle and scanning posture deviation during the scanning acquisition process, the acquired three-dimensional point cloud data often presents an overall tilt state. Existing processing methods usually rely on manual identification of the substrate position or preset fixed height thresholds, making it difficult to stably establish a unified spatial reference benchmark. When the substrate undergoes overall warping deformation due to heat input, this method, which relies on manual experience or simple thresholds, is even more difficult to accurately identify the main plane of the substrate, making it difficult to guarantee the accuracy of subsequent component height analysis and cross-section extraction.

[0004] Furthermore, after establishing the spatial reference, another technical challenge is how to accurately separate each independent component from the point cloud data containing the substrate and multiple components. Existing technologies are mostly focused on single-component scenarios or rely on manual methods to trim and segment the point cloud. In the case of multiple components, due to factors such as the stacking of melt channels, sputtering adhesion, or close distance between components during the additive manufacturing process, different components may exhibit local connections or spatial interference in the point cloud data. When faced with such complex distribution characteristics, existing methods often require manual intervention to select regions and separate adhesions. The operation process is cumbersome and highly subjective, making it difficult to meet the needs of automated detection and batch analysis.

[0005] Furthermore, when performing cross-sectional analysis on additive components, existing methods typically require manual specification of the cross-sectional position and orientation. The results are highly dependent on the operator's experience. When the placement orientation of the component on the substrate is random, and different components exhibit rotational deviations within the plane, the comparability and accuracy of the cross-sectional extraction results cannot be guaranteed if the orientation of each component cannot be standardized. More importantly, existing technologies lack the ability to adaptively determine the orientation of key cross-sections based on the component's own geometric characteristics. This makes it difficult to obtain standardized cross-sectional data with a consistent orientation in batch testing scenarios, thereby affecting the objective evaluation of the component's forming quality and geometric characteristics.

[0006] Therefore, how to design a point cloud-based method for multi-component positioning and cross-section extraction in arc additive manufacturing, and realize the automated extraction of standardized cross-sectional data of multiple components from the original three-dimensional point cloud, so as to provide efficient and objective geometric data support for digital inspection and quality evaluation of arc additive manufacturing, is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0007] In view of this, the present invention provides a point cloud-based method for multi-component positioning and cross-section extraction in arc additive manufacturing. It aims to solve the problems of low processing efficiency, high subjectivity, and poor repeatability caused by the reliance on manual operation for point cloud trimming, region selection, and attitude adjustment in multi-component scenarios in existing technologies, as well as the defects of cross-sectional features that are difficult to compare and quantify due to inconsistent component attitudes. The method achieves automated and highly consistent extraction from the original 3D point cloud to standardized cross-sectional data of multiple components.

[0008] To achieve the above objectives, the present invention adopts the following technical solution:

[0009] A point cloud-based method for multi-component positioning and cross-section extraction in arc additive manufacturing includes the following steps: S1. Obtain the original three-dimensional point cloud containing the substrate and multiple additive components, and construct a global spatial reference based on the main plane of the substrate to obtain the global aligned point cloud. S2. Based on the height distribution characteristics of the global aligned point cloud, separate the substrate point cloud and the component point cloud to obtain a multi-component point cloud set; S3. Perform spatial connectivity analysis and segmentation on the multi-component point cloud set to obtain multiple independent component point clouds corresponding to a single component; S4. Perform main direction identification and attitude standardization processing on the point cloud of each independent component to obtain the component point cloud under the standard attitude. S5. Based on the component point cloud under the standard posture, determine the cross-section position and extract the point set in the neighborhood of the cross-section to generate cross-section point cloud data for characterizing the cross-section morphology of the component.

[0010] Preferably, S1 includes: Using the random sampling consensus algorithm to extract data from the original 3D point cloud Identify the main plane of the substrate ,in For the i-th point cloud sampling point, Let (a, b, c) be its spatial coordinates, (a, b, c) be the normal vector of the principal plane of the substrate, and d be the plane intercept parameter. Based on the normal vector (a, b, c) and the Z-axis of the preset global coordinate system, a rotation matrix is ​​constructed. Translation vector Perform rigid body transformation on the original 3D point cloud P Obtain the globally aligned point cloud. Make the main plane of the substrate In the global coordinate system, z=0.

[0011] Preferably, the random sampling consensus algorithm is used to obtain data from the original 3D point cloud. Identify the main plane of the substrate include: Three non-collinear points are randomly selected from the original 3D point cloud P to construct a candidate plane. Let the equation of this candidate plane be... ; Calculate all points Distance to the candidate plane The statistical distance is less than the preset distance threshold. The number of interior points; Through multiple random sampling iterations, the candidate plane with the most interior points is selected as the final substrate principal plane. .

