A method for generating a simulation model of a particle accumulation body with natural three-dimensional shape characteristics of building materials

Abstract

The present invention discloses a method for generating a simulation model of a particle accumulation body with natural three-dimensional shape characteristics of building materials, comprising the following steps: step 1: making a natural particle simulation model; assigning natural sphericity and roundness to particles of randomly generated three-dimensional coordinate control points to obtain a natural particle simulation model; step 2: making a particle accumulation body simulation model; filling the natural particle simulation model obtained in step 1 with discrete element clusters, and applying gravity deposition to the discrete element clusters through a discrete element method to obtain a particle accumulation body simulation model. The present invention realizes the rapid simulation of the particle accumulation body with natural three-dimensional shape characteristics of building materials without pre-scanning information, provides a solid foundation for exploring the impact of the microscopic geometric characteristics of contact particle accumulation bodies on the macroscopic mechanical properties of civil engineering materials, and provides a new way to efficiently generate a particle accumulation body database with precise contact information.

Description

Technical Field

[0001] The present invention relates to the technical field of building models, and in particular to a method for generating a simulation model of a particle accumulation body with natural three-dimensional shape characteristics of building materials. Background Art

[0002] Granular materials are a key component in civil engineering construction. Randomly stacked particles of varying shapes, characteristics, and gradations form widely used granular aggregates in engineering practice, such as centimeter-scale concrete blocks and meter-scale dam rocks. Over the past few decades, numerous experimental and numerical studies have revealed that the microscopic geometrical characteristics of granular aggregates (including particle shape, size, spatial distribution, and packing state) significantly influence their macroscopic mechanical properties. For example, the flatness and slenderness of sand particles are key factors determining the mechanical properties of interparticle contact and, in turn, the shear strength of the aggregate. Meanwhile, particle roughness and roundness negatively impact the small-strain properties of cement-based and asphalt mixtures. In natural environments, construction granules of various shapes are generated through artificial blasting or environmental weathering. Rocks are broken down into fragments of varying sizes and contours, their surfaces and edges sculpted to varying degrees, exhibiting undulating and sharp features. Over the past few decades, with increasing interest in the three-dimensional shape of granular materials, researchers have provided numerous descriptions aimed at characterizing 3D particle data from laser scanners (surface point clouds of particles) and computed tomography (voxel views of particles). At the particle level, the overall shape of the particle is expressed by sphericity, and the similarity of the particle to the sphere can be classified as isosphere, flat sphere, disc, elongated sphere, blade, and rod ( Figure 1 ), these classifications are based on extensibility and flatness. At the angular level, the sharpness of corners can be expressed by roundness, quantified by the radius of curvature of the corners; while at the microscopic surface level, relief is represented by roughness, defined as the ratio of the particle volume to its convex hull.

[0003] In order to further explore the influence of microscopic geometric features on the macroscopic mechanical response of granular deposits, it is necessary to apply high-tech technologies such as artificial intelligence, discrete element method or material point method to granular deposit simulation experiments. However, the above applications require the generation of granular deposits that conform to large-scale natural shapes. At present, Zheng Junxing and others have successfully achieved the generation of two-dimensional particles using the diffusion model, but two-dimensional particles cannot meet the spatial distribution required for three-dimensional simulation. Using CT technology, Wang Jianfeng and others scanned the granular deposits and segmented the obtained tomographic images using the contact watershed algorithm to obtain voxel data of three-dimensional contact granular deposits. However, this method is time-consuming and labor-intensive, and cannot generate granular deposits of the scale required by artificial intelligence or discrete element methods in a short period of time.

[0004] Therefore, a method for generating a simulation model of a particle accumulation body with natural three-dimensional shape characteristics of building materials is provided to solve the above technical problems. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for generating a simulation model of a particle accumulation body with natural three-dimensional shape characteristics of building materials.

[0006] The object of the present invention is achieved like this:

[0007] A method for generating a simulation model of a particle accumulation body with natural three-dimensional shape characteristics of a building material comprises the following steps:

[0008] Step 1: Create a natural particle simulation model; assign the particles natural sphericity and roundness to the randomly generated three-dimensional coordinate control points to obtain a natural particle simulation model;

[0009] Step 2: Make a particle accumulation simulation model; fill the natural particle simulation model obtained in step 1 with discrete element clusters, and apply gravity deposition to the discrete element clusters through the discrete element method to obtain a particle accumulation simulation model.

