A complex geometric model particle discretization method, system and electronic device

Abstract

The application discloses a complex geometric model particle discrete method and system and electronic equipment, belongs to the field of fluid mechanics, proposes a kind of efficient, accurate method for constructing model geometry surface narrowband area LSDF, and carry out particle compensation, to avoid the problem that particle spacing increases after particle relaxation;According to the above method, by inputting the geometric model and the particle spacing after discretization, the complex geometric model can be discretized into uniform, adaptive particles with specified spacing. The method solves the problem of inaccurate and inefficient LSDF in the prior art, and the volume error of the model after being discretized into particles is large. By inputting the STL geometric model file and the particle spacing, the model can be automatically discretized into a particle model. The particle model can be applied to numerical calculation of the particle method. The accuracy and universality of the method and device in the application are verified by particle discretization of various complex geometries.

Description

Technical Field

[0001] This invention belongs to the field of fluid mechanics technology, specifically relating to a particle discretization method, system, and electronic device for complex geometric models. Background Technology

[0002] Numerical simulation is crucial in fluid machinery, enabling the optimization of equipment performance and safety. When fluid machinery involves multiple, unsteady flows, traditional mesh methods suffer from mesh distortion, requiring continuous mesh reconstruction and resulting in decreased computational accuracy and efficiency. Meshless particle methods, lacking fixed topological relationships, offer unique advantages in calculating large deformations and surface flows, among other unsteady problems. The preprocessing step of particle methods involves generating uniformly distributed particles; however, particle discretization of complex models remains a significant challenge hindering the application of particle methods in industrial fields such as fluid machinery and marine engineering.

[0003] Currently, there are three main particle discretization methods for complex geometric models in particle methods: The first method generates particles on a volumetric mesh. Commercial preprocessing software can easily divide the model into hexahedral or tetrahedral meshes, and then particles are generated by extracting points at the center of the mesh. However, particles generated in this way are not completely uniform. The second method generates initial particles directly on an orthogonal mesh located inside the model. However, this method has jagged edges at the boundaries and does not conform to the geometric surface, which affects computational accuracy at low resolutions. The third method generates initial particles on an orthogonal mesh and then performs particle relaxation based on a signed distance field (SDF).

[0004] The third method can generate particles with smooth boundaries and uniform body fit, but it has two major challenges: First, it requires an efficient and accurate SDF construction algorithm. In the preprocessing modeling of the particle method, only the local signed distance field (LSDF) information of the narrow band region near the geometric surface is needed. Second, the number of particles generated initially is insufficient. After particle relaxation, the particles become sparse, which affects the accuracy of numerical calculation. Summary of the Invention

[0005] In order to overcome the shortcomings of the prior art, the present invention aims to provide a particle discretization method, system and electronic device for complex geometric models, so as to solve the problems of insufficient accuracy and efficiency of LSDF and large volume error after the model is discretized into particles.

[0006] To achieve the above objectives, the present invention employs the following technical solution:

[0007] A particle discretization method for complex geometric models includes the following steps:

[0008] Import the geometric model to construct an LSDF;

[0009] Initial particles are generated based on LSDF, and particle compensation is performed on the initial particles;

[0010] Particle relaxation is performed on the initial particles and the compensated particles until the particles are uniform.

[0011] Preferably, the importing of the geometric model to construct the LSDF specifically includes the following steps:

[0012] S101: Information about each triangular mesh in the analytical geometric model;

[0013] S102: Determine the boundary grid through two grid filtering processes based on the information from the triangular grid;

[0014] S103: Determine the sign value of the boundary grid distance based on the positional relationship between the boundary grid and the adjacent triangular grid, and construct the LSDF of the narrow band region of the geometric surface.

[0015] Preferably, in S102, the first grid screening is as follows:

[0016] S1021: Determine a cubic region that completely contains the triangular mesh based on the information of the triangular mesh, denoted as region I;

[0017] S1022: Expand region I by 1 / 2 the narrow band width to obtain region II.