[0012] Preferably, S2 includes: Extracting the globally aligned point cloud The height values ​​of all points constitute a height set. ; Statistical analysis is performed on the height set Z to determine the substrate height threshold. and will satisfy all The points are incorporated into the component point cloud set. This enables automatic separation of the substrate point cloud.

[0013] Preferably, the statistical analysis includes: constructing a height histogram and identifying its main peak position, and determining a substrate height threshold based on the height interval corresponding to the main peak. .

[0014] Preferably, S3 includes: For component point cloud set Perform spatial connectivity analysis based on a preset distance threshold. Construct a connectivity graph between points to aggregate the component point cloud. Initially divided into multiple subsets , making ,in This represents the point cloud corresponding to the k-th candidate component, where K is the number of candidate components. This is the distance threshold used to determine spatial connectivity between two points; The point cloud of the candidate components is processed according to preset geometric constraint parameters. The point sets that do not meet the criteria are filtered out to obtain the final point cloud of multiple independent components. .

[0015] Preferably, when the candidate component point cloud When there is local adhesion or spatial overlap, a secondary segmentation step is also included: dividing the candidate component point cloud into segments. The projection onto the horizontal plane yields a two-dimensional projection point set, which is then subdivided using a density-based spatial clustering algorithm to separate adjacent or connected components.

[0016] Preferably, S4 includes: Point cloud of any independent component Projecting onto a horizontal plane yields a two-dimensional set of projection points. ; Based on the two-dimensional projection point set Given the geometric distribution, calculate its principal direction vector. ; According to the main direction vector Construct rotation matrix Point clouds of independent components Perform attitude standardization transformation to obtain the transformed point coordinates And component point clouds in standard pose .

[0017] Preferably, S5 includes: For component point clouds in standard posture Determine its length range along the main direction. ,in and These represent the minimum and maximum coordinates of the component's point cloud in the X direction, respectively, and the center height of the cross-section is selected. ; According to the preset cross-sectional thickness parameters Construct the cross-sectional neighborhood and extract the point set. This serves as the cross-sectional point cloud data for the component.

[0018] Preferably, the method further includes: transforming the cross-sectional point cloud data to a two-dimensional plane through projection, and correcting the two-dimensional projection through a rotation matrix so that the height direction of the cross-sectional profile remains upward in the two-dimensional image, thereby obtaining a standardized two-dimensional cross-sectional profile data set for all components.

[0019] As can be seen from the above technical solution, compared with the prior art, the technical solution of the present invention has the following beneficial effects: 1. This method takes the original 3D point cloud as input and constructs a complete automated processing flow through sequentially executed steps such as global spatial benchmark reconstruction, automatic separation of substrate point cloud, multi-component spatial segmentation, single-component posture standardization, and cross-sectional point set generation. The entire process does not require manual intervention for point cloud clipping, region selection, or posture adjustment, effectively avoiding the problems of strong subjectivity and poor repeatability caused by reliance on manual operation in traditional methods, and improving the efficiency and consistency of multi-component point cloud data processing.

[0020] 2. It comprehensively utilizes statistical distribution analysis and spatial connectivity analysis. Based on the automatic removal of substrate point clouds, it segments the point cloud data by combining distance threshold clustering and density clustering. Even under complex conditions such as local adhesion, spatial proximity, or splash noise interference between components, this method can still achieve stable separation of multiple components. Furthermore, it filters the segmentation results by combining preset geometric constraint parameters, thereby ensuring the accuracy and reliability of component identification under different forming conditions.

