Three-dimensional model data processing method for multi-degree-of-freedom inkjet printing of curved surfaces

Abstract

The present invention belongs to the field of multi-degree-of-freedom curved surface additive manufacturing, and particularly relates to a three-dimensional model data processing method used for spatial motion planning in multi-degree-of-freedom inkjet printing of curved surfaces. The method comprises: point cloud extraction for a target spatial curved surface in an STL model, target spatial curved surface fitting, two-dimensional path generation for the target spatial curved surface, and inkjet printing spatial path generation for the target spatial curved surface. In the method, a target curved surface spatial point cloud is rapidly extracted by means of spatial point cloud convex hull or concave hull computing, and then the curved surface is fitted on the basis of deep learning to obtain a spatial function; then the extracted target curved surface spatial point cloud is projected onto a two-dimensional plane generation path and is mapped onto a three-dimensional spatial curved surface to obtain a curved surface inkjet printing path. The method can solve existing problems in the field of curved surface additive manufacturing, such as the high difficulty of path planning, the single-point solving speed decreasing with the increase of the number of model vertices and the number of points to be solved during STL model data access, path planning methods for curved surface forming being limited and the difficulty of implementing complex path planning methods for curved surface forming such as curved surface fractals and curved surface contour offset. The present invention provides an efficient and convenient data access mode for complex path planning methods for multi-degree-of-freedom inkjet printing of curved surfaces, and the single-point solving speed does not decrease with the increase of the number of model vertices and the number of points to be solved; the present invention is therefore of great significance for realizing efficient multi-degree-of-freedom inkje

Description

Data processing method for three-dimensional models of multi-degree-of-freedom inkjet printing of curved surfaces Technical Field

[0001] This invention belongs to the field of multi-degree-of-freedom additive manufacturing of curved surfaces, and relates to a path generation method that combines three-dimensional model data processing with deep learning. Specifically, it is a three-dimensional model data processing method for spatial motion planning in multi-degree-of-freedom inkjet forming of curved surfaces. Background Technology

[0002] Multi-degree-of-freedom inkjet printing technology for curved surfaces belongs to additive manufacturing technology. It is a forming technology that uses multi-axis motion mechanisms, such as multi-joint serial robots or parallel robots, as motion mechanisms, and mounts inkjet printing devices at the ends of these mechanisms for follow-up stacking and forming. The forming process involves key technologies such as 3D model data processing, multi-axis motion mechanism control, follow-up control of the inkjet printing device, and precision closed-loop control. Among these, 3D model data processing is used to read 3D STL models and plan forming motions, which has prominent applications in the field of curved surface electronic printing, and is also expanding into areas such as curved surface filament and tape laying, curved surface welding, and curved surface spraying.

[0003] In the process of multi-degree-of-freedom inkjet forming of curved surfaces, the 3D model data processing also faces many difficulties, making it difficult to meet the requirements of complex path planning.

[0004] (1) There is no topological relationship between the model point data stored in the general three-dimensional model file STL of additive manufacturing, and the forming process only focuses on the outer surface of a fixed area and does not require other model data points.

[0005] (2) In order to realize complex motion path planning during the forming process, it is necessary to repeatedly locate data points. The disordered storage of data points means that all data points need to be sorted for each single point to locate the triangular facet where the target point is located, and then the spatial distance z of the target point is calculated by interpolation. Therefore, the single-point solution efficiency during the forming process decreases as the number of model vertices and the number of solution points increase.

[0006] (3) Existing methods all employ data topology reconstruction and data correlation methods, which are difficult to effectively solve the above problems and are also difficult to implement complex path planning methods. Currently, the forming path design method is relatively simple and is still based on the traditional slicing and layering sweeping method, which cannot meet the diverse forming path design requirements of various performance aspects.