[0010] In step 1, making a natural particle simulation model includes the following three steps, which define the basic shape / size, surface relief, and roundness of the particles in a progressive relationship:

[0011] Step 1.1 Generate different particle contours to control the particle size and main shape, such as aspect ratio and flatness, so that the model approximates the shape of natural particles;

[0012] Step 1.2 generates different surface undulations to simulate natural phenomena such as roughness and weathering on the particle surface, thus enhancing the authenticity of the particles.

[0013] Step 1.3 generates different particle roundness to simulate the sharpness of particles in nature and avoid overly sharp corners.

[0014] The specific operations of step 1.1 are as follows:

[0015] Step 1.1.1 Generate a point cloud P ini , point cloud P ini The coordinates follow the 3D multivariate standard normal distribution; Step 1.1.2 stretches the x and y coordinates and aligns the z axis so that the point cloud P ini Show the sphericity characteristics of natural building particles; Step 1.1.3 for point cloud P ini Stretching includes: Point cloud P ini The projections on the x, y, and z axes are the shortest axis S, longest axis L, and second longest axis I of the particle respectively; Step 1.1.4 respectively converts the point cloud P iniThe x-coordinate of the point cloud P is multiplied by the target slenderness ratio. ini The y coordinate of the point cloud P is multiplied by the inverse of the target flatness. ini With target sphericity values, different particle profiles are obtained.

[0016] The specific operations of step 1.2 are as follows: Step 1.2.1 generates the outer mesh of the closed point cloud by using the α shape algorithm; Step 1.2.2 ensures that the point cloud P ini The external points of are not on the same convex surface to generate natural surface relief. In step 1.2.3, the α value is adjusted to control the density of the generated mesh and the surface relief characteristics.

[0017] The specific operations of step 1.3 are as follows: Step 1.3.1 simulates a more realistic and natural geometry and surface features by identifying and processing sharp corners of the particles; Step 1.3.2 uses the K-nearest neighbor method to identify sharp corners with high curvature and smooths them using the moving least squares method (MLS) to generate particles with varying degrees of roundness. Step 1.1 first defines the basic shape and size of the particles, Step 1.2 adds natural surface contours to this foundation, and Step 1.3 further adjusts the roundness of the particles to refine their detailed features.

[0018] In step 2, making a particle accumulation simulation model includes the following steps:

[0019] Step 2.1 generates a particle accumulation body with contact information and marks the contact area; Step 2.2 generates a particle cluster with contact information and uses the discrete element method DEM to simulate the particle accumulation process; Step 2.3 generates an initial block stone sphere discrete element model and uses random three-dimensional coordinate control points in the unit space to generate corresponding convex hull polyhedron triangles, and establishes the block stone sphere model through the rigid ball algorithm; Step 2.4 uses the gravity deposition method in the discrete element software to accumulate a random number of block stone sphere models.

[0020] In step 2.4, during the accumulation process, the coordinates of the center points of each ball before and after accumulation are recorded to calculate the transformation matrix, and the point cloud P of the corresponding stone is moved according to the obtained transformation matrix. ini Coordinates, thus obtaining the particle point cloud P ini data.

[0021] In step 2.1, the contact information of the particle accumulation body is annotated by the normal vector difference method to mark the contact points; specifically, the K-nearest neighbor method is used to select the neighboring points near the target point, and the PCA algorithm is used to fit the neighboring points into a plane, and the normal vector of the plane is used as the normal vector of the target point; when the point near the target point changes, its corresponding normal vector also changes accordingly; by calculating the normal vector of the newly generated point cloud, and selecting the point where the normal vector changes in each step as the block stone boundary edge point, the generation of the particle contact particle accumulation body is completed.

[0022] The beneficial effects of the present invention are as follows: the natural particle simulation module production method of the present invention gives the particles natural sphericity and roundness by stretching, smoothing and other methods the randomly generated three-dimensional coordinate control points, and can quickly generate a large number of building material particle models. The particle dynamic accumulation module production method first fills the particle model generated by the natural particle simulation module with a high-reduction discrete element cluster, and then applies gravity deposition to the discrete element cluster particles through the discrete element method to realize the simulation of the particle accumulation body. According to the characteristic of the discrete element method for real-time tracking of the center of the discrete element cluster particle, the particle dynamic accumulation module can track the movement trajectory of the stone blocks during the gravity deposition process in real time. The normal vector difference method is used after the particle dynamic accumulation module, and the contact points of the generated particle accumulation body simulation model can also be accurately marked. The present invention realizes the rapid simulation of the particle accumulation body with the natural three-dimensional shape characteristics of the building material without pre-scanning information, provides a solid foundation for exploring the effect of the microscopic geometric characteristics of the contact particle accumulation body on the macroscopic mechanical properties of civil engineering materials, and provides a new way to efficiently generate a particle accumulation body database with precise contact information. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 Schematic diagram of the results of generating different overall shapes of particles for the present invention;