[0018] Preferably, in S102, the second grid screening is as follows:

[0019] Calculate the distance between the center point of the grid and the triangle in the region obtained after the first grid screening. If the distance between the center point of the grid and the triangle is not greater than 1 / 2 of the narrow band width, then the grid belongs to the boundary grid. If it is greater, then the grid does not belong to the boundary grid.

[0020] Preferably, the step of generating initial particles based on LSDF and performing particle compensation on the initial particles includes the following steps:

[0021] S201: Determine the symbol for the non-boundary mesh based on LSDF;

[0022] S202: For non-boundary meshes, extract the center of the internal mesh as the initial particle;

[0023] S203: For boundary meshes, if the signed distance value is less than -1 / 2 times the mesh width, extract the mesh center point as a particle;

[0024] S204: For meshes crossed by LSDF boundaries, calculate the volume error and perform particle compensation based on the volume error.

[0025] Preferably, in step S204, the method for calculating the volume error is as follows:

[0026]

[0027] The volume V of the particle model p for:

[0028]

[0029] If defined The volume error can then be determined by the particle number error σ. N To indicate:

[0030]

[0031] Where V is the actual volume of the model; V p l0 represents the volume of the particle model; N represents the actual number of particles; l0 represents the mesh width, which is also the particle spacing; N t This represents the number of particles in the particle model.

[0032] Preferably, particle migration technology is used to relax the initial particles and the compensated particles until the particles are uniform.

[0033] A complex geometric model particle discrete system, comprising:

[0034] The LSDF build module is used to import geometric models and build LSDF.

[0035] The particle generation module is used to generate initial particles based on LSDF and perform particle compensation on the initial particles.

[0036] The particle relaxation module is used to relax the initial particles and the compensated particles until the particles are uniform.

[0037] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the steps of the particle discretization method for complex geometric models described above.

[0038] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the particle discretization method for complex geometric models described above.

[0039] Compared with the prior art, the present invention has the following beneficial effects:

[0040] This invention proposes a particle discretization method and related apparatus for complex geometric models, which can discretize any complex model into particles with uniformly distributed spacing.

[0041] Furthermore, an efficient and accurate method for constructing the narrow-band LSDF of the model's geometric surface is proposed. The boundary mesh is determined through two screenings without the need to construct a global distance field. The sign and distance of the boundary mesh are obtained by accurate numerical calculation, and the non-boundary mesh is quickly determined using an traversal approach, which significantly reduces the amount of computation.

[0042] Furthermore, a particle compensation model is proposed. Based on the positional relationship between the mesh and the geometric boundary at the boundary, the missing volume at the boundary is statistically analyzed and particles are added to effectively reduce the volume error after the model is discretized into particles, thereby avoiding the problem of increased particle spacing after particle relaxation.

[0043] Furthermore, a particle migration algorithm is used to adjust the particle positions while constraining the particles to not flow out of the boundary, so that the particles are attached to the model boundary and uniformly distributed; CFL conditions are introduced in particle migration to improve the stability of the relaxation process.

[0044] This invention addresses the shortcomings of existing LSDF methods, such as insufficient accuracy and efficiency, and large volume errors after model discretization into particles. By inputting an STL geometric model file and particle spacing, the model can be automatically discretized into a body-fitting, uniform particle model, which can be applied to numerical calculations using particle methods. The accuracy and versatility of the method and apparatus in this invention have been verified through particle discretization of various complex geometries. Attached Figure Description

[0045] Figure 1 This is a flowchart of the method of the present invention;

[0046] Figure 2 In constructing the LSDF, a schematic diagram of the boundary mesh is determined through two screening processes; where (a) is the xz-plane diagram and (b) is the xy-plane diagram.

[0047] Figure 3 Three cases are used to determine the relationship between a point P and the inside / outside of a triangular mesh: (a) point Q is inside the triangle; (b) point Q is on the edge of the triangle; (c) point Q is at the vertex of the triangle.

[0048] Figure 4 This is a schematic diagram illustrating the generation of initial particles;

[0049] Figure 5 A schematic diagram of boundary particle constraints;

[0050] Figure 6 A centrifugal pump model and the effect after particle discretization;

[0051] Figure 7The particle distribution in the impeller model discretization process is shown in three states: initial particle distribution, particle compensation, and particle relaxation. (a) Initial particle distribution; (b) Particle compensation distribution; (c) Particle distribution after relaxation.