[0007] In recent years, with the continuous development of computer technology and the gradual improvement of artificial intelligence frameworks, it has become possible to fit any spatial function using neural networks. This can effectively avoid the repeated sorting and calculation of point data when locating spatial points, and it is robust to the increase in the number of model vertices and solution points. Complex path planning methods can be implemented by repeatedly accessing the model to meet various performance requirements. However, current surface forming motion planning methods are still derived from data-based models and data processing methods, and have not proposed effective solutions to the various difficulties mentioned above. Summary of the Invention

[0008] To address the aforementioned issues, this invention discloses a data processing method for three-dimensional models of multi-degree-of-freedom inkjet forming on curved surfaces. This method overcomes the limitations of existing motion planning methods for multi-degree-of-freedom inkjet forming on curved surfaces, which struggle to implement complex path planning and where single-point solution efficiency decreases with increasing model vertices and solution points. This provides an effective solution for efficient multi-degree-of-freedom inkjet forming on curved surfaces.

[0009] To achieve the above objectives, the technical solution adopted in this proposal is as follows:

[0010] A method for processing data of a three-dimensional model of a curved surface multi-degree-of-freedom inkjet printing process includes the following steps:

[0011] Read the STL model, a common file in additive manufacturing, and obtain the point cloud data of the target spatial surface in the model through convex hull or concave hull calculation methods. Use this spatial point cloud data as the basic dataset for spatial surface fitting.

[0012] Construct a point cloud data processing network model suitable for spatial surface fitting;

[0013] In the established point cloud data processing network model, the planar coordinates and spatial coordinates of the model points in the training base dataset are used to obtain the weights after training.

[0014] Project the target spatial surface onto a two-dimensional plane and generate a two-dimensional path based on the two-dimensional point cloud;

[0015] The obtained two-dimensional path is mapped to a point cloud data processing network model, and the trained weights are used to output a multi-degree-of-freedom inkjet forming spatial path for curved surfaces.

[0016] As a further design of this scheme, the method for reading STL files and obtaining the target surface is as follows: After reading the spatial coordinates of the unordered points on the surface of the STL model in ASCII code format, the spatial point cloud data of the target surface in the model is quickly extracted by convex hull or concave hull calculation in the three-dimensional point cloud processing method. The point cloud data (x,y) is used as input and z is used as label as training dataset.

[0017] As a further design of this scheme, the method for constructing a point cloud data processing network model for spatial surface fitting is as follows: a point cloud segmentation module is built to segment the target surface point cloud into several block point clouds that are easy to fit; a block point cloud rotation module is built to rotate the block point clouds obtained in the point cloud segmentation module to obtain a rigid rotation matrix with the maximum projected area; and a deep neural network fitting module is built to fit the rigidly rotated block point cloud obtained in the block point cloud rotation module.

[0018] As a further design of this scheme, the point cloud segmentation module is used to segment the obtained two-dimensional coordinates into multiple block point clouds according to the projection grid. The construction method is as follows: project the target curved surface point cloud onto the two-dimensional plane, and use the minimum square bounding algorithm to calculate the minimum square bounding of the point cloud. Then, use equal-distance interpolation in the square bounding to divide the point cloud into multiple block point clouds of equal size.

[0019] As a further design of this scheme, the block point cloud rotation module is used to rotate the block point cloud obtained in the point cloud segmentation module, obtain the rigid rotation matrix of the maximum two-dimensional projection, the rotated point cloud data, and the rigid inverse rotation matrix. The construction method is as follows: read in several block point clouds obtained in S2, take the two points with the largest Euclidean distance in each point cloud as the rotation axis and also the X vector of the rigid rotation matrix, take the midpoint O of the two points as the origin of the coordinate system, generate 720 Y vectors in the plane passing through O and perpendicular to the X axis, with the angle difference between the plane intersection lines passing through O and being 0.5°, calculate the Z vector of the rotation matrix by performing the outer product of the X vector and each Y vector, calculate the projected area under the point cloud rotation and iterate the rigid rotation matrix under the maximum projected area, and output the rotated point cloud data and the rigid inverse rotation matrix.

