Model quality detection method, system and electronic equipment

Abstract

The present invention provides a model quality detection method, system and electronic device, including: obtaining a model to be detected sent by a user; determining the grid size based on the correspondence between the model scene type and the preset model scene type grid size; dividing the model to be detected into multiple local models; judging whether the division of the local model is successful; if the division of the local model fails, identifying abnormal geometric features of the model to be detected, and marking the identified abnormal geometric features; if the division of the local model is successful, performing grid quality detection on each local model; identifying the grid quality detection result of each local model; if the grid quality detection result is that the grid is unqualified, identifying abnormal geometric features of the local model, and marking the identified abnormal geometric features. This method can identify abnormal edges in the model that are prone to cause stress concentration and finite element model quality problems, thereby improving the efficiency of model detection and the reliability of numerical analysis.

Description

Technical Field

[0001] The present invention relates to the technical field of three-dimensional model detection, and in particular to a model quality detection method, system and electronic equipment. Background Art

[0002] With the development of intelligent manufacturing, numerical calculation has become crucial in the product design and manufacturing process, and is widely used in fields such as finite element analysis (FEA) and computational fluid dynamics (CFD) to predict the performance of products under different working conditions. Although cloud-based numerical calculation industrial apps have simplified the complex numerical analysis process and allowed non-professional users to participate, the 3D models uploaded by users are usually not pre-processed, and there are quality problems such as geometric discontinuity, meshing difficulties, and stress concentration, which seriously affect the accuracy of numerical analysis. Existing model quality detection methods mostly rely on manual inspection for abnormal edge identification, which is prone to missed detection or misjudgment. Summary of the invention

[0003] In view of this, the purpose of the present invention is to provide a model quality detection method, a model quality detection system and an electronic device, which can accurately identify abnormal edges in the model that are likely to cause stress concentration and finite element model quality problems, thereby reducing the complexity and risk of misjudgment of manual inspection, and further improving the efficiency of model quality detection and the reliability of numerical analysis.

[0004] In the first aspect, an embodiment of the present invention provides a model quality detection method, including: obtaining a model to be detected sent by a user, geometric feature data corresponding to the model to be detected, and a model scene type; the geometric feature data is used to characterize the correspondence between vertices, edges, and faces in the model to be detected; based on the model scene type and the preset model scene type grid size correspondence, determining the grid size corresponding to the model to be detected; based on the grid size, dividing the model to be detected into multiple local models; judging whether the division of the local model is successful; if the division of the local model fails, performing abnormal geometric feature recognition on the model to be detected, and marking the recognized abnormal geometric features; if the division of the local model is successful, performing mesh quality detection on each local model; identifying the mesh quality detection result of each local model; if the mesh quality detection result is that the mesh is unqualified, performing abnormal geometric feature recognition on the local model, and marking the recognized abnormal geometric features.

[0005] Furthermore, the step of obtaining geometric feature data corresponding to the model to be detected includes: extracting geometric elements corresponding to the model to be detected; the geometric elements include vertex data, edge data and face data; performing topological analysis on the geometric elements, identifying each edge in the model to be detected and the adjacent faces connected to each edge, and establishing a mapping relationship between the edges and the adjacent faces; wherein the mapping relationship between the edges and the adjacent faces is used to characterize the association relationship between the edge ID corresponding to the edge and the face ID corresponding to the adjacent face; combining the geometric elements and the mapping relationship between the edges and the adjacent faces to form the geometric feature data corresponding to the model to be detected.

[0006] Furthermore, abnormal geometric feature recognition includes sharp edge recognition; abnormal geometric features include sharp edges; the steps of performing abnormal geometric feature recognition on the model to be detected and marking the identified abnormal geometric features include: determining the angle value between adjacent faces connected to each edge in the model to be detected based on the geometric feature data; comparing each angle value with a pre-set angle threshold, and marking the edge corresponding to the angle value less than the angle threshold as a sharp edge.

[0007] Furthermore, before the step of comparing each angle value with the angle threshold, the method also includes: performing topological analysis on the geometric elements to determine the face complexity, edge length difference, vertex density, curvature change normalization value and topological structure complexity corresponding to the model to be detected; based on the face complexity, edge length difference, vertex density, curvature change normalization value, topological structure complexity and preset complexity weights, calculating the geometric complexity score corresponding to the model to be detected; based on the geometric complexity score and the mapping relationship between the preset geometric complexity score and the angle threshold, determining the angle threshold corresponding to the model to be detected.

[0008] Furthermore, based on the mapping relationship between the geometric complexity score and the preset geometric complexity score and the angle threshold, the step of determining the angle threshold corresponding to the model to be detected includes: identifying the geometric complexity score; if the geometric complexity score is less than the preset first threshold, determining that the model to be detected is a simple model, and determining the angle threshold to be the first angle threshold; if the geometric complexity score is greater than or equal to the preset first threshold and less than the preset second threshold, determining that the model to be detected is a medium model, and determining the angle threshold to be the second angle threshold; if the geometric complexity score is greater than or equal to the preset first threshold, determining that the model to be detected is a complex model, and determining the angle threshold to be the third angle threshold; wherein, the preset first threshold is less than the preset second threshold; the first angle threshold is greater than the second angle threshold and greater than the third angle threshold.

[0009] Furthermore, abnormal geometric feature recognition includes edge length recognition; abnormal geometric features include short edges; the steps of performing abnormal geometric feature recognition on the model to be detected and marking the identified abnormal geometric features include: determining the length of each edge in the model to be detected based on the geometric feature data; comparing each edge length with the grid size, and marking the edges corresponding to the edge lengths smaller than the grid size as short edges.

[0010] Furthermore, the step of performing mesh quality detection on each local model includes: determining the number of faces in each local model based on geometric feature data; determining the local model type of the local model based on the geometric feature data and the number of faces; the local model type is a tetrahedral mesh or a hexahedral mesh; calculating the mesh quality detection parameters corresponding to each local model based on the local model type and a preset mesh quality detection method; the mesh quality detection parameters include aspect ratio and / or distortion and / or warping factor; judging whether each mesh quality detection parameter is within a preset mesh quality detection parameter threshold; if the mesh quality detection parameter is within the preset mesh quality detection parameter threshold, determining that the mesh quality detection result of the local model corresponding to the mesh quality detection parameter is a qualified mesh; if the mesh quality detection parameter is not within the preset mesh quality detection parameter threshold, determining that the mesh quality detection result of the local model corresponding to the mesh quality detection parameter is an unqualified mesh.