[0024] Figure 2 This is a schematic diagram of the fluctuation changes of the particle surface generated by different α values ​​of the present invention;

[0025] Figure 3 This is a schematic diagram of the generation of a contact particle accumulation body under the action of gravity in the present invention;

[0026] Figure 4 This is a schematic diagram of the contact process of calibrating a particle accumulation body using the normal vector difference method of the present invention;

[0027] Figure 5 Schematic diagram of particle surface edge recognition based on K proximity in the present invention;

[0028] Figure 6 A shape parameter characteristic distribution diagram of a virtual building particle database is generated for the present invention. DETAILED DESCRIPTION

[0029] The present invention will be further described below with reference to the accompanying drawings and examples.

[0030] A method for generating a simulation model of a particle accumulation body with natural three-dimensional shape characteristics of a building material comprises the following steps:

[0031] Step 1: Create a natural particle simulation model; assign the particles' natural sphericity and roundness to the randomly generated three-dimensional coordinate control points to obtain a natural particle simulation model.

[0032] In step 1, making a natural particle simulation model includes the following three steps, which define the basic shape / size, surface relief, and roundness of the particles in a progressive relationship:

[0033] Step 1.1 generates different particle contours. The technical purpose is to control the size and main shape of the particles, such as aspect ratio and flatness, so that the model approximates the shape of natural particles, such as Figure 1 As shown; the specific operations of step 1.1 are as follows:

[0034] Step 1.1.1 Generate a point cloud P ini , point cloud P ini The coordinates follow the 3D multivariate standard normal distribution; Step 1.1.2 stretches the x and y coordinates and aligns the z axis so that the point cloud P ini Show the sphericity characteristics of natural building particles; Step 1.1.3 for point cloud P ini Stretching includes: Point cloud P ini The projections on the x, y, and z axes are the shortest axis S, longest axis L, and second longest axis I of the particle respectively; Step 1.1.4 respectively converts the point cloud P ini The x-coordinate of the point cloud P is multiplied by the target slenderness ratio. ini The y coordinate of the point cloud P is multiplied by the inverse of the target flatness. ini With target sphericity values, different particle profiles are obtained.

[0035] Step 1.2 generates different surface undulations to simulate natural phenomena such as roughness and weathering on the particle surface and enhance the authenticity of the particles; Figure 2 As shown in the figure, different α values ​​will directly control the fluctuation of the particle surface. The specific operations of step 1.2 are as follows: step 1.2.1 generates the outer mesh of the closed point cloud through the α shape algorithm; step 1.2.2 ensures that the point cloud P ini The external points of are not on the same convex surface to generate natural surface relief. In step 1.2.3, the α value is adjusted to control the density of the generated mesh and the surface relief characteristics.

[0036] Step 1.3 generates particles of varying roundness to simulate the sharpness of particles in nature and avoid overly sharp corners. Specifically, step 1.3.1 identifies and processes sharp corners of particles to simulate more realistic, natural geometry and surface features. Step 1.3.2 uses the K-nearest neighbor method to identify sharp corners with high curvature and smooths them using the moving least squares method (MLS) to generate particles of varying roundness.

[0037] Step 1.1 first defines the basic shape and size of the particles, step 1.2 adds natural surface undulations, and step 1.3 further adjusts the roundness of the particles and refines their detailed features.

[0038] Step 2: Make a particle accumulation simulation model; fill the natural particle simulation model obtained in step 1 with discrete element clusters, and apply gravity deposition to the discrete element clusters through the discrete element method to obtain a particle accumulation simulation model.