[0020] As a further design of this scheme, a deep neural network fitting module is used to fit the rotated point cloud output by the block point cloud rotation module and add a rigid inverse rotation module. The construction method is as follows: read the rotated point cloud data output by the block point cloud rotation module, capture the spatial topological relationship of the point cloud through a deep neural network, form the mapping relationship between two-dimensional plane points and three-dimensional spatial positions in the curved surface, and output the spatial position directly by inputting the plane path points.

[0021] As a further design of this scheme, the training of the point cloud data processing network model includes the following steps: reading the target surface point cloud data and processing it through the point cloud segmentation module and the block point cloud rotation module to obtain several rotated block point clouds and their rigid inverse rotation matrices; normalizing the input point cloud data, calculating the spatial distance of the current input (x,y) calculated by the network and multiplying it by the rigid inverse rotation matrix to obtain the calculated spatial distance, comparing the calculated spatial distance with the actual spatial distance in the label, and making the calculated spatial distance equal to the actual spatial distance; repeating the second step to optimize and adjust the parameters of each network node of the model, and finally obtaining an optimal solution. The optimal solution makes the calculated spatial distance equal to the actual spatial distance, and this optimal solution is the weight of the point cloud data processing network model.

[0022] As a further design of this scheme, the method for generating a curved surface forming two-dimensional path is as follows: read the target space point cloud and project it into a two-dimensional plane, calculate the minimum bounding of the two-dimensional point cloud through convex hull or concave hull algorithms, and then generate the two-dimensional forming path through calculation methods such as fixed-angle straight line scanning, contour offset, parting curve and mesh tiling.

[0023] As a further design of this scheme, the method for outputting the multi-degree-of-freedom inkjet forming spatial path of the curved surface is as follows: read the obtained two-dimensional forming path and the trained point cloud data processing model, and import the two-dimensional forming path points into the point cloud data processing model in batches to generate the curved surface inkjet forming spatial path.

[0024] The beneficial effects of this invention are:

[0025] (1) In the data processing method of multi-degree-of-freedom inkjet forming three-dimensional model of curved surface designed in this invention, the target curved surface point cloud is segmented and rotated, then imported into a deep learning-based network fitting model and then rotated inversely to output. This can realize the direct mapping from two-dimensional data to three-dimensional data and effectively avoid the repeated sorting and calculation of point data when locating spatial points.

[0026] (2) In the data processing method for multi-degree-of-freedom inkjet forming three-dimensional model of curved surface designed in this invention, a complex path planning method for curved surface forming motion that is difficult to achieve by traditional calculation methods can be realized by mapping two-dimensional to three-dimensional. This method includes curved surface contour offset, curved surface fractal and curved surface mesh.

[0027] (3) In the three-dimensional model data processing method for multi-degree-of-freedom inkjet forming of curved surfaces designed in this invention, the spatial point positioning calculation efficiency is robust to the increase of the number of model vertices and the number of solution points, and is suitable for large-scale point cloud computing, which can provide technical support for efficient multi-degree-of-freedom inkjet forming of curved surfaces. Attached Figure Description

[0028] Figure 1 is a schematic diagram of point cloud extraction of the target surface of the STL model according to the embodiment of the present invention. The left image is the initial three-dimensional image of the point cloud, the middle image is the three-dimensional image of the surface point cloud annotation calculation, and the right image is the three-dimensional image of the surface point cloud after removing the internal point cloud.

[0029] Figure 2 is a diagram of the network model architecture for point cloud data processing used for spatial surface fitting according to an embodiment of the present invention.

[0030] Figure 3 is a schematic diagram of the point cloud segmentation module described in the embodiment of the present invention.

[0031] Figure 4 is a schematic diagram of the block point cloud rotation module according to the embodiment of the present invention. The left figure shows the maximum projected area and minimum envelope of the block point cloud rotation, and the right figure shows the curve of the change of the projected area of ​​the block point cloud rotation.

[0032] Figure 5 is a schematic diagram of the deep neural network fitting module described in the embodiment of the present invention, wherein the left figure is the three-dimensional image output by the helmet model fitting, and the right figure is the error change curve.