[0011] Furthermore, after the step of marking the identified abnormal geometric features, the method also includes: obtaining key data of the edge corresponding to each abnormal geometric feature; the key data includes the edge ID and angle value corresponding to the edge; generating an abnormal edge data file based on the key data; loading the model to be detected and the abnormal edge data file corresponding to the model to be detected on a pre-set visualization platform, and highlighting each abnormal geometric feature using preset color coding.

[0012] Furthermore, on a pre-set visualization platform, after loading the model to be detected and the abnormal edge data file corresponding to the model to be detected, and highlighting each abnormal geometric feature using a preset color code, the method also includes: identifying the user's interactive operations on the visualization platform; if the interactive operation is a preset abnormal edge identification operation, obtaining the abnormal edge position information corresponding to the interactive operation, and displaying the key data corresponding to the edge corresponding to the abnormal edge position information.

[0013] In the second aspect, an embodiment of the present invention provides a model quality detection system, including: a geometric feature data determination module, which is used to obtain the model to be detected sent by the user, the geometric feature data corresponding to the model to be detected, and the model scene type; the geometric feature data is used to characterize the correspondence between the vertices, edges and faces in the model to be detected; a grid division module, which is used to determine the grid size corresponding to the model to be detected based on the model scene type and the preset model scene type grid size correspondence; the grid division module is also used to divide the model to be detected into multiple local models based on the grid size; an abnormal geometric feature marking module, which is used to determine whether the division of the local model is successful; the abnormal geometric feature marking module is also used to identify abnormal geometric features of the model to be detected if the division of the local model fails, and mark the identified abnormal geometric features; the abnormal geometric feature marking module is also used to perform grid quality detection on each local model if the division of the local model is successful; identify the grid quality detection result of each local model; if the grid quality detection result is that the grid is unqualified, identify abnormal geometric features of the local model, and mark the identified abnormal geometric features.

[0014] The embodiment of the present invention provides a model quality detection method, system and electronic device, including: obtaining a model to be detected sent by a user, geometric feature data corresponding to the model to be detected and a model scene type; the geometric feature data is used to characterize the correspondence between vertices, edges and faces in the model to be detected; based on the model scene type and the preset model scene type grid size correspondence, determine the grid size corresponding to the model to be detected; based on the grid size, divide the model to be detected into multiple local models; judge whether the division of the local model is successful; if the division of the local model fails, identify abnormal geometric features of the model to be detected, and mark the identified abnormal geometric features; if the division of the local model is successful, perform grid quality detection on each local model; identify the grid quality detection result of each local model; if the grid quality detection result is that the grid is unqualified, identify abnormal geometric features of the local model, and mark the identified abnormal geometric features. In this way, abnormal edges in the model that are prone to cause stress concentration and finite element model quality problems can be accurately identified, thereby reducing the complexity and misjudgment risk of manual inspection, thereby improving the efficiency of model detection and the reliability of numerical analysis.

[0015] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description, claims and drawings.

[0016] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0018] Figure 1 A flow chart of a model quality detection method provided in Embodiment 1 of the present invention;

[0019] Figure 2 A flow chart of a method for setting geometric feature data provided in Embodiment 1 of the present invention;

[0020] Figure 3 A flowchart of a method for identifying and marking identified abnormal geometric features when the abnormal geometric feature identification provided in the first embodiment of the present invention includes sharp edge identification;

[0021] Figure 4 A flow chart of a method for determining an angle threshold provided in Embodiment 1 of the present invention;

[0022] Figure 5 A flow chart of a method for setting a clamping threshold value provided in Embodiment 1 of the present invention;

[0023] Figure 6 A flowchart of a method for identifying and marking identified abnormal geometric features when the abnormal geometric feature identification provided in the first embodiment of the present invention includes short edge identification;

[0024] Figure 7 A flow chart of a method for performing mesh quality detection on each local model provided in the first embodiment of the present invention;

[0025] Figure 8 A schematic diagram of a warping factor mapping relationship provided in Embodiment 1 of the present invention;

[0026] Fig. 9 A schematic diagram of setting the clamping threshold provided in the first embodiment of the present invention;

[0027] Fig.10 A schematic diagram of the model sharp edge detection result provided in the first embodiment of the present invention;

[0028] Fig.11 A schematic diagram of a model quality detection system provided in Embodiment 2 of the present invention;

[0029] Fig.12 A schematic diagram of another model quality detection system provided in Embodiment 2 of the present invention.

[0030] Icons: 1-geometric feature data determination module; 2-grid division module; 3-abnormal geometric feature marking module; 4-model display module. DETAILED DESCRIPTION

[0031] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0032] With the continuous development of intelligent manufacturing, numerical calculation plays an increasingly important role in product design and manufacturing. Numerical analysis techniques, such as finite element analysis (FEA) and computational fluid dynamics (CFD), have become indispensable tools for engineers in the product development process. These technologies can effectively predict the performance of products under different working conditions and improve product quality by optimizing design and reducing costs. However, traditional numerical analysis software is highly complex and has a high operating threshold. It has high requirements on the user's professional background and technical level, and often requires professional engineers to operate and analyze. For users without a professional background, this threshold undoubtedly hinders the popularization and application of numerical analysis technology.

[0033] In recent years, the rapid development of cloud computing and big data technology has brought new opportunities for the promotion and application of numerical computing technology. Cloud-based numerical computing industrial apps have emerged. These apps encapsulate complex numerical analysis processes in a simple and easy-to-use user interface, making it easy for non-professional users to get started. However, in practical applications, most of the three-dimensional models provided by users are often original models without pre-processing, rather than numerical analysis models that can be directly meshed and post-processed. These original models may have many quality problems, such as geometric discontinuity, meshing difficulties, stress concentration, etc., which affect the accuracy and effectiveness of numerical analysis. Among the many quality problems, sharp edges, that is, edges with angles less than a certain threshold, are often the main source of stress concentration. If not handled, these sharp edges may cause distortion of numerical analysis results. Short edges, that is, edges with a length shorter than the mesh size, often lead to poor meshing quality and inaccurate numerical calculation results.

[0034] This embodiment provides a model quality detection method, which can identify abnormal edges that are likely to cause stress concentration and meshing difficulties, thereby performing effective model quality detection before numerical analysis, identifying and processing potential problems in the model, and thus improving the reliability and efficiency of numerical analysis.

[0035] To facilitate understanding of this embodiment, the embodiment of the present invention is described in detail below.

[0036] Embodiment 1:

[0037] Figure 1 This is a flow chart of the model quality detection method provided in Example 1 of the present invention.

[0038] Reference Figure 1 ,Model quality detection methods, including:

[0039] Step S101, obtaining the model to be detected sent by the user, geometric feature data corresponding to the model to be detected, and the model scene type; the geometric feature data is used to characterize the corresponding relationship between vertices, edges and faces in the model to be detected.