[0052] Figure 8 A diagram showing the velocity distribution of fluid particles after the centrifugal pump is filled with fluid.

[0053] Figure 9 This is a cross-sectional view of the pressure distribution during the rotation of a centrifugal pump.

[0054] Figure 10 This is a schematic diagram of particle compensation.

[0055] Wherein, 1-volute; 2-impeller. Detailed Implementation

[0056] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. 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 should fall within the scope of protection of the present invention.

[0057] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0058] The present invention will now be described in further detail with reference to the accompanying drawings:

[0059] See Figure 1 This application discloses a particle discretization method for complex geometric models, comprising the following steps:

[0060] S1: Import the geometric model and construct the LSDF;

[0061] S2: Generate initial particles based on LSDF and perform particle compensation on the initial particles;

[0062] S3: Perform particle relaxation on the initial particles and the compensated particles until the particles are uniform.

[0063] A highly efficient and accurate method for constructing narrow-band LSDF on the geometric surface of a model is proposed, and particle compensation is performed to avoid the problem of increased particle spacing after particle relaxation. Based on this method, by inputting the geometric model to be discretized and the discretized particle spacing, complex geometric models can be discretized into uniform, body-fitting particles with a specified spacing. This solves the difficulties of insufficient accuracy and efficiency of existing LSDF techniques, as well as the large volume error after discretizing the model into particles. By inputting an STL geometric model file and particle spacing, the model can be automatically discretized into a body-fitting, uniform particle model, which can be applied to numerical calculations using particle methods. The accuracy and universality of the method and apparatus in this invention are verified by particle discretization of various complex geometries.

[0064] In some embodiments, the import of the geometric model to construct the LSDF specifically includes the following steps:

[0065] S101: Information about each triangular mesh in the analytical geometric model;

[0066] S102: Determine the boundary grid through two grid filtering processes based on the information from the triangular grid;

[0067] S103: Determine the sign value of the boundary grid distance based on the positional relationship between the boundary grid and the adjacent triangular grid, and construct the LSDF of the narrow band region of the geometric surface.

[0068] A method for constructing narrow-band LSDF on the geometric surface of a model is proposed. The boundary mesh is determined by two screenings without the need to construct a global distance field. The sign and distance of the boundary mesh are obtained by precise numerical calculation, and the non-boundary mesh is quickly determined by an ergonomic approach, which greatly reduces the amount of computation.

[0069] In some embodiments, in S102, the first grid filtering is as follows:

[0070] S1021: Determine a cubic region that completely contains the triangular mesh based on the information of the triangular mesh, denoted as region I;

[0071] S1022: Expand region I by 1 / 2 the narrowband width to obtain region II.

[0072] In some embodiments, in S102, the second grid filtering is as follows:

[0073] Calculate the distance between the center point of the grid and the triangle in the region obtained after the first grid screening. If the distance between the center point of the grid and the triangle is not greater than 1 / 2 of the narrow band width, then the grid belongs to the boundary grid. If it is greater, then the grid does not belong to the boundary grid.

[0074] In some embodiments, generating initial particles based on LSDF and performing particle compensation on the initial particles includes the following steps:

[0075] S201: Determine the symbol for the non-boundary mesh based on LSDF;

[0076] S202: For non-boundary meshes, extract the center of the internal mesh as the initial particle;

[0077] S203: For boundary meshes, if the signed distance value is less than -1 / 2 times the mesh width, extract the mesh center point as a particle;

[0078] S204: For meshes crossed by LSDF boundaries, calculate the volume error and perform particle compensation based on the volume error.

[0079] A particle compensation model is proposed. Based on the positional relationship between the grid and the geometric boundary at the boundary, the missing volume at the boundary is statistically analyzed and particles are added. This effectively reduces the volume error after the model is discretized into particles, thereby avoiding the problem of increased particle spacing after particle relaxation.