[0021] 3. By identifying the main orientation and standardizing the attitude of each independent component, the spatial attitude differences caused by the random placement angle of the components on the substrate are eliminated, and a unified local reference benchmark is established for subsequent cross-sectional analysis. On this basis, the cross-sectional position is adaptively determined based on the geometric scale of the component itself, and cross-sectional point cloud data with consistent orientation are extracted. This enables quantitative comparison and analysis of the cross-sectional morphology features of different components in the same batch or between different batches in a unified coordinate system, providing an objective geometric data basis for the stability assessment of additive manufacturing process and the inspection of forming quality. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0023] Figure 1 A flowchart of a method for positioning and extracting multiple components in arc additive manufacturing based on point cloud, provided in an embodiment of the present invention; Figure 2 This is a global attitude correction effect diagram provided by an embodiment of the present invention; Figure 3 This is a diagram illustrating the substrate cutting and separation effect provided in an embodiment of the present invention. Figure 4 This is a diagram illustrating the connected component clustering and segmentation effect provided in an embodiment of the present invention. Figure 5 This is a diagram illustrating the standardized posture effect of a single component provided in an embodiment of the present invention. Figure 6 This is a schematic diagram of cross-sectional positioning provided in an embodiment of the present invention; Figure 7 The image shows the effect of generating a standardized cross-sectional image as provided in an embodiment of the present invention. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] like Figure 1 As shown, this embodiment provides a method for multi-component positioning and cross-section extraction in arc additive manufacturing based on point clouds, including the following steps: S1. Obtain the original three-dimensional point cloud containing the substrate and multiple additive components, and construct a global spatial reference based on the main plane of the substrate to obtain the global aligned point cloud. S2. Based on the height distribution characteristics of the global aligned point cloud, separate the substrate point cloud and the component point cloud to obtain a multi-component point cloud set; S3. Perform spatial connectivity analysis and segmentation on the multi-component point cloud set to obtain multiple independent component point clouds corresponding to a single component; S4. Perform main direction identification and attitude standardization processing on the point cloud of each independent component to obtain the component point cloud under the standard attitude. S5. Based on the component point cloud under the standard posture, determine the cross-section position and extract the point set in the neighborhood of the cross-section to generate cross-section point cloud data for characterizing the cross-section morphology of the component.

[0026] By constructing a fully automated processing step from global benchmark reconstruction, automatic substrate separation, multi-component spatial segmentation, single-component posture standardization to cross-sectional point set generation, it achieves automatic processing of 3D point cloud data containing substrates and multiple additive components without relying on manual intervention. It can stably and accurately identify and separate multiple components in complex forming scenarios where components have local adhesion, spatial proximity, or inconsistent postures, and obtain standardized cross-sectional point cloud data with unified spatial reference and comparability, providing an objective geometric data basis for forming quality inspection and process evaluation in arc additive manufacturing.

[0027] The following provides further explanation of each step and related features in the above method; In this embodiment, S1, an original three-dimensional point cloud containing a substrate and multiple additive components is obtained, and a global spatial reference is constructed based on the original three-dimensional point cloud to identify the main plane of the substrate, thereby obtaining a globally aligned point cloud; including: Using the random sampling consensus algorithm to extract data from the original 3D point cloud Identify the main plane of the substrate ,in For the i-th point cloud sampling point, Let (a, b, c) be its spatial coordinates, (a, b, c) be the normal vector of the principal plane of the substrate, and d be the plane intercept parameter. Based on the normal vector (a, b, c) and the Z-axis of the preset global coordinate system, a rotation matrix is ​​constructed. Translation vector Perform rigid body transformation on the original 3D point cloud P Obtain the globally aligned point cloud. Make the main plane of the substrate In the global coordinate system, z=0.

[0028] Furthermore, a random sample consensus algorithm is used to analyze the original 3D point cloud. Identify the main plane of the substrate include: Three non-collinear points are randomly selected from the original 3D point cloud P to construct a candidate plane. Let the equation of this candidate plane be... ; Calculate all points Distance to the candidate plane The statistical distance is less than the preset distance threshold. The number of interior points; Through multiple random sampling iterations, the candidate plane with the most interior points is selected as the final substrate principal plane. .

[0029] In the specific implementation process, the original 3D point cloud can be obtained through a laser scanner or structured light scanning equipment. During scanning, it is essential to ensure complete coverage of the substrate area and all additive components. Due to the influence of the equipment installation tilt angle or component placement posture during scanning, the original point cloud is usually tilted overall, which can lead to errors when directly performing height analysis. Therefore, this step uses the Random Sample Consensus (RANSAC) algorithm to identify the substrate principal plane from the original 3D point cloud. The advantage of this algorithm is that it can effectively eliminate interference from component point clouds and noise points. Even when the substrate has slight thermal deformation, it can still stably fit the principal plane representing the substrate. In multiple random sampling iterations, the distance threshold... The value is set from 0.5mm to 2mm based on the point cloud density and scanning accuracy to ensure the accuracy of interior point determination; like Figure 2 As shown, after plane recognition is completed, a rotation matrix is ​​constructed to align the substrate normal vector with the global Z-axis, and the substrate plane is zeroed through translation transformation, thereby establishing a unified spatial reference for all subsequent geometric processing. After global attitude correction, the original tilted point cloud is uniformly aligned to the z=0 horizontal plane, establishing a stable spatial reference for subsequent processing.