[0040] Here, the model to be tested is an unprocessed three-dimensional model. Before uploading the unprocessed three-dimensional model to the numerical calculation APP (such as developing scenario-based lightweight numerical calculation industrial applications), the sharp edges, short edges, and small angles in the three-dimensional model, which are prone to stress concentration, are identified, thereby reducing the problem of calculation failure caused by low quality of user models and improving the functions and effects of the numerical calculation APP.

[0041] In one embodiment, referring to Figure 2 , the steps of step S101 include:

[0042] Step S201, extracting geometric elements corresponding to the model to be detected; the geometric elements include vertex data, edge data and face data.

[0043] Here, the 3D model file corresponding to the model to be detected is read, and common file formats include STL, OBJ, STEP, etc. The geometric elements corresponding to the model to be detected are extracted, and the geometric elements include vertex data, edge data, and face data.

[0044] Vertex data may be the coordinates of the vertices, edge data may be the connection relationship of the edges, and face data may be which vertices constitute a face.

[0045] The geometric elements are organized according to the relationship between edges, faces and vertices and stored in the form of arrays.

[0046] The model scene type refers to the specific purpose or product category of the model to be inspected, such as shelves, tool parts, mechanical parts, etc. Different types of models usually have different geometric complexity and structural characteristics, so different meshing strategies need to be selected.

[0047] Step S202, performing topological analysis on the geometric elements, identifying each edge in the model to be detected and the adjacent faces connected to each edge, and establishing a mapping relationship between the edge and the adjacent faces; wherein the mapping relationship between the edge and the adjacent faces is used to characterize the association relationship between the edge ID corresponding to the edge and the face ID corresponding to the adjacent face.

[0048] Here, a topological analysis is performed on the geometric elements, that is, the connection relationship between vertices, edges and faces is analyzed in the geometric space. All edges are identified, and the adjacent faces connected to each edge are found. A unique edge ID is assigned to each edge based on a preset coding rule, and a unique face ID is assigned to each adjacent face. The preset coding rule is pre-set according to the actual situation.

[0049] Based on the edge ID, the face ID, and the corresponding relationship between each edge and the adjacent face, a mapping relationship between the edge and the adjacent face is established.

[0050] Specifically, when retrieving each edge, the edge ID will be recorded and the face connected to the edge will be retrieved and its face ID will be recorded. At this time, each edge will have a corresponding face ID, and there is a mapping relationship between the edge and face ID numbers. Each edge corresponds to a set of feature information (including the corresponding edge and face IDs and the corresponding geometric elements).

[0051] Step S203: combining the geometric elements and the mapping relationship between the edges and the adjacent faces to form geometric feature data corresponding to the model to be detected.

[0052] Here, the geometric feature data includes vertices, edges, adjacent faces, vertex data, edge data, face data and the mapping relationship between edges and adjacent faces corresponding to the model to be detected. Among them, the edge data includes the coordinate information of the internal points of the edge, and the face data includes the coordinates of the midpoints of the face and the internal nodes of the face.

[0053] By matching the position information of all faces and edges, the two adjacent faces of each edge can be determined (the position of a common edge is shared by two adjacent faces, and the face data contains the edge data of all edges).

[0054] Specifically, assuming that a user uploads a cube model file, the system will read the data of the cube's 8 vertices and 12 edges:

[0055] Vertex data: V = [[0, 0, 0], [1, 0, 0], [1, 1, 0], [0, 1, 0], [0, 0, 1],[1, 0, 1], [1, 1, 1], [0, 1, 1]].

[0056] Edge data: E = [[0, 1], [1, 2], [2, 3], [3, 0], [4, 5], [5, 6], [6, 7],[7, 4], [0, 4], [1, 5], [2, 6], [3, 7]].

[0057] Face data: Each face is composed of four vertices. For example, the bottom face is composed of vertices [0, 1, 2, 3].

[0058] For a bottom edge [0, 1] in the cube model, this edge connects the bottom face [0, 1, 2, 3] and the front face [0, 1, 5, 4] at the same time, and the mapping relationship between the edge ID of this edge and the face ID of its adjacent face is recorded. Among them, the edge ID is E0, and its corresponding face IDs are F0 (bottom face) and F1 (front face).

[0059] Step S102, based on the corresponding relationship between the model scene type and the preset model scene type grid size, determine the grid size corresponding to the model to be detected.

[0060] Here, the correspondence between model scene type grid size is the default grid size of different types of models pre-set according to actual conditions. The grid size will be used for model meshing to ensure that the geometric features of the model are represented while avoiding excessive computational complexity due to too dense a grid.

[0061] For example, some models (such as shelf products) can accept a larger mesh size, while power tool parts with more details require a smaller mesh size. Specifically, the default mesh size for shelf products is 10mm. The default mesh size for power tool parts is 5mm.

[0062] Step S103: dividing the model to be detected into multiple local models based on the grid size.

[0063] Specifically, suppose a user uploads a 3D model of a power tool part. It is identified as a model scene type of power tool parts. Based on the preset model scene type grid size correspondence, 5 mm is determined as the grid size of the model to be detected.

[0064] The model is meshed to divide it into multiple local models. In this process, an attempt is made to generate mesh elements of appropriate size and shape to ensure that the entire model geometry is covered.

[0065] Step S104, determining whether the division of the local model is successful.

[0066] Step S105: If the division of the local model fails, abnormal geometric features are identified on the model to be detected, and the identified abnormal geometric features are marked.

[0067] Here, if the grid where a local model is located is too distorted or has an inappropriate density, it is determined that the division of the local model fails.

[0068] Abnormal geometric features are those geometric characteristics that may cause meshing failure or poor quality, such as short edges and sharp edges.

[0069] In one embodiment, the identification of unusual geometric features includes identification of sharp edges; the unusual geometric features include sharp edges.

[0070] Reference Figure 3 In step S105, the step of identifying abnormal geometric features of the model to be detected and marking the identified abnormal geometric features includes:

[0071] Step S301, based on the geometric feature data, determine the angle value between adjacent faces connected to each edge in the model to be detected.

[0072] Here, based on the geometric feature data, the normal vector corresponding to each adjacent face is calculated.

[0073] Here, assuming that a target surface is defined by points A, B, and C, when calculating the normal vector, we first need to extract the coordinates of the three vertices on the target surface and calculate the normal vector of the target surface using the following cross product formula (1):

[0074] (1)

[0075] Where n represents the normal vector of the target surface, BA represents the vector from point A to point B, and CA represents the vector from point A to point C.

[0076] Based on the normalization algorithm, each normal vector is converted to a unit normal vector.

