Geometric model mesh generation method and device based on machine learning, medium and equipment
Through the geometric model meshing method based on machine learning, the problem of insufficient efficiency and adaptability of the existing technology when dealing with complex models is solved, and efficient and accurate meshing is achieved, which is suitable for simulation and numerical calculation of complex models.
Patent Information
- Application Number
- CN202411504042.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-25
- Publication Date
- 2025-03-07
AI Technical Summary
The prior art has limitations in handling complex models, improving division efficiency and enhancing adaptability, making it difficult to generate high-quality grids and avoid errors and distortions.
Using a machine learning-based geometric model meshing method, pre-processing geometric models, using machine learning models for meshing, and performing quality inspection and optimization to generate high-quality mesh that is highly consistent with the actual model.
It improves grid division efficiency, reduces labor costs, generates high-quality grids, ensures the accuracy and credibility of division results, and is suitable for simulation and numerical calculations of complex models.
Smart Images
Figure CN119578206A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a method, device, medium and equipment for meshing a geometric model based on machine learning, and belongs to the technical field of computer program artificial intelligence. Background Art
[0002] In the field of meshing, a variety of technologies have been proposed and applied. These technologies cover the following aspects: (1) Rule-based meshing technology: It mainly relies on preset rules and algorithms to divide the mesh through mathematical calculations. It is suitable for meshing simple models, but its efficiency and accuracy are limited when dealing with complex and changing solution domains. (2) Parameterized meshing technology: It controls the meshing by adjusting parameters, which requires professionals to make tedious adjustments to the parameters, and different parameter settings may be required for solution domains of different shapes and sizes. (3) Adaptive meshing technology: It attempts to explore adaptive meshing, but lacks sufficient intelligence and flexibility, making it difficult to cope with various complex situations, and there are still challenges in mesh quality control. In summary, although the existing technology can solve some problems to a certain extent, it still has obvious limitations in dealing with complex models, improving meshing efficiency, and enhancing adaptive capabilities. Summary of the invention
[0003] In view of this, the present invention provides a geometric model mesh division method, device, medium and equipment based on machine learning, which improves the division efficiency and reduces labor costs; automatically adjusts the mesh division strategy to generate high-quality meshes; ensures that the mesh division results are highly consistent with the actual model, avoids errors and distortions that may occur in traditional methods, provides a reliable foundation for subsequent finite element analysis or other numerical calculations, and improves the accuracy and credibility of simulation results, making them more suitable for practical use.
[0004] In order to achieve the above first purpose, the technical solution of the geometric model meshing method based on machine learning provided by the present invention is as follows:
[0005] The geometric model meshing method based on machine learning provided by the present invention comprises the following steps:
[0006] Obtain the geometric model to be meshed;
[0007] Preprocessing the geometric model of the to-be-meshed grid to obtain a preprocessed geometric model of the to-be-meshed grid;
[0008] The preprocessed geometric model to be meshed is input into a mesh partitioning prediction model obtained through machine learning. The mesh partitioning prediction model obtained through machine learning performs mesh partitioning on the geometric model to be meshed to obtain a mesh partitioning result of the geometric model.
[0009] The geometric model meshing method based on machine learning provided by the present invention can also be further implemented by adopting the following technical measures.
[0010] Preferably, the preprocessed geometric model to be meshed is input into a mesh partitioning prediction model obtained through machine learning, and the mesh partitioning prediction model obtained through machine learning performs mesh partitioning on the geometric model to be meshed, and after the step of obtaining the mesh partitioning result of the geometric model, the following steps are also included:
[0011] Performing a quality inspection and / or a simulation calculation verification on the meshing result of the geometric model to obtain a quality inspection and / or a simulation calculation verification result of the meshing result of the geometric model;
[0012] According to the quality inspection and / or simulation calculation verification result of the mesh division result of the geometric model, the mesh division result of the geometric model is optimized to obtain the optimized mesh division result of the geometric model.