[0080] In some embodiments, the method for calculating the volume error in step S203 is as follows:

[0081]

[0082] The volume V of the particle model p for:

[0083]

[0084] If defined The volume error can then be determined by the particle number error σ. N To indicate:

[0085]

[0086] Where V is the actual volume of the model; V p l0 represents the volume of the particle model; N represents the actual number of particles; l0 represents the mesh width, which is also the particle spacing; N t This represents the number of particles in the particle model.

[0087] In some embodiments, particle migration technology is used to relax the initial particles and the compensated particles until the particles are uniform. The particle migration algorithm is used to adjust the particle positions while constraining the particles to not flow out of the boundary, so that the particles are attached to the model boundary and uniformly distributed; the CFL condition is introduced in the particle migration to improve the stability of the relaxation process.

[0088] [Example] See Figure 6 Let's take a centrifugal pump as an example for implementation:

[0089] S1: Import the geometric model and construct the LSDF;

[0090] After importing the geometric model, a background mesh is constructed, and the signed distance information of the mesh to the model surface within a narrow band region is calculated (a positive sign indicates the mesh is outside the model, and a negative sign indicates it is inside the model). The steps for constructing an LSDF are as follows:

[0091] S101: Analytical geometric model data. STL format models are widely used in industries such as industry and 3D printing. The file records detailed information about each triangle mesh that makes up the geometric model, including the coordinates of the three vertices and the normal vector.

[0092] S102: Determine the boundary mesh. This invention uses a two-stage filtering method to quickly determine the boundary mesh, reducing computational load. First mesh filtering: Based on the vertex coordinates of the triangles, determine a cubic region (denoted as region I) that completely contains them. Figure 2 (As shown in the red dashed box); Expand region 1 by δ layers (2δ is the narrow band width) to obtain region II (as shown in the red dashed box). Figure 2 (As shown in the red solid box in the middle), the grid in this area is the grid after the first screening; the second grid screening: calculate the distance and position relationship between the center point of the grid and the triangle after the first screening. If the distance from the center point of the grid to the triangle is less than or equal to δ, then the grid belongs to the boundary grid.

[0093] S103: Determine the sign of the boundary grid distance. Let P be the center point of boundary grid i, and Q be the point on the triangular grid closest to P. This can be determined by the normal vector (pseudo-normal vector) n of the triangle. p The angle between N and vector QP is used to determine this: if N p If QP≤0, then point P is located inside the geometry, and the distance value of the boundary grid is negative; if N p If QP > 0, then point P is located outside the geometry, and the distance value of the boundary grid is positive. Based on the positional relationship between point Q and the triangular grid, there are three cases.

[0094] Case 1: Point Q is located inside the triangle. For example... Figure 3 As shown in (a). At this time, n p It is equal to the normal vector n0 of the triangle containing point Q;

[0095] Case 2: Point Q is located on the edge of the triangle, such as... Figure 3 As shown in (b). At this point, it is necessary to construct a pseudo-normal vector using the normal vectors of the two triangles, n. p Calculated using formula (2):

[0096]

[0097] In the formula, n0 and n1 are the normal vectors of the two triangles containing point Q.

[0098] Case 3: Point Q is located at a vertex of the triangle, such as... Figure 3 As shown in (c). At this point, it is necessary to construct a pseudo-normal vector by weighting the angles of all triangles (at least three) that possess this vertex, as shown in formula (3):

[0099]

[0100] In the formula n i Let α be the normal vector of the triangle with vertex Q. i Let Q be the angle between the two sides of the triangle originating from point Q.

[0101] Thus, through three steps, a narrow band width of 2δ near the geometric surface was constructed.

[0102] S2: Generate initial particles based on LSDF and perform particle compensation on the initial particles.

[0103] S201: Determine the sign of non-boundary meshes. Using the sign values ​​of the boundary meshes, the sign values ​​of all non-boundary meshes within the computational domain can be quickly determined using a traversal approach: Traverse the background mesh along any axis; if the preceding and following edges of a row of non-boundary meshes are both inner boundary meshes, then this row of meshes consists of inner meshes.

[0104] S202: Particles are generated at the center of the internal mesh. See also... Figure 4 Typically, the center of boundary particles is offset inwards by 1 / 2 particle spacing from the model boundary to prevent particle overlap at the interface of multiple models. This offset boundary line is called the particle center boundary. For non-boundary meshes, since they are far from the geometric boundary, the mesh center of all internal non-boundary meshes can be extracted as particle points.