[0030] In this embodiment S2, based on the height distribution characteristics of the globally aligned point cloud, the substrate point cloud and the component point cloud are separated to obtain a multi-component point cloud set; including: Extracting the globally aligned point cloud The height values ​​of all points constitute a height set. ; Statistical analysis is performed on the height set Z to determine the substrate height threshold. and will satisfy all The points are incorporated into the component point cloud set. This enables automatic separation of the substrate point cloud.

[0031] Further statistical analysis includes: constructing a height histogram and identifying its main peak position, and determining the substrate height threshold based on the height interval corresponding to the main peak. .

[0032] In the globally aligned point cloud, the substrate region, due to its large area and dense point count, typically exhibits a distinct single-peak distribution on the height histogram. Meanwhile, the component point cloud is distributed across different height ranges above the substrate. The substrate height threshold is determined by constructing a histogram of the height set Z and identifying its dominant peak position. The group interval of the histogram is adaptively set according to the height range and number of points in the point cloud. The main peak position can be identified using a peak detection algorithm, with the maximum height value corresponding to the main peak or its upper boundary used as the substrate height threshold. , with a height greater than All points are included in the component point cloud set, thereby realizing the automatic separation of the substrate point cloud and the component point cloud; It does not require a preset fixed height threshold and can adaptively process point cloud data under different substrate thicknesses or scanning postures, thus improving the method's versatility. Figure 3 As shown, the substrate point cloud was effectively cut and removed, and the remaining point cloud contained only additive components and a small amount of spatter noise, providing clean input data for subsequent multi-component segmentation.

[0033] In this embodiment, S3, spatial connectivity analysis and segmentation are performed on the multi-component point cloud set to obtain multiple independent component point clouds corresponding to a single component; including: For component point cloud set Perform spatial connectivity analysis based on a preset distance threshold. Construct a connectivity graph between points to aggregate the component point cloud. Initially divided into multiple subsets , making ,in This represents the point cloud corresponding to the k-th candidate component, where K is the number of candidate components. This is the distance threshold used to determine spatial connectivity between two points; The point cloud of the candidate components is processed according to preset geometric constraint parameters. The point sets that do not meet the criteria are filtered out to obtain the final point cloud of multiple independent components. ; Furthermore, when the candidate component point cloud When there is local adhesion or spatial overlap, a secondary segmentation step is also included: dividing the candidate component point cloud into segments. The projection onto the horizontal plane yields a two-dimensional projection point set, which is then subdivided using a density-based spatial clustering algorithm to separate adjacent or connected components.

[0034] This step first performs spatial connectivity analysis on the component point cloud based on a preset distance threshold, initially segmenting the overall point cloud into multiple candidate component subsets. To address local adhesion issues caused by melt flow stacking, spatter adhesion, or excessively close component spacing during additive manufacturing, a secondary segmentation strategy based on planar projection is further introduced: candidate components with adhesion are projected onto a horizontal plane and subdivided using a density-based spatial clustering algorithm. Subsequently, the segmentation results are filtered using preset geometric constraint parameters (such as minimum height and minimum span), eliminating invalid point sets such as spatter noise and substrate residue, ultimately obtaining multiple independent component point clouds corresponding to the actual additively manufactured components. Figure 4 As shown, this step, through a combination of connectivity analysis and density clustering, can accurately separate even locally bonded components into independent subsets of the point cloud.

[0035] In this embodiment S4, the main direction identification and attitude standardization processing are performed on the point cloud of each independent component to obtain the component point cloud under the standard attitude; including: Point cloud of any independent component Projecting onto a horizontal plane yields a two-dimensional set of projection points. ; Based on the two-dimensional projection point set Given the geometric distribution, calculate its principal direction vector. The principal direction vector at this location By calculating the two-dimensional projection point set The minimum directed bounding box is obtained; According to the main direction vector Construct rotation matrix Point clouds of independent components Perform attitude standardization transformation to obtain the transformed point coordinates And component point clouds in standard pose .