[0077] Here, the normalized unit normal vector retains the directionality of the normal vector, but its length becomes 1.

[0078] Each normal vector is converted to a unit normal vector by calculating the following formula (2).

[0079] (2)

[0080] in, is the unit normal vector, is the magnitude (length) of the normal vector.

[0081] The magnitude of the normal vector is obtained by calculating the square root of the sum of the squares of the vector components using the following formula (3).

[0082] (3)

[0083] Based on the unit normal vector corresponding to each adjacent face and the mapping relationship between the edge and the adjacent face, the angle value corresponding to each edge ID is determined.

[0084] Here, it is assumed that the normal vectors of the two adjacent faces of an edge are and , angle Calculated by formula (4):

[0085] (4)

[0086] in, represents the dot product of the normal vector, and They represent the magnitude of the normal vector respectively.

[0087] The dot product is used to calculate the angle between two vectors. The larger the dot product (closer to 1), the smaller the angle; the smaller the dot product (closer to -1), the larger the angle.

[0088] Since the normal vector has been normalized, the formula is simplified to formula (5):

[0089] (5)

[0090] Calculate the angle value of each edge and store the angle value corresponding to each edge in association with the edge ID. The stored data structure can be a list containing edge IDs and angle values, for example:

[0091]

[0092] in, represents the edge ID, Indicates the angle value corresponding to the edge.

[0093] Step S302: Compare each angle value with an angle threshold, and mark the edge corresponding to the angle value smaller than the angle threshold as a sharp edge.

[0094] In one embodiment, referring to Figure 4 In step S302, before the step of comparing each angle value with the angle threshold, the method further includes:

[0095] Step S401 , performing topological analysis on geometric elements to determine the face complexity, edge length difference, vertex density, curvature change normalization value and topological structure complexity corresponding to the model to be detected.

[0096] Here, the number of facets in the model to be detected and the number of edges corresponding to each facet are identified, and the sum of the number of edges corresponding to each facet is determined as the facet complexity.

[0097] The polygonal patches in the model are more complex than the triangular patches, so the patch complexity of the model to be detected is calculated based on the number of patches and the number of edges corresponding to each face.

[0098] Specifically, if there are 200 facets in the model to be detected, and each facet consists of 3 edges, the facet complexity M is calculated to be 600.

[0099] The side length corresponding to each side in the model to be detected is identified, and the quotient of the side with the largest side length and the side with the smallest side length is determined as the side length difference.

[0100] Here, the greater the difference in the length of the sides in the model to be detected is, the higher the geometric complexity corresponding to the model to be detected is.

[0101] Specifically, assuming that the longest side is 10 units and the shortest side is 1 unit, the side length difference B is 10 / 1=10.

[0102] The number of vertices and the surface area of ​​the model to be inspected are identified, and the quotient of the number of vertices and the surface area is determined as the vertex density.

[0103] Here, the vertex density of the model to be detected is high, indicating that the geometric complexity of the model to be detected is high.

[0104] Specifically, assuming 100 vertices and a surface area of ​​500 square units, the vertex density D is 1000 / 500 = 2 (vertices / square unit).

[0105] The curvature variation range corresponding to the model to be detected is calculated, and the curvature variation normalized value is calculated based on the curvature variation range and the preset reference maximum value.

[0106] Here, the normal vector corresponding to each adjacent face is calculated, and the local curvature value of each face is calculated based on the change of the normal vector of the adjacent face. The curvature value K of each vertex is interpolated based on the curvature of the faces adjacent to each vertex.

[0107] Get the maximum curvature value K max and minimum value K min , calculate the curvature variation range corresponding to the model to be detected .

[0108] The larger the curvature variation range, the greater the curvature difference of the model to be detected and the more complex it is.

[0109] Based on the normalization formula, the curvature change is standardized within the range of 0 to 1, and the calculation is ,in, Normalized value for curvature change, preset reference maximum value is the maximum curvature change value preset based on the preset benchmark model, which is used as the denominator for normalization.

[0110] The normalized value of curvature change is between [0,1]. The closer the normalized value of curvature change is to 1, the more drastic the surface change is, and the closer it is to 0, the smoother the surface is.

[0111] The number of complex structures of the model to be tested is identified and determined as the topological complexity; the complex structures include holes, closed areas, and non-manifold geometry.

[0112] Here, the more holes, closed areas and non-manifold geometries there are in the model to be detected, the higher the topological complexity of the model to be detected. Among them, non-manifold geometries refer to an edge in the model to be detected that connects more than two faces, or a vertex that connects more than one continuous boundary.

[0113] Specifically, assuming there are 3 holes and 2 closed areas, these complex structures are identified and the topological complexity J is determined to be 5.

[0114] Step S402, based on the patch complexity, edge length difference, vertex density, curvature change normalization value, topological structure complexity and preset complexity weight, calculate and obtain the geometric complexity score corresponding to the model to be detected.

[0115] Here, in the preset complexity weights, the face complexity weight is set to W1, the edge length difference weight is set to W2, the vertex density weight is set to W3, the curvature change normalization value weight is set to W4, and the topological structure complexity weight is set to W5.

[0116] The geometric complexity score is .

[0117] Step S403: determining an angle threshold corresponding to the model to be detected based on the geometric complexity score and a preset mapping relationship between the geometric complexity score and the angle threshold.

[0118] In one embodiment, referring to Figure 5 , the steps of step S403 include:

[0119] Step S501, identifying a geometric complexity score.

[0120] Step S502: If the geometric complexity score is less than a preset first threshold, the model to be detected is determined to be a simple model, and the angle threshold is determined to be a first angle threshold.

[0121] Here, the preset first threshold and the first angle threshold are preset according to actual conditions. The first threshold can be set to 50, and the first angle threshold can be set to 30°.

[0122] Specifically, a model to be detected has a geometric complexity score of 40 and is identified as a simple model. The angles of a few edges are 15 or 20 degrees, which are lower than the threshold of 30°, so these edges are identified and marked as sharp edges.

[0123] Step S503: If the geometric complexity score is greater than or equal to the preset first threshold and less than the preset second threshold, the model to be detected is determined to be a medium model, and the angle threshold is determined to be the second angle threshold.

[0124] Here, the preset second threshold and the second angle threshold are preset according to actual conditions. The second threshold can be set to 80, and the second angle threshold can be set to 15°.

[0125] Specifically, a model to be detected has a geometric complexity score of 65 and is identified as a medium model. The angles of the edges in some areas are 8° or 9°, which is lower than the threshold of 15°. These edges will be marked as sharp edges.