[0039] The production of particle accumulation simulation model includes the following steps: Figure 3 As shown in , step 2.1 generates a particle accumulation body with contact information and marks the contact area; Figure 4 As shown in the figure, the contact information of the particle accumulation body is annotated by the normal vector difference method; specifically, the K-nearest method is used to select the neighboring points near the target point, and the PCA algorithm is used to fit the neighboring points into a plane, and the plane normal vector is used as the target point normal vector; when the points near the target point change, the corresponding normal vector also changes accordingly; by calculating the normal vector of the newly generated point cloud, and selecting the point where the normal vector changes in each step as the block stone boundary edge point, the generation of the particle contact particle accumulation body is completed.

[0040] Step 2.2 generates particle clusters with contact information and uses the discrete element method (DEM) to simulate the particle accumulation process.

[0041] Step 2.3 generates the initial stone sphere discrete element model by using random three-dimensional coordinate control points in the unit space to generate corresponding convex hull polyhedron triangles, and establishes the stone sphere model through the rigid sphere algorithm.

[0042] Step 2.4: A random number of stone sphere models are piled up using the gravity deposition method in the discrete element software. During the accumulation process, the coordinates of the center points of each ball before and after the accumulation are recorded to calculate the transformation matrix, and the point cloud P of the corresponding stone is moved according to the obtained transformation matrix. ini Coordinates, thus obtaining the particle point cloud P ini data.

[0043] Next, we will first generate the closed P ini The grid represents the particles The surface points can be sampled from the grid as if by laser scanning, and the particle The voxel data of can be formed by enclosing a binary 3D matrix with a grid. Moderate undulations on the generated grid are necessary to generate particles with natural surface textures. In our method, the random strategy makes P ini The points in P cannot be on the same convex surface, thus ensuring that ini There will inevitably be fluctuations at the external points of P. ini The number of points in is set to 1000, and the α shape algorithm (which is an effective method for generating triangular boundary meshes of controllable complexity point clouds) is used to generate closed P ini Specifically, the α shape algorithm attempts to ini Find the outer grid such that a sphere of radius α cannot penetrate the grid. As α increases from 0 to +∞, the α shape will provide a finite spectrum containing P ini itself (α=0), The convex hull of , and the gradual changes between these two extremes. In this spectrum, there is always a critical value , represents the generation of closed all P ini However, when When the generated mesh is The tightest closure of the particle size is more extreme than the surface texture of natural particles. According to our experimental experience, α is set to 1.5 to 2 times The random value of P is sufficient for a particle size of 1 ini Generate a mesh that preserves the undulations and approximates the surface of natural particles. Afterwards, virtual particles with various overall morphologies and surface undulations can be generated by uniformly sampling 10,000 points on the mesh and using this mesh to close a binary 3D matrix with a resolution of 0.3. and Note that voxel data can be converted to point clouds based on voxel resolution and coordinates. and are all represented as 3D point sets:

[0044]

[0045] in, In N is equal to 10000, and in , the number of elements of a voxel is 1.

[0046] Generate particles with varying degrees of roundness: and Only very sharp corners generated during the sampling and closing process are included. These corners are generated by the process of connecting triangles. In order to generate particles with corners of different sharpness, we first use the K-nearest neighbor method to identify the existing sharp corners. These corners are regarded as clusters of high curvature points, such as Figure 5 These corners are then smoothed at different levels using the widely used Moving Least Squares (MLS) method. Specifically, for each point in the corner cluster , MLS will update to a position on a smooth surface , the surface is obtained by the neighborhood Fitted, which includes the distance Neighboring vertices within By solving the optimization problem:

[0047]

[0048] minimize Means to reduce The z coordinate value of and projection values The weighted sum of the differences between . is a Gaussian weight function, indicating that the distance Closer points will have higher weights, thus Provide more local topological features. Therefore, for each , changing the value of r will result in different The discovery of points can further fit different surfaces with different smoothing effects .

[0049] Generation of contact aggregates and contact marking: Currently, the dynamic processes of various granular material aggregates, such as landslides and compression tests, can be effectively simulated by discrete element methods (DEM). This invention uses a highly reduced multi-sphere discrete element cluster model to transmit the forces between particles, thereby achieving high-precision simulation that takes into account the particle shape characteristics. Therefore, using the method proposed by (Angelidakis) and The grid is converted into discrete element clusters, and the cluster aggregation simulated by DEM is obtained under gravity deposition.