[0036] Because the orientation of components on the substrate is random, and different components exhibit rotational deviations in the horizontal plane, this step projects the point cloud of each individual component onto the horizontal plane to obtain a two-dimensional projection point set. The principal direction vector of the component is determined by calculating the minimum directed bounding box (OBB) of this projection point set. Based on this principal direction, a rotation matrix is ​​constructed to perform a rotation transformation on the component point cloud, aligning the component's major axis with the X-axis or Y-axis of the global coordinate system. Simultaneously, the component's centroid is translated to the origin of the coordinate system. For example... Figure 5 As shown, through this attitude standardization process, after the main direction is identified and rotated, the major axis of each independent component is aligned with the coordinate axis and the centroid is located at the origin, thus establishing a unified geometric reference for subsequent cross-sectional analysis.

[0037] In this embodiment S5, based on the component point cloud under the standard posture, the cross-section position is determined and the point set in the neighborhood of the cross-section is extracted to generate cross-section point cloud data for characterizing the cross-section morphology of the component; including: For component point clouds in standard posture Determine its length range along the main direction. ,in and These represent the minimum and maximum coordinates of the component's point cloud in the X direction, respectively, and the center height of the cross-section is selected. The main direction at this location is determined by the component point cloud under the aforementioned standard attitude. The coordinate axis direction is determined as follows: when the component's scale in the X direction is greater than its scale in the Y direction, the cross-section normal vector is (1,0,0); otherwise, it is (0,1,0). According to the preset cross-sectional thickness parameters Construct the cross-sectional neighborhood and extract the point set. This serves as the cross-sectional point cloud data for the component.

[0038] After the component's attitude is standardized, this step adaptively determines the cross-sectional orientation based on the component's geometric scale relationship in the standardized coordinate system: when the component's scale in the X direction is greater than its scale in the Y direction, the cross-sectional normal vector is (1,0,0); otherwise, it is (0,1,0); subsequently, as follows... Figure 6 As shown, with the center of the component's length along the main direction as the reference position, the cross-section position is adaptively selected at the midpoint of the component's length direction, ensuring that the cross-section reflects the overall geometric characteristics of the component. A spatial neighborhood is constructed using preset cross-section thickness parameters, and the point cloud within this neighborhood is extracted as the cross-section point set. This adaptive cross-section positioning strategy eliminates the need for manual specification of the cross-section position and orientation, automatically determining a representative cross-section based on the component's own geometric characteristics. The obtained cross-section point cloud data has a unified spatial orientation and can be further used for geometric feature analysis such as two-dimensional contour fitting and dimensional measurement, providing a standardized data foundation for evaluating the forming quality of additively manufactured components.

[0039] To further facilitate analysis and visualization, the cross-sectional point cloud data can be projected onto a two-dimensional plane, and a two-dimensional rotation matrix can be constructed using the angle between the local basis vectors and the global height direction for correction, ensuring that the height direction of the cross-sectional profile remains upward in the two-dimensional image; for example... Figure 7 As shown, a standardized two-dimensional cross-sectional profile data set of all components is finally obtained. After projection transformation and orientation correction, the extracted cross-sectional point cloud generates a standardized two-dimensional cross-sectional image with a unified orientation, which is convenient for subsequent size measurement and morphology analysis.

[0040] This embodiment constructs a fully automated processing method through the sequential execution of the above-mentioned steps, from global spatial reference reconstruction, automatic substrate separation, multi-component spatial segmentation, single-component posture standardization to cross-sectional point set generation. This method uses three-dimensional point cloud as the only input and can automatically process complex point cloud data containing substrate and multiple additive components without manual intervention. It can effectively deal with complex situations in actual forming scenarios such as local adhesion of components, spatial proximity, and inconsistent posture. The standardized cross-sectional point cloud data obtained in the end has a unified spatial reference and comparability, providing an objective and reliable geometric data basis for forming quality inspection, dimensional measurement, and process evaluation of arc additive manufacturing components.

[0041] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple; relevant parts can be referred to the method section.

[0042] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for multi-component positioning and cross-section extraction in arc additive manufacturing based on point clouds, characterized in that, Includes the following steps: S1. Obtain the original three-dimensional point cloud containing the substrate and multiple additive components, and construct a global spatial reference based on the main plane of the substrate to obtain the global aligned point cloud. S2. Based on the height distribution characteristics of the global aligned point cloud, separate the substrate point cloud and the component point cloud to obtain a multi-component point cloud set; S3. Perform spatial connectivity analysis and segmentation on the multi-component point cloud set to obtain multiple independent component point clouds corresponding to a single component; S4. Perform main direction identification and attitude standardization processing on the point cloud of each independent component to obtain the component point cloud under the standard attitude. S5. Based on the component point cloud under the standard posture, determine the cross-section position and extract the point set in the neighborhood of the cross-section to generate cross-section point cloud data for characterizing the cross-section morphology of the component.