[0126] Step S504, if the geometric complexity score is greater than or equal to the preset first threshold, the model to be detected is determined to be a complex model, and the angle threshold is determined to be the third angle threshold; wherein the preset first threshold is less than the preset second threshold; the first angle threshold is greater than the second angle threshold which is greater than the third angle threshold.

[0127] Here, the third angle threshold is preset according to actual conditions, and the third angle threshold can be set to 10°.

[0128] Specifically, a model to be tested, whose geometric complexity score is 90, is identified as a complex model, and only the edges with angles less than 10° are marked as sharp edges.

[0129] In one embodiment, each angle value is compared with an angle threshold, and if the angle is less than the threshold, the edge is marked as a sharp edge. The judgment formula (6) is as follows:

[0130] ) (6)

[0131] in, represents the edge ID, Indicates the angle value corresponding to the edge ID. ) is a Boolean value, Indicates the angle threshold.

[0132] The comparison result is used to determine whether to mark the edge as a sharp edge.

[0133] In one embodiment, the abnormal geometric feature identification includes edge length identification; the abnormal geometric feature includes a short edge.

[0134] Reference Figure 6In step S105, the step of identifying abnormal geometric features of the model to be detected and marking the identified abnormal geometric features includes:

[0135] Step S601, based on the geometric feature data, determine the length of each edge in the model to be detected.

[0136] Here, based on the geometric feature data, the geometric information of each edge in the model to be tested is extracted one by one and the edge length is calculated.

[0137] Step S602: Compare each side length with the grid size, and mark the side corresponding to the side length smaller than the grid size as a short side.

[0138] Here, after determining the side length, the side length of each side is compared with the grid size. The grid size is a preset size based on the model scene type, representing a reasonable grid unit division size.

[0139] If the length of an edge is smaller than the mesh size, the edge is marked as short. Marking short edges helps users identify potential problem areas in the model. These areas may produce low-quality meshes when meshing or cause insufficient accuracy in numerical simulation calculations.

[0140] Step S106, if the division of the local model is successful, perform a mesh quality check on each local model; identify the mesh quality check result of each local model; if the mesh quality check result is that the mesh is unqualified, identify abnormal geometric features of the local model and mark the identified abnormal geometric features.

[0141] In one embodiment, referring to Figure 7 In step S106, the step of performing mesh quality detection on each local model includes:

[0142] Step S701, determining the number of faces in each local model based on the geometric feature data.

[0143] Here, after the meshing is successfully done, the number of faces corresponding to each local model obtained by the meshing is read one by one.

[0144] Step S702, determining the local model type of the local model based on the geometric feature data and the number of faces; the local model type is a tetrahedral mesh or a hexahedral mesh.

[0145] Here, the tetrahedral mesh is composed of triangular faces and the hexahedral mesh is composed of quadrilateral faces.

[0146] Step S703, based on the local model type and the preset mesh quality detection method, calculate the mesh quality detection parameters corresponding to each local model; the mesh quality detection parameters include aspect ratio and / or distortion and / or warping factor.

[0147] Here, the preset grid quality detection method includes an aspect ratio detection method, a distortion detection method and a warping factor detection method.

[0148] The mesh quality detection parameters corresponding to the tetrahedral mesh include aspect ratio and distortion, and the mesh quality detection parameters corresponding to the hexahedral mesh include aspect ratio, distortion and warping factor.

[0149] Aspect ratio detection methods include:

[0150] Extract the vertices of the triangular or quadrilateral faces corresponding to each local model and calculate their longest and shortest sides. Calculate the aspect ratio corresponding to each local model: .

[0151] Distortion detection methods include:

[0152] Extract the surface mesh in each local model and determine the shape and characteristics of the surface mesh.

[0153] For tetrahedral meshes, the face elements are triangles, and the ideal angle is 60° (equilateral triangle).

[0154] For hexahedral meshes, the face elements are quadrilaterals, and the ideal angle is 90° (rectangular).

[0155] Extract the face mesh angle value for each face mesh.

[0156] Warpage factor detection methods include:

[0157] For a hexahedral mesh, four nodes of a quadrilateral face of each local model are obtained, for example, numbered as node 1, node 2, node 3, and node 4.

[0158] Reference Figure 8 , first select node 1 as the warping point, and the remaining nodes 2, 3, and 4 form a plane. Calculate the mapping point of node 1 on the plane formed by nodes 2, 3, and 4 (that is, the vertical projection of the point on the plane). Based on the warping factor formula: .

[0159] Step S704, determining whether each mesh quality detection parameter is within a preset mesh quality detection parameter threshold.

[0160] Here, the grid quality detection parameter threshold is preset according to actual conditions.

[0161] Step S705: if the mesh quality detection parameter is within a preset mesh quality detection parameter threshold, determine that the mesh quality detection result of the local model corresponding to the mesh quality detection parameter is a qualified mesh.

[0162] Step S706: if the mesh quality detection parameter is not within the preset mesh quality detection parameter threshold, determine that the mesh quality detection result of the local model corresponding to the mesh quality detection parameter is mesh unqualified.

[0163] Here, the grid quality detection parameter threshold corresponding to the aspect ratio is set to 20.

[0164] When the aspect ratio of the local model is in the range of 1-20, the grid unit is considered normal; if the aspect ratio is ≥20, it is considered abnormal and the grid quality test result of the local model is determined to be unqualified.

[0165] The mesh quality detection parameter threshold corresponding to the surface mesh angle value is set to 150°.

[0166] When the surface mesh angle corresponding to a local model is ≥150°, the distortion is considered too high and abnormal, and the mesh quality test result of the local model is determined to be unqualified.

[0167] The mesh quality detection parameter threshold corresponding to the warping factor is set to 1.0.

[0168] When the warping factor is ≥1.0, the local model warping is considered abnormal, and the mesh quality test result of the local model is determined to be unqualified.

[0169] Furthermore, the mesh quality detection result of each local model is identified; if the mesh quality detection result is that the mesh is unqualified, abnormal geometric features are identified for the local model, and the identified abnormal geometric features are marked.

[0170] Here, the abnormal geometric feature recognition method of each local model is the same as the abnormal geometric feature recognition method of the model to be detected.

[0171] In one embodiment, after the step of marking the identified abnormal geometric features in step S105 and step S106, the method further includes:

[0172] The key data of the edge corresponding to each abnormal geometric feature is obtained; the key data includes the edge ID and the angle value corresponding to the edge.

[0173] Generate abnormal edge data files based on key data.

[0174] Here, the abnormal edge data file stores the edge ID and angle value corresponding to the associated sharp edge, and the edge ID and angle value corresponding to the associated short edge. The abnormal edge data file is a VTK format file.