[0013] Preferably, the method for constructing the grid partition prediction model obtained through machine learning comprises the following steps:
[0014] Acquire a geometric model training sample, wherein a mesh division result of the geometric model training sample is known;
[0015] Inputting the geometric model training samples into a set machine learning model so that the set machine learning model obtains a mesh division result corresponding to the geometric model training samples one by one;
[0016] The parameters of the set machine learning model are continuously adjusted so that the mesh division results of the geometric model obtained through the set machine learning model correspond one-to-one with the training samples of the geometric model, thereby obtaining a mesh division prediction model obtained through machine learning.
[0017] Preferably, the geometric model meshing method based on machine learning further comprises the following steps:
[0018] Each geometric model of the grid to be divided and the grid division result corresponding to the geometric model of the grid to be divided are added to the training samples.
[0019] Preferably, in the step of preprocessing the geometric model to be meshed to obtain the preprocessed geometric model to be meshed, the preprocessing includes at least one of feature extraction, data cleaning and standardization, wherein:
[0020] The features extracted by the feature extraction process include at least any one of curvature, boundary conditions, and stress concentration areas;
[0021] The data cleaning and standardization processing includes cleaning and standardizing the parameters of the features extracted by the feature extraction process, so that the noise and abnormal values of the features extracted by the feature extraction process are eliminated, thereby ensuring the consistency and comparability of the data.
[0022] Preferably, the algorithm of the grid partition prediction model obtained through machine learning is selected from any one of a convolutional neural network algorithm, a random forest or a support vector machine algorithm.
[0023] Preferably, according to the quality inspection and / or simulation calculation verification result of the mesh division result of the geometric model, the mesh division result of the geometric model is optimized, and obtaining the optimized mesh division result of the geometric model specifically comprises the following steps:
[0024] Predict the optimal meshing strategy, where a mesh uniformity metric might be defined as the variance of all element sizes in the mesh:
[0025]
[0026] Among them, S i is the size of the i-th grid cell, SS is the average of all grid cell sizes, and N is the total number of grid cells;
[0027] Dynamically adjust the grid density, where the grid density adjustment algorithm is:
[0028] NewDensity=OldDensity×(1-α×Gradient Density )
[0029] Among them, NewDensity is the adjusted grid density, OldDensity is the current grid density, α is the learning rate, and Gradient Density is the gradient of the optimization objective function with respect to the grid density.
[0030] Select a mesh type that is suitable for the geometric model to be meshed.
[0031] In order to achieve the above second purpose, the technical solution of the geometric model meshing device based on machine learning provided by the present invention is as follows:
[0032] The geometric model meshing device based on machine learning provided by the present invention comprises:
[0033] A geometric model acquisition module to be divided is used to obtain a geometric model of a grid to be divided;
[0034] A data preprocessing module, used for preprocessing the geometric model of the to-be-meshed grid to obtain a preprocessed geometric model of the to-be-meshed grid;
[0035] The mesh division prediction model obtained through machine learning is used to input the preprocessed geometric model of the mesh to be divided into it, and it meshes the geometric model of the mesh to be divided to obtain the mesh division result of the geometric model.
[0036] In order to achieve the third objective above, the technical solution of the computer-readable storage medium provided by the present invention is as follows:
[0037] The computer-readable storage medium provided by the present invention stores a geometric model mesh division program based on machine learning. When the geometric model mesh division program based on machine learning is executed by a processor, the steps of the geometric model mesh division method based on machine learning provided by the present invention are implemented.
[0038] In order to achieve the fourth objective, the technical solution of the electronic device provided by the present invention is as follows:
[0039] The electronic device provided by the present invention includes a memory and a processor, wherein the memory stores a geometric model mesh division program based on machine learning, and when the geometric model mesh division program based on machine learning is executed by the processor, the steps of the geometric model mesh division method based on machine learning provided by the present invention are implemented.