[0033] Figure 6 is a schematic diagram of the three-dimensional path generation described in the embodiment of the present invention.

[0034] Figure 7 is a flowchart of the method of the present invention. Detailed Implementation

[0035] The present invention will be further illustrated below with reference to the accompanying drawings and specific embodiments. It should be understood that the following specific embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without inventive effort are within the protection scope of the present invention.

[0036] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0037] As shown in Figure 7, this embodiment of the invention discloses a method for processing three-dimensional model data of multi-degree-of-freedom inkjet printing of curved surfaces, including the following steps:

[0038] (1) Read the STL model, a common file in additive manufacturing, and obtain the target spatial surface point cloud data in the model through the convex hull calculation method. Use the spatial point cloud data as the basic dataset for spatial surface fitting.

[0039] (2) Construct a point cloud data processing network model suitable for spatial surface fitting;

[0040] (3) In the established point cloud data processing network model, the planar coordinates and spatial coordinates of the model points in the training base dataset are used to obtain the weights after training.

[0041] (4) Project the target spatial surface onto a two-dimensional plane and generate a two-dimensional path based on the two-dimensional point cloud;

[0042] (5) Map the obtained two-dimensional path to the network fitting model and use the weights obtained from training to output the multi-degree-of-freedom inkjet forming spatial path of the surface.

[0043] As a preferred embodiment of the present invention, referring to FIG1, the process of obtaining the dataset in step (1) is as follows:

[0044] (1-1) Read the ASCII-formatted STL model through the mesh interface;

[0045] (1-2) Calculate the surface point cloud using the convex hull algorithm in the PIL interface;

[0046] (1-3) Use (x,y) in the point cloud data as input and z as label to form the training dataset.

[0047] As a preferred embodiment of the present invention, referring to FIG2, in step (2), the point cloud data processing network model for fitting the spatial surface includes: a point cloud segmentation module for segmenting the target surface point cloud into several block point clouds that are easy to fit; a block point cloud rotation module for rotating the block point cloud obtained in the point cloud segmentation module to obtain a rigid rotation matrix with the maximum projected area; and a deep neural network fitting module for fitting the rigidly rotated block point cloud obtained in the block point cloud rotation module.

[0048] As a preferred embodiment of the present invention, referring to FIG3, the process of the point cloud segmentation module includes: projecting the target curved surface point cloud onto a two-dimensional plane, calculating the minimum square bounding of the point cloud using the minimum square bounding algorithm, and then dividing the point cloud into block-shaped point clouds of size 50×50 using equidistant interpolation with a spacing of 50 in the square bounding.

[0049] As a preferred embodiment of the present invention, referring to Figure 4, the process of the block point cloud rotation module includes: reading in several block point clouds obtained in S2, taking the two points with the largest Euclidean distance in each point cloud as the rotation axis and also the X vector of the rigid rotation matrix, taking the midpoint O of the two points as the origin of the coordinate system, generating 720 Y vectors passing through O and perpendicular to the X axis in a plane, with the angle difference between the plane intersection lines being 0.5°, calculating the Z vector of the rotation matrix by performing an outer product of the X vector and each Y vector, calculating the projected area under the point cloud rotation and iterating the rigid rotation matrix under the maximum projected area, and outputting the rotated point cloud data and the rigid inverse rotation matrix. In Figure 4, the projected area is the largest when the point cloud rotation angle is 157°.

[0050] As a preferred embodiment of the present invention, referring to FIG5, the process of the deep neural network fitting module includes: reading the rotated point cloud data output from the block point cloud rotation module, capturing the spatial topological relationship of the point cloud through a multilayer perceptron, wherein the batch size of the multilayer perceptron is 256, the epoch is 10, the optimizer is 'adam', the loss function is 'mse', and the activation function is 'sigmoid', thereby forming a mapping relationship between two-dimensional plane points and three-dimensional spatial positions in the curved surface, and the spatial position can be directly output by inputting plane path points.