[0050] Since the relative positions of the DEM clusters to the grid and to the grid of point clouds and voxels remain unchanged, the DEM clusters can be replaced back into the original point clouds and voxels to generate clusters that simulate laser scans and CT scans, such as Figure 3As shown. In addition, during this replacement process, the neighboring points of some replaced points will change with the replacement of adjacent particles. By observation, we found that these points correspond to the particle contact areas that need to be marked, such as Figure 4 Therefore, in the present invention, the contact points are marked as the points whose normal vectors calculated by the neighboring points change during the replacement process.

[0051] Application Cases:

[0052] The stone's outline is controlled using randomly generated three-dimensional coordinate points. When generating different particle outlines, the distribution and number of control points must be determined to ensure that the particle shape closely resembles natural stone. Furthermore, a preliminary convex hull polyhedron model is generated by stretching the control points. Finally, a triangle patch algorithm is used to triangulate the convex hull to form the final discrete element cluster model of the stone.

[0053] In this embodiment, 100-500 three-dimensional coordinate control points are first randomly generated in the unit space. After random stretching, the control points are used to generate corresponding convex hull polyhedron triangles. Then, the rigid sphere algorithm is used to establish a stone sphere model. At the same time, the PCL computing library is used to convert the convex hull polyhedron triangles into point cloud format to obtain point cloud data of a single stone block.

[0054] In the initial stage of generating particles with various overall shapes and surface reliefs, a point cloud P is generated. ini , whose coordinates follow a three-dimensional multivariate standard normal distribution. Therefore, P ini The external points of P are randomly distributed along the x, y, and z axes, similar to a sphere. ini The x- and y-coordinates of the target stretch ratio (S / I) and the inverse of the target flatness (I / L) are multiplied respectively, and the x, y, and z axes can be aligned with the shortest axis (S), the longest axis (L), and the middle axis (I) so that P ini Shows the overall morphology of naturally constructed particles, e.g. Figure 1 As shown in . Randomly assigning the slenderness ratio and flatness in the range of 0.3 to 1.0 can generate almost all forms in nature, such as Figure 6 shown.

[0055] To mimic the generation of natural particles, we combined some mature 3D data processing techniques to carve the initial spherical point cloud into particles with surface point cloud and voxel data. The open source code AlphaShapeToolbox (https: / / github.com / bellockk / alphashape.git) and the point cloud library (http: / / pointclouds.org) were used in the process. We successfully constructed a virtual architectural particle database containing 25,000 virtual particles. These particles have many important morphological characteristics, including overall shape, roundness, and roughness. The shape parameter distribution of the particles is as follows: Figure 6 As shown, the morphological types of most particles found in nature are covered, and the distribution of shape parameters is broad and diverse, fully reproducing the geometric characteristics of particles in the natural environment. These goals are achieved by stretching the overall shape, extracting and modifying external surface points, and polishing corners. In addition, all particle sizes are normalized to 1 and dimensionless, making it easy for users to obtain particles of the desired shape and size.

[0056] In the process of particle edge recognition and smoothing, by randomly setting The particles with a diameter of 1 are successfully smoothed to a circularity in the range of 0.5 to 0.85 within the range of 0.05 to 0.3. All MLS operations are implemented using the KdTree and MovingLeastSquares functions in PCL.

[0057] In this example, 10 discrete element cluster models of rocks were randomly selected from a database of 25,000 virtual particles. The open-source discrete element software YADE was used to simulate the gravity deposition of the rock models. During the simulation, the rocks accumulated under the action of gravity to form a pile. Furthermore, the coordinates of the center points of each ball before and after accumulation were recorded to calculate a transformation matrix. The point cloud coordinates of the corresponding rocks were moved according to the obtained transformation matrix to generate accurate point cloud data of the particle aggregate, as shown in Figure 2. Figure 3 shown.

[0058] In the generated point cloud data, the normal vector difference method is applied to mark the contact areas between the blocks. The contact points are identified and marked by calculating the normal vector difference of each point. The marked areas are refined by combining the K-nearest neighbor method and the principal component analysis (PCA) algorithm to ensure the accuracy of the contact areas. The specific method is as follows: Relative contact information Expressed as:

[0059]

[0060] in, yes The number of points in . A value of 1 indicates a contact point, and a value of 0 indicates a non-contact point. are local features such as normal vectors and curvatures. Therefore, contact points are marked as points where the normal vectors calculated by neighboring points change during the replacement process, such as Figure 4 shown.