[0040] The present invention provides a geometric model meshing method, device, medium and equipment based on machine learning. First, the method has excellent efficiency. Traditional meshing is often time-consuming and labor-intensive, requiring a lot of manual intervention, while the machine learning method can automatically perform meshing, greatly improving the efficiency of meshing and reducing labor costs. Secondly, the method has excellent adaptability. It can automatically adjust the meshing strategy according to the geometric shape and complexity of the model to generate high-quality meshes. This adaptability enables the method to cope with various complex models without the need for tedious manual adjustments and optimizations. Finally, the method also shows extremely high accuracy. Through the optimization of the machine learning algorithm, it can ensure that the meshing result is highly consistent with the actual model, avoiding the errors and distortions that may occur in the traditional method, providing a reliable basis for subsequent finite element analysis or other numerical calculations, and improving the accuracy and credibility of the simulation results. In summary, an efficient meshing method based on machine learning shows significant advantages in terms of efficiency, adaptability and accuracy. These advantages make the method have broad application potential and market prospects in the field of meshing, and inject new vitality into the development of engineering simulation and numerical calculation. BRIEF DESCRIPTION OF THE DRAWINGS
[0041] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:
[0042] Figure 1 A flowchart of the steps of a method for meshing a geometric model based on machine learning provided in an embodiment of the present invention;
[0043] Figure 2 A schematic diagram of the signal flow relationship between various functional modules in a geometric model meshing device based on machine learning provided in an embodiment of the present invention;
[0044] Figure 3 A schematic diagram of the structure of a geometric model meshing device based on machine learning in the hardware operating environment provided by an embodiment of the present invention. DETAILED DESCRIPTION
[0045] In order to solve the problems existing in the prior art, the present invention provides a geometric model mesh division method, device, medium and equipment based on machine learning, which improves the division efficiency and reduces labor costs; automatically adjusts the mesh division strategy to generate high-quality meshes; ensures that the mesh division results are highly consistent with the actual model, avoids errors and distortions that may occur in traditional methods, provides a reliable basis for subsequent finite element analysis or other numerical calculations, and improves the accuracy and credibility of simulation results, making them more suitable for practical use.
[0046] Glossary
[0047] Machine learning: is a branch of artificial intelligence that enables computers to automatically learn and extract features from data without explicit programming, thereby improving and optimizing meshing performance.
[0048] Meshing: It is a process of dividing a continuous geometric area or object into a finite number of discrete units (i.e., meshes). Through meshing, continuous physical fields can be discretized so that numerical methods can be applied to each mesh unit to solve mathematical problems such as partial differential equations. The quality and efficiency of meshing directly affect the accuracy and computational cost of subsequent numerical calculations.
[0049] Numerical simulation: is a technology that uses mathematical models and calculation methods to simulate and analyze actual systems or processes on computers. It discretizes continuous physical or engineering problems and uses numerical methods to solve partial differential equations or integral equations to obtain approximate solutions. Numerical simulation is widely used in various fields, such as engineering, physics, chemistry, biology, etc., and provides a powerful tool for scientific research, engineering design and optimization.
[0050] Stress concentration: It is a phenomenon in which the stress level in some areas or points of an object or structure is significantly higher than that in other areas or points due to the uneven distribution of geometry, material properties or external loads.
[0051] Boundary conditions: refers to the constraints set to describe the behavior of physical systems or engineering problems on specific boundaries when solving partial differential equations or performing numerical simulations.
[0052] Mesh density: refers to the number of meshes per unit area or volume, and is an indicator of the density of the mesh. The higher the mesh density, the higher the calculation accuracy is usually, because the changes in physical phenomena can be captured more accurately.
[0053] Mesh quality: It is an indicator to measure the quality of mesh geometry and topology. It is a comprehensive indicator, mainly including mesh shape quality, mesh size uniformity, mesh topology, mesh smoothness, mesh adaptability, etc. Therefore, optimizing mesh quality is one of the key goals to improve the efficiency and accuracy of numerical simulation.