[0175] On the pre-set visualization platform, the model to be detected and the abnormal edge data file corresponding to the model to be detected are loaded, and each abnormal geometric feature is highlighted using the preset color coding.

[0176] Here, the visualization platform may be Paraview (an open source industrial simulation platform).

[0177] The user uploads the 3D model of the model to be tested through the visualization platform interface, and the model's geometric structure data such as vertices, edges, faces, etc. are loaded into the visualization platform.

[0178] After loading the 3D model, the user further loads the abnormal edge data file, which integrates the geometric structure of the model with the abnormal edge information.

[0179] On the visualization platform, preset color codes can be used to highlight edges marked as short edges, edges marked as sharp edges, and local models with mesh quality detection results showing unqualified meshes.

[0180] For example, all edges identified as sharp edges may be highlighted in red in the 3D model.

[0181] On the visualization platform, preset color codes can also be used to mark models of different geometric complexity.

[0182] For example, the sharp edges corresponding to the simple model may be marked as red, the sharp edges corresponding to the medium-surface model may be marked as yellow, and the sharp edges corresponding to the complex model may be marked as green.

[0183] Here, the purpose of highlighting is to make sharp edges visually distinguishable from other ordinary edges, thereby providing the maximum visual contrast effect.

[0184] In one embodiment, on a preset visualization platform, after loading the model to be detected and the abnormal edge data file corresponding to the model to be detected, and highlighting each abnormal geometric feature using a preset color coding, the method further includes:

[0185] Identify user interactions on the visualization platform.

[0186] Here, interactive operations can be rotating, scaling the model, or selecting specific sharp edges. Users can rotate and scale the model to view the location and number of sharp edges at different perspectives and scales. This highlighting method allows users to quickly identify potential problem areas (i.e. sharp edges) in the model, thereby helping users make better decisions during model evaluation and optimization.

[0187] If the interactive operation is a preset abnormal edge identification operation, abnormal edge position information corresponding to the interactive operation is obtained, and key data corresponding to the edge corresponding to the abnormal edge position information is displayed.

[0188] Here, hovering the mouse over a certain edge is preset as an abnormal edge identification operation.

[0189] When the user hovers the mouse over an abnormal edge, an interactive prompt box pops up.

[0190] The prompt box displays detailed information about the sharp edge, including but not limited to key data such as edge ID, corresponding angle value, position coordinates, etc. The interactive design of the prompt box provides users with a way to quickly understand the specific characteristics and potential problems of the sharp edge.

[0191] When the user clicks or performs a specific interactive operation (such as selecting an abnormal edge identification operation), the abnormal edge location information corresponding to the current operation is obtained, and the key data related to the abnormal edge, such as the edge ID and angle value, are displayed in detail. This interactive feedback mechanism not only makes the model detection results more intuitive, but also provides effective data support for users in subsequent model modification and optimization.

[0192] In one embodiment, a bicycle frame model is used as a test case. The user uploads the bicycle frame model in the UI (User Interface) of the numerical computing APP. Because the model quality detection method in this embodiment is integrated in the numerical computing APP, the user only needs to click the "Upload Model" button and select the local model file.

[0193] The model scene type of the model to be tested is obtained as a frame product, and the grid size corresponding to the bicycle frame model is determined to be 10 mm.

[0194] The model to be detected is divided into multiple local models based on the grid size, and the division failure of the local model is determined.

[0195] Based on the topological analysis of geometric elements, the bicycle frame model is determined to be a simple model. Fig. 9 , and determine that the angle threshold corresponding to the bicycle frame model is 30°.

[0196] After identifying the sharp edges of the bicycle frame model, refer to Fig.10 , all sharp edges with angles less than or equal to 30 degrees in the bicycle frame model are displayed on the visualization platform. The sharp edges are displayed in red, and users can see the distribution of sharp edges at a glance. The visualization platform also provides sharp edge statistics, showing the total number of sharp edges detected and their distribution in various parts of the model. In addition, users can use the interactive tools on the visualization platform (such as the magnifying glass button) to zoom in on any sharp edge and check the key data of the sharp edge in detail (such as position, angle value, and adjacent surface normal vector).

[0197] The embodiment of the present invention provides a model quality detection method, including: obtaining a model to be detected sent by a user, geometric feature data corresponding to the model to be detected, and a model scene type; the geometric feature data is used to characterize the correspondence between vertices, edges, and faces in the model to be detected; based on the model scene type and the preset model scene type grid size correspondence, determine the grid size corresponding to the model to be detected; based on the grid size, divide the model to be detected into multiple local models; judge whether the division of the local model is successful; if the division of the local model fails, identify abnormal geometric features of the model to be detected, and mark the identified abnormal geometric features; if the division of the local model is successful, perform grid quality detection on each local model; identify the grid quality detection result of each local model; if the grid quality detection result is that the grid is unqualified, identify abnormal geometric features of the local model, and mark the identified abnormal geometric features. In this method, by meshing the model to be detected, the abnormal edges that are prone to stress concentration and grid division difficulties are accurately identified, and then effective model quality detection is performed before numerical analysis, and potential problems in the model are identified and processed to improve the reliability and efficiency of numerical analysis. At the same time, the model quality detection method is integrated into the numerical calculation APP, allowing users to complete the entire process of model uploading, detection, optimization and numerical analysis on the cloud web page, thereby effectively reducing the operational threshold and complexity of numerical analysis.

[0198] Embodiment 2:

[0199] Fig.11 This is a schematic diagram of a model quality detection system provided in Embodiment 2 of the present invention.

[0200] Reference Fig.11 , the model quality detection system includes:

[0201] The geometric feature data determination module 1 is used to obtain the model to be detected sent by the user, the geometric feature data corresponding to the model to be detected and the model scene type; the geometric feature data is used to characterize the correspondence between vertices, edges and faces in the model to be detected.

[0202] The grid division module 2 is used to determine the grid size corresponding to the model to be detected based on the corresponding relationship between the model scene type and the preset model scene type grid size.

[0203] The grid division module 2 is also used to divide the model to be detected into multiple local models based on the grid size.

[0204] Abnormal geometric feature marking module 3, used to determine whether the division of the local model is successful;

[0205] The abnormal geometric feature marking module 3 is also used to identify abnormal geometric features of the model to be detected and mark the identified abnormal geometric features if the division of the local model fails;

[0206] The abnormal geometric feature marking module 3 is also used to perform mesh quality detection on each local model if the division of the local model is successful; identify the mesh quality detection result of each local model; if the mesh quality detection result is that the mesh is unqualified, identify abnormal geometric features of the local model and mark the identified abnormal geometric features.