[0061] In this example, a contact-aggregation particle database of 5,000 particle aggregates was generated using discrete element simulation. To conserve computational memory and time, each aggregate consisted of 10 particles of varying shapes randomly selected from the contact-aggregation particle database and downsampled to 5,000 points. The size of each particle in the aggregate was a random value between 20 and 60, achieved by directly multiplying the coordinates of a particle with a diameter of 1 by the target diameter.

Claims

1. A method for generating a simulation model of a particle accumulation body with natural three-dimensional shape characteristics of building materials, characterized by: The following steps are involved: Step 1: Create a natural particle simulation model; assign the particles natural sphericity and roundness to the randomly generated three-dimensional coordinate control points to obtain a natural particle simulation model; The following steps are involved in making a natural particle simulation model: Step 1.1 generates different particle contours to control the size and main shape of the particles so that the natural particle simulation model approximates the shape of natural particles. The specific operations of step 1.1 are as follows: Step 1.1.1 Generate a point cloud P ini , point cloud P ini The coordinates of follow a 3D multivariate standard normal distribution; Step 1.1.2 stretches the x and y coordinates and aligns the z axis so that the point cloud P ini Shows the sphericity characteristics of natural building particles; Step 1.1.3 Point cloud P ini Stretching includes: Point cloud P ini The projections on the x, y, and z axes are the shortest axis S, longest axis L, and second longest axis I of the particle, respectively; Step 1.1.4: Point cloud P ini The x-coordinate of the point cloud P is multiplied by the target slenderness ratio. ini The y coordinate of the point cloud P is multiplied by the inverse of the target flatness. ini With target sphericity value, different particle profiles can be obtained; Step 1.2 generates different surface reliefs to enhance the realism of the particles; Step 1.3 generates different particle roundness to simulate the sharpness of particles in nature and avoid overly sharp corners; Step 2: Create a particle accumulation simulation model; fill the natural particle simulation model obtained in step 1 with discrete element clusters, and apply gravity deposition to the discrete element clusters using the discrete element method to obtain a particle accumulation simulation model; The following steps are involved in making a particle accumulation simulation model: Step 2.1 Generate a particle accumulation body with contact information and mark the contact area; the particle accumulation body contact information is marked by the normal vector difference method to mark the contact points; Step 2.2 generates particle clusters with contact information and uses the discrete element method (DEM) to simulate the particle accumulation process; Step 2.3 generates the initial block stone sphere discrete element model. The corresponding convex hull polyhedron triangles are generated using the random three-dimensional coordinate control points in the unit space, and the block stone discrete element cluster model is established using the rigid sphere algorithm. Step 2.4: The random number of stone discrete element cluster models are piled up using the gravity deposition method in the discrete element software. During the accumulation process, the coordinates of the center points of each ball before and after the accumulation are recorded to calculate the transformation matrix, and the point cloud P of the corresponding stone is moved according to the obtained transformation matrix. ini Coordinates, thus obtaining the particle point cloud P ini data.

2. The method for generating a simulation model of a particle accumulation body with natural three-dimensional shape characteristics of a building material according to claim 1, characterized in that: The specific operations of step 1.2 are as follows: Step 1.2.1 Generate a surface mesh of the closed point cloud using the alpha shape algorithm; Step 1.2.2 Ensure that the point cloud P is obtained by random strategy ini The outer points of are not on the same convex surface to generate natural surface relief; In step 1.2.3, the α value is adjusted to control the density of the generated mesh and the surface relief characteristics.

3. The method for generating a simulation model of a particle accumulation body with natural three-dimensional shape characteristics of a building material according to claim 1, characterized in that: The specific operations of step 1.3 are as follows: Step 1.3.1: Simulate real, natural geometry and surface features by identifying and processing sharp corners on the particle surface. In step 1.3.2, the K-nearest neighbor method is used to identify sharp corners with high curvature, and the sharp corners are smoothed by the moving least squares method (MLS) to generate particles with different roundness.

4. The method for generating a simulation model of a particle accumulation body with natural three-dimensional shape characteristics of a building material according to claim 1, characterized in that: The step 2.1 is specifically as follows: first, the K-nearest neighbor method is used to select neighboring points near the target point, and the PCA algorithm is used to fit the neighboring points into a plane, with the plane normal vector being used as the target point normal vector; when the points near the target point change, the corresponding normal vector also changes accordingly; by calculating the normal vector of the newly generated point cloud, and selecting the point where the normal vector changes in each step as the block stone boundary edge point, the generation of the particle contact particle accumulation body is completed.

Images