[0054] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the geometric model meshing method, device, medium and equipment based on machine learning proposed by the present invention, its specific implementation method, structure, characteristics and effects, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "embodiment" does not necessarily refer to the same embodiment. In addition, the features, structures, or characteristics in one or more embodiments may be combined in any suitable form.
[0055] The term "and / or" in this article is merely a description of the association relationship of associated objects, indicating that three relationships may exist, for example, A and / or B. Specifically, it is understood that: A and B may be included at the same time, A may exist alone, or B may exist alone, and any of the above three situations may be met.
[0056] A geometric model meshing method based on machine learning
[0057] See attached Figure 1 The geometric model meshing method based on machine learning provided by the embodiment of the present invention comprises the following steps:
[0058] Step S1: Obtain a geometric model to be meshed;
[0059] Step S2: preprocessing the geometric model to be meshed to obtain a preprocessed geometric model to be meshed.
[0060] Specifically, data preprocessing includes:
[0061] (1) Collect training data sets. Collect meshing cases with different geometric shapes and boundary conditions as training data sets. These cases should include geometric shapes of various complexities and meshing results under different boundary conditions.
[0062] (2) Feature extraction. Feature extraction is performed on the geometric model, including key parameters such as curvature, boundary conditions, and stress concentration areas. These feature parameters will be used as input to the machine learning model to predict the optimal meshing strategy.
[0063] (3) Data cleaning and standardization: The extracted feature parameters are cleaned and standardized to eliminate noise and outliers and ensure the consistency and comparability of the data.
[0064] Step S3: input the preprocessed geometric model to be meshed into the meshing prediction model obtained through machine learning, and mesh the geometric model to be meshed using the meshing prediction model obtained through machine learning to obtain a meshing result of the geometric model.
[0065] Specifically,
[0066] The machine learning model building method includes the following steps:
[0067] (1) Select a machine learning algorithm. Select a suitable machine learning algorithm based on the characteristics of the problem and the characteristics of the data. In the present invention, algorithms such as convolutional neural network (CNN), random forest or support vector machine (SVM) can be considered.
[0068] (2) Construct a mesh partitioning prediction model. Use the selected machine learning algorithm to construct a mesh partitioning prediction model. The model should be able to learn the mapping relationship between geometric features and mesh partitioning strategies and output the optimal mesh partitioning strategy.
[0069] (3) Model training and validation. Use the training data set to train the machine learning model and evaluate the model performance through methods such as cross-validation. During the training process, the model parameters and training strategies should be continuously adjusted to improve the model's prediction accuracy and generalization ability.
[0070] The automatic meshing method includes the following steps:
[0071] (1) Calling meshing software. According to the prediction results of the machine learning model and the meshing strategy, the meshing software is automatically called to perform meshing. Commonly used meshing software includes ANSYS Mesh, Gmsh, etc.
[0072] (2) Real-time monitoring and adjustment. During the meshing process, the meshing progress and mesh quality are monitored in real time, and dynamic adjustments are made as needed. For example, when it is found that the mesh quality does not meet the requirements, the mesh density or type can be automatically adjusted to ensure that the mesh quality meets the calculation requirements.
[0073] (3) Output meshing results. When meshing is completed, output meshing results, including mesh files, mesh quality reports, etc. These results will be used for subsequent numerical simulation calculations.
[0074] The present invention provides a geometric model meshing method based on machine learning. First, the method has excellent efficiency. Traditional meshing is often time-consuming and labor-intensive, requiring a lot of manual intervention, while the machine learning method can automatically perform meshing, greatly improving the efficiency of meshing and reducing labor costs. Secondly, the method has excellent adaptability. It can automatically adjust the meshing strategy according to the geometric shape and complexity of the model to generate high-quality meshes. This adaptability enables the method to cope with various complex models without the need for tedious manual adjustments and optimizations. Finally, the method also shows extremely high accuracy. Through the optimization of the machine learning algorithm, it can ensure that the meshing result is highly consistent with the actual model, avoiding the errors and distortions that may occur in the traditional method, providing a reliable basis for subsequent finite element analysis or other numerical calculations, and improving the accuracy and credibility of the simulation results. In summary, an efficient meshing method based on machine learning shows significant advantages in terms of efficiency, adaptability and accuracy. These advantages make the method have broad application potential and market prospects in the field of meshing, and inject new vitality into the development of engineering simulation and numerical calculation.