[0207] Furthermore, the geometric feature data determination module 1 is also used for:

[0208] Extract geometric elements corresponding to the model to be detected; geometric elements include vertex data, edge data and face data.

[0209] Perform topological analysis on geometric elements, identify each edge in the model to be detected and the adjacent faces connected to each edge, and establish a mapping relationship between edges and adjacent faces; wherein the mapping relationship between edges and adjacent faces is used to characterize the association relationship between the edge ID corresponding to the edge and the face ID corresponding to the adjacent face.

[0210] The mapping relationship between geometric elements and edges and adjacent faces is combined to form geometric feature data corresponding to the model to be detected.

[0211] Furthermore, the abnormal geometric feature recognition includes sharp edge recognition; the abnormal geometric feature includes a sharp edge. The abnormal geometric feature marking module 3 is also used for:

[0212] Based on the geometric feature data, the angle value between adjacent faces connected to each edge in the model to be detected is determined.

[0213] Each angle value is compared with the angle threshold, and the edges corresponding to the angle values ​​less than the angle threshold are marked as sharp edges.

[0214] Furthermore, the abnormal geometric feature marking module 3 is also used for:

[0215] Perform topological analysis on geometric elements to determine the face complexity, edge length difference, vertex density, curvature change normalization value and topological structure complexity corresponding to the model to be tested.

[0216] Based on the patch complexity, edge length difference, vertex density, curvature change normalization value, topological structure complexity and preset complexity weight, the geometric complexity score corresponding to the model to be tested is calculated.

[0217] Based on the geometric complexity score and the preset mapping relationship between the geometric complexity score and the angle threshold, the angle threshold corresponding to the model to be detected is determined.

[0218] Furthermore, the abnormal geometric feature marking module 3 is also used for:

[0219] Identify the geometric complexity score.

[0220] If the geometric complexity score is less than a preset first threshold, the model to be detected is determined to be a simple model, and the angle threshold is determined to be the first angle threshold.

[0221] If the geometric complexity score is greater than or equal to the preset first threshold and less than the preset second threshold, the model to be detected is determined to be a medium model, and the angle threshold is determined to be the second angle threshold.

[0222] If the geometric complexity score is greater than or equal to the preset first threshold, the model to be detected is determined to be a complex model, and the angle threshold is determined to be the third angle threshold; wherein the preset first threshold is less than the preset second threshold; the first angle threshold is greater than the second angle threshold, which is greater than the third angle threshold.

[0223] Furthermore, the abnormal geometric feature recognition includes side length recognition; the abnormal geometric feature includes a short side. The abnormal geometric feature marking module 3 is also used for:

[0224] Based on the geometric feature data, the length of each edge in the model to be detected is determined.

[0225] Compare each edge length with the grid size, and mark the edges whose length is smaller than the grid size as short edges.

[0226] Furthermore, the abnormal geometric feature marking module 3 is also used for:

[0227] Based on the geometric feature data, the number of faces in each partial model is determined.

[0228] Based on the geometric feature data and the number of faces, a local model type of the local model is determined; the local model type is a tetrahedral mesh or a hexahedral mesh.

[0229] Based on the local model type and the preset mesh quality detection method, the mesh quality detection parameters corresponding to each local model are calculated; the mesh quality detection parameters include aspect ratio and / or distortion and / or warping factor.

[0230] Determine whether each mesh quality detection parameter is within a preset mesh quality detection parameter threshold.

[0231] If the mesh quality detection parameter is within a preset mesh quality detection parameter threshold, it is determined that the mesh quality detection result of the local model corresponding to the mesh quality detection parameter is a qualified mesh.

[0232] If the mesh quality detection parameter is not within the preset mesh quality detection parameter threshold, it is determined that the mesh quality detection result of the local model corresponding to the mesh quality detection parameter is a mesh failure.

[0233] In one embodiment, referring to Fig.12 The model quality detection system further includes: a model display module 4, the model display module 4 is used to:

[0234] The key data of the edge corresponding to each abnormal geometric feature is obtained; the key data includes the edge ID and the angle value corresponding to the edge.

[0235] Generate abnormal edge data files based on key data.

[0236] On the pre-set visualization platform, the model to be detected and the abnormal edge data file corresponding to the model to be detected are loaded, and each abnormal geometric feature is highlighted using the preset color coding.

[0237] Furthermore, the model display module 4 is also used for:

[0238] Identify user interactions on the visualization platform.

[0239] If the interactive operation is a preset abnormal edge identification operation, abnormal edge position information corresponding to the interactive operation is obtained, and key data corresponding to the edge corresponding to the abnormal edge position information is displayed.

[0240] The embodiment of the present invention provides a model quality detection system that can accurately identify abnormal edges in the model that are likely to cause stress concentration and finite element model quality problems, thereby reducing the complexity and risk of misjudgment of manual inspection, and further improving the efficiency of model detection and the reliability of numerical analysis.

[0241] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the system and device described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0242] In addition, in the description of the embodiments of the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0243] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

[0244] Finally, it should be noted that the above-described embodiments are only specific implementations of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The protection scope of the present invention is not limited thereto. Although the present invention is described in detail with reference to the above-described embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-described embodiments within the technical scope disclosed by the present invention, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A model quality detection method, characterized in that: include: Obtaining the model to be detected sent by the user, the geometric feature data corresponding to the model to be detected, and the model scene type; The geometric feature data is used to characterize the corresponding relationship between vertices, edges and faces in the model to be detected; Determining the grid size corresponding to the model to be detected based on the corresponding relationship between the model scene type and the preset model scene type grid size; Based on the grid size, dividing the model to be detected into a plurality of local models; Determining whether the division of the local model is successful; If the division of the local model fails, abnormal geometric features are identified on the model to be detected, and the identified abnormal geometric features are marked; If the division of the local model is successful, perform a mesh quality test on each of the local models; identify the mesh quality test result of each of the local models; if the mesh quality test result is that the mesh is unqualified, perform the abnormal geometric feature identification on the local model, and mark the identified abnormal geometric feature; The abnormal geometric feature identification includes sharp edge identification; the abnormal geometric feature includes a sharp edge; The step of identifying abnormal geometric features of the model to be detected and marking the identified abnormal geometric features includes: Based on the geometric feature data, determine the angle value between adjacent faces connected to each of the edges in the model to be detected; Compare each of the angle values ​​with an angle threshold, and mark the edge corresponding to the angle value smaller than the angle threshold as a sharp edge; Before the step of comparing each of the angle values ​​with the angle threshold, the method further includes: Performing topological analysis on geometric elements to determine the face complexity, edge length difference, vertex density, curvature change normalization value and topological structure complexity corresponding to the model to be detected; the geometric elements include vertex data, edge data and face data; Based on the face complexity, the edge length difference, the vertex density, the curvature change normalization value, the topological structure complexity and the preset complexity weight, a geometric complexity score corresponding to the model to be detected is calculated; Determining the angle threshold corresponding to the model to be detected based on the geometric complexity score and a preset mapping relationship between the geometric complexity score and the angle threshold; The step of determining the angle threshold corresponding to the model to be detected based on the mapping relationship between the geometric complexity score and the preset geometric complexity score and the angle threshold comprises: identifying the geometric complexity score; If the geometric complexity score is less than a preset first threshold, the model to be detected is determined to be a simple model, and the angle threshold is determined to be a first angle threshold; If the geometric complexity score is greater than or equal to the preset first threshold and less than the preset second threshold, the model to be detected is determined to be a medium model, and the angle threshold is determined to be a second angle threshold; If the geometric complexity score is greater than or equal to a preset first threshold, the model to be detected is determined to be a complex model, and the angle threshold is determined to be a third angle threshold; wherein the preset first threshold is less than the preset second threshold; the first angle threshold is greater than the second angle threshold which is greater than the third angle threshold.