[0105] S203: Boundary grid located in the narrow band region. If the distance φ from the center point of the grid to the boundary is less than -0.5l0, then the center point of the grid is extracted as a particle point.

[0106] S204: Particle Compensation. For complex geometries, after generating particles at the center of the internal mesh in S202 and S203, the volume of the internal mesh is often smaller than the actual volume of the model. This results in increased particle spacing after particle relaxation, affecting the accuracy of numerical calculations. To solve this problem, this invention proposes a particle compensation method. For meshes crossed by LSDF boundaries, the volume error is calculated, and particle compensation is performed based on the volume error. The method for calculating the volume error is as follows:

[0107]

[0108] The volume V of the particle model p for:

[0109]

[0110] If defined The volume error can then be determined by the particle number error σ. N To indicate:

[0111]

[0112] Where V is the actual volume of the model; V p l0 represents the volume of the particle model; N represents the actual number of particles; l0 represents the mesh width, which is also the particle spacing; N t This represents the number of particles in the particle model.

[0113] The principle of particle compensation is as follows: traverse and find the meshes that are crossed by the model boundary, and calculate the volume of the meshes inside and outside the model boundary (denoted as V respectively). 内 V 外 If the sum equals 1), and the mesh does not generate particles in S202 and S203, then a portion of the volume is missing, such as... Figure 10 As shown, the total volume error σ = σ - V 内 If the volume error σ < -C (where C is the particle compensation threshold, typically 0.8) during the traversal, particles are generated at the center of the grid to compensate for the volume error, and the volume error σ = σ + 1. This process is repeated to calculate all grids and complete particle compensation.

[0114] S3: Perform particle relaxation on the initial particles and the compensated particles until the particles are uniform.

[0115] The initially generated and compensated particle surface exhibits a matrix shape, which does not conform to the geometric model surface, requiring particle relaxation and homogenization. This invention employs particle migration technology to achieve this process. The distance that particle i moves each time is shown in formula (6):

[0116]

[0117] In the formula Cshift C is the migration coefficient. shift ≤0.5; l0 is the initial particle spacing. Let be the particle number density gradient. To ensure the stability of the relaxation effect, the CFL condition is introduced, where the particle velocity is...

[0118]

[0119] Where: Δt init This is the initial time step. The formula for the CFL condition is:

[0120]

[0121] In the formula: u max Let U be the particle with the highest velocity among all particles, and C be the CFL constant, which is taken as 0.15. After calculating u... max Then, if the CFL condition is satisfied, then Δt = Δt init Otherwise, Δt = Cl0 / u max Therefore, the particle travels a distance in one iteration is: δr i ′=u i Δt

[0122] like Figure 5 As shown, during the relaxation process, particles near the boundary expand outwards because there are no particles on the outside. This paper employs a boundary particle constraint method, which moves the outer particles back to the particle center boundary line when they exceed the boundary line, ensuring that particles near the boundary remain close to the geometric surface.

[0123] After multiple relaxations, the particles gradually become uniform. This is because the average distance between the boundary particles and the centerline boundary of the particle system... The particle migration ends when certain conditions are met. The calculation formula is

[0124]

[0125] Where: φ j Let n be the distance from particle j to the surface of the geometry; b Let be the number of boundary particles. The condition for determining the number of boundary particles is:

[0126]

[0127] In the formula C b The boundary particle judgment coefficient is typically chosen to be 0.8. The criterion for achieving stability during relaxation is:

[0128]

[0129] In the formula: n is the number of particle migration iterations, Cps This is the particle migration convergence coefficient, typically taken as 0.015.