[0075] Among them, the pre-processed geometric model to be meshed is input into the meshing prediction model obtained by machine learning, and the meshing prediction model of machine learning meshes the geometric model to be meshed, and after the step of obtaining the meshing result of the geometric model, the following steps are also included:
[0076] Performing quality inspection and / or simulation calculation verification on the meshing result of the geometric model to obtain quality inspection and / or simulation calculation verification results of the meshing result of the geometric model;
[0077] According to the quality inspection and / or simulation calculation verification results of the meshing results of the geometric model, the meshing results of the geometric model are optimized to obtain the optimized meshing results of the geometric model.
[0078] Specifically, data post-processing and verification include the following steps:
[0079] (1) Mesh quality check. Check the quality of the divided mesh, including mesh shape, mesh size, mesh connectivity, etc. Ensure that the mesh quality meets the calculation requirements to improve the accuracy and stability of numerical simulation.
[0080] (2) Simulation verification. Import the divided mesh into numerical simulation software for calculation to verify the effectiveness of the mesh division. By comparing the simulation results with experimental data or theoretical solutions, the accuracy and calculation efficiency of the mesh division can be evaluated.
[0081] (3) Feedback and optimization. Feedback is used to adjust the parameters and training strategies of the machine learning model based on the simulation results and actual application requirements. Through continuous iterative optimization, the prediction accuracy and generalization ability of the machine learning model are improved, and the efficiency and quality of meshing are further improved.
[0082] The method for constructing the grid partition prediction model obtained through machine learning includes the following steps:
[0083] Acquire a geometric model training sample, wherein a mesh division result of the geometric model training sample is known;
[0084] Inputting the geometric model training samples into the set machine learning model, so that the set machine learning model obtains a meshing result corresponding to the geometric model training samples one by one;
[0085] The parameters of the set machine learning model are continuously adjusted so that the mesh division results of the geometric model obtained through the set machine learning model correspond one-to-one with the geometric model training samples, thereby obtaining a mesh division prediction model obtained through machine learning.
[0086] In this case, the mesh division prediction model obtained through machine learning provided by the embodiment of the present invention can be continuously iterated and optimized by adding geometric models, thereby improving the prediction accuracy and generalization ability of the machine learning model, and further improving the efficiency and quality of mesh division.
[0087] The geometric model meshing method based on machine learning also includes the following steps:
[0088] Each geometric model to be meshed and the meshing result corresponding to the geometric model to be meshed are added to the training samples.
[0089] In the step of preprocessing the geometric model to be meshed to obtain the preprocessed geometric model to be meshed, the preprocessing includes at least one of feature extraction, data cleaning and standardization, wherein:
[0090] The features extracted by the feature extraction process include at least any one of curvature, boundary conditions, and stress concentration areas;
[0091] Data cleaning and standardization processing includes cleaning and standardizing the parameters of the features extracted by the feature extraction process, so that the noise and outliers of the features extracted by the feature extraction process are eliminated, ensuring the consistency and comparability of the data.
[0092] The algorithm of the grid partition prediction model obtained through machine learning is selected from any one of a convolutional neural network algorithm, a random forest algorithm or a support vector machine algorithm.
[0093] According to the quality inspection and / or simulation calculation verification results of the meshing results of the geometric model, the meshing results of the geometric model are optimized to obtain the optimized meshing results of the geometric model, which specifically includes the following steps:
[0094] (1) Predict the optimal meshing strategy. For a new geometric model, input its characteristic parameters into the machine learning model to predict the optimal meshing strategy. The prediction content includes mesh type, mesh density, cell order, etc. These indicators can be calculated by specific formulas. For example, the mesh uniformity indicator may be defined as the variance of all cell sizes in the mesh:
[0095]
[0096] Among them, S i is the size of the i-th grid cell, SS is the average of all grid cell sizes, and N is the total number of grid cells.