2. The model quality detection method according to claim 1, characterized in that: The step of obtaining geometric feature data corresponding to the model to be detected includes: Extracting the geometric elements corresponding to the model to be detected; Performing topological analysis on the geometric elements, identifying each edge and the adjacent faces connected to each edge in the model to be detected, and establishing a mapping relationship between the edge and the adjacent faces; wherein the mapping relationship between the edge and the adjacent faces is used to characterize the association relationship between the edge ID corresponding to the edge and the face ID corresponding to the adjacent face; The geometric elements and the mapping relationship between the edges and adjacent faces are combined to form the geometric feature data corresponding to the model to be detected.

3. The model quality detection method according to claim 1, characterized in that: The abnormal geometric feature identification includes side length identification; the abnormal geometric feature includes a short side; The step of identifying abnormal geometric features of the model to be detected and marking the identified abnormal geometric features includes: Based on the geometric feature data, determining the length of each edge in the model to be detected; Each side length is compared with the grid size, and the side corresponding to the side length smaller than the grid size is marked as a short side.

4. The model quality detection method according to claim 1, characterized in that: The step of performing mesh quality detection on each of the local models comprises: Based on the geometric feature data, determining the number of faces in each of the local models; Determining a local model type of the local model based on the geometric feature data and the number of faces; the local model type is a tetrahedral mesh or a hexahedral mesh; Based on the local model type and the preset grid quality detection method, a grid quality detection parameter corresponding to each of the local models is calculated; the grid quality detection parameter includes an aspect ratio and / or a distortion degree and / or a warping factor; Determining whether each of the grid quality detection parameters is within a preset grid quality detection parameter threshold; If the mesh quality detection parameter is within the preset mesh quality detection parameter threshold, determining that the mesh quality detection result of the local model corresponding to the mesh quality detection parameter is a qualified mesh; If the mesh quality detection parameter is not within the preset mesh quality detection parameter threshold, it is determined that the mesh quality detection result of the local model corresponding to the mesh quality detection parameter is a mesh failure.

5. The model quality detection method according to claim 1, characterized in that: After the step of marking the identified abnormal geometric features, the method further comprises: Acquire key data of the edge corresponding to each of the abnormal geometric features; the key data includes the edge ID and the angle value corresponding to the edge; Generate an abnormal edge data file based on the key data; On a preset visualization platform, the model to be detected and the abnormal edge data file corresponding to the model to be detected are loaded, and each abnormal geometric feature is highlighted using a preset color coding.

6. The model quality detection method according to claim 5, characterized in that: After the step of loading the model to be detected and the abnormal edge data file corresponding to the model to be detected on the preset visualization platform, and highlighting each abnormal geometric feature using a preset color coding, the method further includes: Identifying the interactive operation of the user on the visualization platform; If the interactive operation is a preset abnormal edge identification operation, abnormal edge position information corresponding to the interactive operation is obtained, and the key data corresponding to the edge corresponding to the abnormal edge position information is displayed.

7. A model quality detection system, characterized in that: include: A geometric feature data determination module is used to obtain the model to be detected sent by the user, the geometric feature data corresponding to the model to be detected, and the model scene type; The geometric feature data is used to characterize the corresponding relationship between vertices, edges and faces in the model to be detected; A grid division module, used to determine the grid size corresponding to the model to be detected based on the corresponding relationship between the model scene type and the preset model scene type grid size; The grid division module is further used to divide the model to be detected into multiple local models based on the grid size; An abnormal geometric feature marking module is used to determine whether the division of the local model is successful; The abnormal geometric feature marking module is further used to identify abnormal geometric features of the model to be detected and mark the identified abnormal geometric features if the division of the local model fails; The abnormal geometric feature marking module is further used to perform mesh quality detection on each of the local models if the division of the local models is successful; identify the mesh quality detection result of each of the local models; if the mesh quality detection result is that the mesh is unqualified, perform the abnormal geometric feature identification on the local model, and mark the identified abnormal geometric feature; The abnormal geometric feature identification includes sharp edge identification; the abnormal geometric feature includes a sharp edge; The abnormal geometric feature marking module is further used to determine the angle value between adjacent faces connected to each edge in the model to be detected based on the geometric feature data; compare each angle value with an angle threshold, and mark the edge corresponding to the angle value less than the angle threshold as a sharp edge; The abnormal geometric feature marking module is also used to perform topological analysis on geometric elements to determine the face complexity, edge length difference, vertex density, curvature change normalization value and topological structure complexity corresponding to the model to be detected; based on the face complexity, edge length difference, vertex density, curvature change normalization value, topological structure complexity and preset complexity weight, calculate the geometric complexity score corresponding to the model to be detected; Based on the geometric complexity score and the mapping relationship between the preset geometric complexity score and the angle threshold, determining the angle threshold corresponding to the model to be detected; the geometric elements include vertex data, edge data and face data; The abnormal geometric feature marking module is also used to identify the geometric complexity score; If the geometric complexity score is less than the preset first threshold, the model to be detected is determined to be a simple model, and the angle threshold is determined to be the first angle threshold; if the geometric complexity score is greater than or equal to the preset first threshold and less than the preset second threshold, the model to be detected is determined to be a medium model, and the angle threshold is determined to be the second angle threshold; if the geometric complexity score is greater than or equal to the preset first threshold, the model to be detected is determined to be a complex model, and the angle threshold is determined to be the third angle threshold; wherein, the preset first threshold is less than the preset second threshold; the first angle threshold is greater than the second angle threshold, which is greater than the third angle threshold.

Images