[0051] As a preferred embodiment of the present invention, referring to FIG2, in step (3), the point cloud data processing network model training process is as follows:

[0052] (3-1) Read the target surface point cloud data and obtain several rotated block point clouds and their rigid inverse rotation matrices after processing by the point cloud segmentation module and the block point cloud rotation module.

[0053] (3-2) Normalize the input point cloud data, calculate the spatial distance of the current input (x,y) obtained by the network and multiply it by the rigid inverse rotation matrix to obtain the calculated spatial distance, compare the calculated spatial distance with the actual spatial distance in the label, and make the calculated spatial distance equal to the actual spatial distance.

[0054] (3-3) Repeat (3-2) to optimize and adjust the parameters of each network node in the model, and finally obtain an optimal solution. The optimal solution makes the computational space distance equal to the actual space distance. This optimal solution is the weight of the point cloud data processing network model, and it is stored together with the network model as an h5 file.

[0055] As a preferred embodiment of the present invention, referring to FIG6, in step (4), the point cloud data processing network model training process is as follows:

[0056] (4-1) Read the target space point cloud and project it onto a two-dimensional plane, and calculate the minimum bounding space of the two-dimensional projected point cloud of the radome, the nose of the aircraft and the helmet model using the convex hull algorithm.

[0057] (4-2) Then, a two-dimensional forming path is generated by using a fixed-angle straight-line scan on the radome and the front nose of the aircraft, and by using a contour offset calculation method on the helmet.

[0058] As a preferred embodiment of the present invention, referring to FIG6, in step (5), the point cloud data processing network model training process is as follows:

[0059] (5-1) Read the two-dimensional forming paths of the antenna radome, the nose of the aircraft and the helmet model, the trained point cloud data processing model and the corresponding weights.

[0060] (5-2) Import the two-dimensional forming path points into the point cloud data processing model in batches to generate the curved surface inkjet forming spatial path.

[0061] Finally, the antenna radome, aircraft nose, and helmet models solved for 10,000 path points. In 20 experiments, the average single-point solution speed was 0.0244ms / point, 0.0243ms / point, and 0.0246ms / point, respectively. The single-point solution speed did not increase with the number of vertices and decreased with the number of solution points.

[0062] The technical means disclosed in this invention are not limited to those disclosed in the above embodiments, but also include technical solutions composed of any combination of the above technical features.

Claims

1. A curved surface multi-degree of freedom inkjet shaping three-dimensional model data processing method, characterized in that, Includes the following steps: S1: Read the STL model file, a common file in additive manufacturing, and obtain the target spatial surface point cloud data in the model through convex hull or concave hull calculation methods. Use this spatial point cloud data as the basic dataset for spatial surface fitting. S2: Construct a point cloud data processing network model suitable for spatial surface fitting; S3: In the point cloud data processing network model established in S2, train the planar coordinates and spatial coordinates of the model points in the basic dataset obtained in S1 to obtain the weights after training. S4: Project the target space surface obtained in S1 onto a two-dimensional plane and generate a two-dimensional path based on the two-dimensional point cloud; S5: Map the two-dimensional path obtained in S4 to the point cloud data processing network model constructed in S2, and use the weights obtained in S3 to output the multi-degree-of-freedom inkjet forming spatial path of the surface.

2. The method of claim 1, wherein the method is a method of processing three-dimensional model data for inkjet forming a curved surface having multiple degrees of freedom, characterized by In S1, the spatial coordinates of unordered points on the surface of the STL model in ASCII code format are read. Then, the spatial point cloud data of the target surface in the model is quickly extracted by convex hull or concave hull calculation in the 3D point cloud processing method. The point cloud data (x,y) is used as input and z is used as label to form the training dataset.