[0097] (2) Dynamically adjust the mesh density. According to the prediction results of the machine learning model, the mesh density is dynamically adjusted. A denser mesh is used in areas with stress concentration or drastic flow field changes to improve the calculation accuracy, and a sparser mesh is used in relatively flat areas or areas with small changes to reduce the calculation cost. The adjustment can be achieved through a specific formula, such as adjusting the mesh density according to the mesh quality index;
[0098] NewDensity=OldDensity×(1-α×Gradient Density )
[0099] Among them, NewDensity is the adjusted grid density, OldDensity is the current grid density, α is the learning rate, and Gradient Densityis the gradient of the optimization objective function with respect to the grid density.
[0100] (3) Select the appropriate mesh type. Select the appropriate mesh type based on the geometry and computational requirements. For example, for complex geometries, consider using tetrahedrons or unstructured meshes, and for simple geometries, consider using hexahedrons or structured meshes.
[0101] A geometric model meshing device based on machine learning
[0102] See attached Figure 2 , the geometric model meshing device based on machine learning provided by the embodiment of the present invention includes:
[0103] A geometric model acquisition module to be divided is used to obtain a geometric model of a grid to be divided;
[0104] A data preprocessing module is used to preprocess the geometric model to be meshed to obtain a preprocessed geometric model to be meshed;
[0105] The meshing prediction model obtained through machine learning is used to input the preprocessed geometric model to be meshed into it, which meshes the geometric model to be meshed to obtain the meshing result of the geometric model.
[0106] The present invention provides a geometric model meshing device based on machine learning. First, the method has excellent efficiency. Traditional meshing is often time-consuming and labor-intensive, requiring a lot of manual intervention, while the machine learning method can automatically perform meshing, greatly improving the efficiency of meshing and reducing labor costs. Secondly, the method has excellent adaptability. It can automatically adjust the meshing strategy according to the geometric shape and complexity of the model to generate high-quality meshes. This adaptability enables the method to cope with various complex models without the need for tedious manual adjustments and optimizations. Finally, the method also shows extremely high accuracy. Through the optimization of the machine learning algorithm, it can ensure that the meshing result is highly consistent with the actual model, avoiding the errors and distortions that may occur in the traditional method, providing a reliable basis for subsequent finite element analysis or other numerical calculations, and improving the accuracy and credibility of the simulation results. In summary, an efficient meshing method based on machine learning shows significant advantages in terms of efficiency, adaptability and accuracy. These advantages make the method have broad application potential and market prospects in the field of meshing, and inject new vitality into the development of engineering simulation and numerical calculation.
[0107] Computer readable storage medium
[0108] The computer-readable storage medium provided in an embodiment of the present invention stores a geometric model mesh division program based on machine learning. When the geometric model mesh division program based on machine learning is executed by a processor, the steps of the geometric model mesh division method based on machine learning provided in the present invention are implemented.
[0109] The present invention provides a computer-readable storage medium, which can be used to implement a geometric model meshing method based on machine learning. First, the method has excellent efficiency. Traditional meshing is often time-consuming and labor-intensive, requiring a lot of manual intervention, while the machine learning method can automatically perform meshing, greatly improving the efficiency of meshing and reducing labor costs. Secondly, the method has excellent adaptability. It can automatically adjust the meshing strategy according to the geometric shape and complexity of the model to generate high-quality meshes. This adaptability enables the method to cope with various complex models without the need for tedious manual adjustments and optimizations. Finally, the method also shows extremely high accuracy. Through the optimization of the machine learning algorithm, it can ensure that the meshing result is highly consistent with the actual model, avoiding the errors and distortions that may occur in the traditional method, providing a reliable basis for subsequent finite element analysis or other numerical calculations, and improving the accuracy and credibility of the simulation results. In summary, an efficient meshing method based on machine learning shows significant advantages in terms of efficiency, adaptability and accuracy. These advantages make the method have broad application potential and market prospects in the field of meshing, and inject new vitality into the development of engineering simulation and numerical calculation.