[0130] See Figure 7 , Figure 8 and Figure 9 As can be seen, this application proposes an efficient and accurate method for constructing a narrow-band LSDF for the geometric surface of a model. The boundary mesh is determined through two screening processes without requiring the construction of a global distance field. The sign and distance of the boundary mesh are obtained through precise numerical calculations, while the non-boundary mesh is quickly determined using an traversal-progression method, significantly reducing computational load. A particle compensation model is proposed, which, based on the positional relationship between the boundary mesh and the geometric boundary, statistically analyzes the missing volume at the boundary and supplements it with particles, effectively reducing the volume error after the model is discretized into particles, thus avoiding the problem of increased particle spacing after particle relaxation. A particle migration algorithm is used to adjust the particle position while constraining particles to not flow out of the boundary, ensuring that particles adhere to the model boundary and are uniformly distributed. A CFL condition is introduced in particle migration to improve the stability of the relaxation process. This invention solves the difficulties of insufficient accuracy and efficiency of existing LSDF methods and the large volume error after the model is discretized into particles. By inputting an STL geometric model file and particle spacing, the model can be automatically discretized into a body-fitting, uniform particle model, which can be applied to numerical calculations using particle methods. The accuracy and universality of the method and apparatus in this invention are verified by particle discretization of various complex geometries.

[0131] This application also discloses a particle discretization system with complex geometric models, including:

[0132] The LSDF build module is used to import geometric models and build LSDF.

[0133] The particle generation module is used to generate initial particles based on LSDF and perform particle compensation on the initial particles.

[0134] The particle relaxation module is used to relax the initial particles and the compensated particles until the particles are uniform.

[0135] This application also discloses an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the particle discretization method for complex geometric models described above.

[0136] This application also discloses a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the particle discretization method for complex geometric models described above.

[0137] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, apparatus, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0138] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0139] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0140] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable apparatus for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0141] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A particle discretization method for complex geometric models, characterized in that, Includes the following steps: Importing the geometric model of the centrifugal pump to construct the LSDF involves the following steps: S101: Information about each triangular mesh in the analytical geometric model; S102: Determine the boundary grid through two grid filtering processes based on the information from the triangular grid; S103: Determine the sign value of the boundary grid distance based on the positional relationship between the boundary grid and the adjacent triangular grid, and construct the LSDF of the narrow band region of the geometric surface; The process of generating initial particles based on LSDF and performing particle compensation on the initial particles includes the following steps: S201: Determine the symbol for the non-boundary mesh based on LSDF; S202: For non-boundary meshes, extract the center of the internal mesh as the initial particle; S203: For boundary meshes, if a signed distance value is less than -1 / 2 times the mesh width, extract the mesh center point as a particle; S204: For the mesh that is crossed by the LSDF boundary, calculate the volume error and perform particle compensation based on the volume error; Particle migration technology is used to relax the initial particles and the compensated particles until the particles are uniform.

2. The particle discretization method for complex geometric models according to claim 1, characterized in that, In step S102, the first grid screening is as follows: S1021: Determine a cubic region that completely contains the triangular mesh based on the information of the triangular mesh, denoted as region I; S1022: Expand region I by 1 / 2 the narrow band width to obtain region II.

3. The particle discretization method for complex geometric models according to claim 1, characterized in that, In step S102, the second grid filtering is as follows: Calculate the distance between the center point of the grid and the triangle in the region obtained after the first grid screening. If the distance between the center point of the grid and the triangle is not greater than 1 / 2 of the narrow band width, then the grid belongs to the boundary grid. If it is greater, then the grid does not belong to the boundary grid.

4. The particle discretization method for complex geometric models according to claim 1, characterized in that, In step S204, the method for calculating the volume error is as follows: （3） Volume of the particle model for: （4） If defined = The volume error can then be determined by the particle number error. To indicate: （5） Where V is the actual volume of the model; The volume of the particle model is N; the actual number of particles is N. This represents the grid width, and also the particle spacing. This represents the number of particles in the particle model.

5. A particle discrete system with a complex geometric model, characterized in that, A particle discretization method for complex geometric models as described in any one of claims 1 to 4, comprising: The LSDF build module is used to import geometric models and build LSDF. The particle generation module is used to generate initial particles based on LSDF and perform particle compensation on the initial particles. The particle relaxation module is used to relax the initial particles and the compensated particles until the particles are uniform.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the steps of the particle discretization method for complex geometric models according to any one of claims 1-4.

7. A computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the particle discretization method for complex geometric models according to any one of claims 1-4.

Images