3. The method of claim 1, wherein the method is a method of processing three-dimensional model data for inkjet forming a curved surface having multiple degrees of freedom, characterized by, In S2, the point cloud network processing model for fitting spatial curved surfaces includes: a point cloud segmentation module, a block point cloud rotation module, and a deep neural network fitting module. The point cloud segmentation module is used to segment the target curved surface point cloud into several block point clouds that are easy to fit. The block point cloud rotation module is used to rotate the block point clouds obtained in the point cloud segmentation module to obtain a rigid rotation matrix with the maximum projected area. The deep neural network fitting module is used to fit the rigidly rotated block point cloud obtained in the block point cloud rotation module.

4. The method of claim 3, wherein the method further comprises: The point cloud segmentation module divides the obtained two-dimensional coordinates into multiple block point clouds according to the projected grid. The process of generating block point clouds includes: The point cloud obtained in S1 is projected onto a two-dimensional plane, and the minimum square bounding algorithm is used to calculate the minimum square bounding of the point cloud. Then, the point cloud is divided into multiple block-shaped point clouds of equal size by equal-distance interpolation within the square bounding.

5. The method of claim 3, wherein the method further comprises: The process of obtaining the rigid rotation matrix for the maximum two-dimensional projection, the rotated point cloud data, and the rigid inverse rotation matrix from the block point cloud obtained in the block point cloud rotation module and the point cloud segmentation module includes: Read in several block point clouds obtained from S2, take the two points with the largest Euclidean distance in each point cloud as the rotation axis and also the X vector of the rigid rotation matrix, take the midpoint O of the two points as the origin, generate 720 Y vectors in the plane passing through O and perpendicular to the X axis, with the angle difference between the intersecting lines of the planes being 0.5°, and calculate the Z vector of the rotation matrix by performing the outer product of the X vector and each Y vector. Calculate the projected area of ​​the point cloud under rotation and iterate the rigid rotation matrix under the maximum projected area, and output the point cloud data after rotation and the rigid inverse rotation matrix.

6. The method of claim 3, wherein the method further comprises: The process of using a deep neural network fitting module to fit the rotated point cloud output by the block point cloud rotation module and adding a rigid inverse rotation module includes: The system reads the rotated point cloud data output from the block point cloud rotation module, captures the spatial topological relationship of the point cloud through a deep neural network, and forms a mapping relationship between two-dimensional plane points and three-dimensional spatial positions in the curved surface. The spatial position can be directly output by inputting plane path points.

7. The method for processing three-dimensional model data of multi-degree-of-freedom inkjet printing of curved surfaces according to claim 1, characterized in that, In S3, the process of training the point cloud data processing network model includes the following steps: S3-1, Several rotated block point clouds and their rigid inverse rotation matrices are obtained after the target surface point cloud data is processed by the point cloud segmentation module and block point cloud rotation module in S2. S3-2, normalize the input point cloud data, calculate the spatial distance of the current input (x,y) obtained by the network and multiply it by the rigid inverse rotation matrix to obtain the calculated spatial distance, compare the calculated spatial distance with the actual spatial distance in the label, and make the calculated spatial distance equal to the actual spatial distance. S3-3, repeat S3-2 to optimize and adjust the parameters of each network node in the model, and finally obtain an optimal solution. The optimal solution makes the computational spatial distance equal to the actual spatial distance. This optimal solution is the weight of the point cloud data processing network model.

8. The method of claim 1, wherein the method is a method of processing three-dimensional model data for inkjet forming a curved surface with multiple degrees of freedom, characterized in that, In S4, the target spatial point cloud obtained in S1 is read and projected onto a two-dimensional plane. The minimum bounding space of the two-dimensional point cloud is calculated using convex hull or concave hull algorithms. Then, a two-dimensional forming path is generated using calculation methods such as fixed-angle straight line scanning, contour offset, parting curve, and mesh tiling.

9. The method of claim 1, wherein the method is a method of processing three-dimensional model data for inkjet forming a curved surface with multiple degrees of freedom, characterized in that, In S5, the two-dimensional forming path obtained in S4 and the point cloud data processing model trained in S3 are read, and the two-dimensional forming path points are imported into the point cloud data processing model in batches to generate the curved surface inkjet forming spatial path.

Images