2. The method for multi-component positioning and cross-section extraction in arc additive manufacturing based on point cloud as described in claim 1, characterized in that, S1 includes: Using the random sampling consensus algorithm to extract data from the original 3D point cloud Identify the main plane of the substrate ,in For the i-th point cloud sampling point, Let (a, b, c) be its spatial coordinates, (a, b, c) be the normal vector of the principal plane of the substrate, and d be the plane intercept parameter. Based on the normal vector (a, b, c) and the Z-axis of the preset global coordinate system, a rotation matrix is ​​constructed. Translation vector Perform rigid body transformation on the original 3D point cloud P Obtain the globally aligned point cloud. Make the main plane of the substrate In the global coordinate system, z=0.

3. The method for multi-component positioning and cross-section extraction in arc additive manufacturing based on point cloud as described in claim 2, characterized in that, The random sampling consensus algorithm is used to extract data from the original 3D point cloud. Identify the main plane of the substrate include: Three non-collinear points are randomly selected from the original 3D point cloud P to construct a candidate plane. Let the equation of this candidate plane be... ; Calculate all points Distance to the candidate plane The statistical distance is less than the preset distance threshold. The number of interior points; Through multiple random sampling iterations, the candidate plane with the most interior points is selected as the final substrate principal plane. .

4. The method for multi-component positioning and cross-section extraction in arc additive manufacturing based on point cloud as described in claim 1, characterized in that, S2 includes: Extracting the globally aligned point cloud The height values ​​of all points constitute a height set. ; Statistical analysis is performed on the height set Z to determine the substrate height threshold. and will satisfy all The points are incorporated into the component point cloud set. This enables automatic separation of the substrate point cloud.

5. The method for multi-component positioning and cross-section extraction in arc additive manufacturing based on point cloud as described in claim 4, characterized in that, The statistical analysis includes: constructing a height histogram and identifying its main peak position, and determining the substrate height threshold based on the height interval corresponding to the main peak. .

6. The method for multi-component positioning and cross-section extraction in arc additive manufacturing based on point cloud as described in claim 1, characterized in that, S3 includes: For component point cloud set Perform spatial connectivity analysis based on a preset distance threshold. Construct a connectivity graph between points to aggregate the component point cloud. Initially divided into multiple subsets , making ,in This represents the point cloud corresponding to the k-th candidate component, where K is the number of candidate components. This is the distance threshold used to determine spatial connectivity between two points; The point cloud of the candidate components is processed according to preset geometric constraint parameters. The point sets that do not meet the criteria are filtered out to obtain the final point cloud of multiple independent components. .

7. The method for multi-component positioning and cross-section extraction in arc additive manufacturing based on point cloud as described in claim 6, characterized in that, When the candidate component point cloud When there is local adhesion or spatial overlap, a secondary segmentation step is also included: dividing the candidate component point cloud into segments. The projection onto the horizontal plane yields a two-dimensional projection point set, which is then subdivided using a density-based spatial clustering algorithm to separate adjacent or connected components.

8. The method for multi-component positioning and cross-section extraction in arc additive manufacturing based on point cloud as described in claim 1, characterized in that, S4 includes: Point cloud of any independent component Projecting onto a horizontal plane yields a two-dimensional set of projection points. ; Based on the two-dimensional projection point set Given the geometric distribution, calculate its principal direction vector. ; According to the main direction vector Construct rotation matrix Point clouds of independent components Perform attitude standardization transformation to obtain the transformed point coordinates And component point clouds in standard pose .

9. The method for multi-component positioning and cross-section extraction in arc additive manufacturing based on point cloud as described in claim 1, characterized in that, S5 includes: For component point clouds in standard posture Determine its length range along the main direction. ,in and These represent the minimum and maximum coordinates of the component's point cloud in the X direction, respectively, and the center height of the cross-section is selected. ; According to the preset cross-sectional thickness parameters Construct the cross-sectional neighborhood and extract the point set. This serves as the cross-sectional point cloud data for the component.

10. The method for multi-component positioning and cross-section extraction in arc additive manufacturing based on point cloud as described in claim 1, characterized in that, Also includes: The cross-sectional point cloud data is transformed to a two-dimensional plane through projection, and the two-dimensional projection is corrected by a rotation matrix so that the height direction of the cross-sectional profile remains upward in the two-dimensional image, thus obtaining a standardized two-dimensional cross-sectional profile data set for all components.

Images