[0110] Electronic devices
[0111] The electronic device provided by an embodiment of the present invention includes a memory and a processor. The memory stores a geometric model mesh division program based on machine learning. When the geometric model mesh division program based on machine learning is executed by the processor, the steps of the geometric model mesh division method based on machine learning provided by the present invention are implemented.
[0112] The present invention provides an electronic device that can be used to implement a geometric model meshing method based on machine learning. First, the method has excellent efficiency. Traditional meshing is often time-consuming and labor-intensive, requiring a lot of manual intervention, while the machine learning method can automatically perform meshing, greatly improving the efficiency of meshing and reducing labor costs. Secondly, the method has excellent adaptability. It can automatically adjust the meshing strategy according to the geometric shape and complexity of the model to generate high-quality meshes. This adaptability enables the method to cope with various complex models without the need for tedious manual adjustments and optimizations. Finally, the method also shows extremely high accuracy. Through the optimization of the machine learning algorithm, it can ensure that the meshing result is highly consistent with the actual model, avoiding the errors and distortions that may occur in the traditional method, providing a reliable basis for subsequent finite element analysis or other numerical calculations, and improving the accuracy and credibility of the simulation results. In summary, an efficient meshing method based on machine learning shows significant advantages in terms of efficiency, adaptability and accuracy. These advantages make the method have broad application potential and market prospects in the field of meshing, and inject new vitality into the development of engineering simulation and numerical calculation.
[0113] Reference Figure 3 , Figure 3 A schematic diagram of the structure of a geometric model meshing device based on machine learning in the hardware operating environment involved in an embodiment of the present invention.
[0114] like Figure 3 As shown, the geometric model meshing device based on machine learning may include: a processor 1001, such as a central processing unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. Among them, the communication bus 1002 is used to realize the connection and communication between these components. The user interface 1003 may include a display screen (Display), an input unit such as a keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a wireless fidelity (WIreless-FIdelity, WI-FI) interface). The memory 1005 may be a high-speed random access memory (Random Access Memory, RAM) memory, or a stable non-volatile memory (Non-Volatile Memory, NVM), such as a disk memory. The memory 1005 may also be a storage device independent of the aforementioned processor 1001.
[0115] Those skilled in the art will understand that Figure 3The structure shown in does not constitute a limitation on the geometric model meshing device based on machine learning, and may include more or fewer components than shown in the figure, or a combination of certain components, or a different arrangement of components.
[0116] like Figure 3 As shown, the memory 1005 as a storage medium may include an operating system, a data storage module, a network communication module, a user interface module, and a geometric model meshing program based on machine learning.
[0117] exist Figure 3 In the geometric model meshing device based on machine learning shown, the network interface 1004 is mainly used for data communication with the network server; the user interface 1003 is mainly used for data interaction with the user; the processor 1001 and the memory 1005 in the geometric model meshing device based on machine learning of the present invention can be set in the geometric model meshing device based on machine learning, and the geometric model meshing device based on machine learning calls the geometric model meshing program based on machine learning stored in the memory 1005 through the processor 1001, and executes the geometric model meshing method based on machine learning provided in an embodiment of the present invention.
[0118] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.
[0119] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.
Claims
1. A geometric model meshing method based on machine learning, characterized in that: The following steps are involved: Obtain the geometric model to be meshed; Preprocessing the geometric model of the to-be-meshed grid to obtain a preprocessed geometric model of the to-be-meshed grid; The preprocessed geometric model to be meshed is input into a mesh partitioning prediction model obtained through machine learning. The mesh partitioning prediction model obtained through machine learning performs mesh partitioning on the geometric model to be meshed to obtain a mesh partitioning result of the geometric model.
2. The geometric model meshing method based on machine learning according to claim 1, characterized in that: The method further comprises the following steps: inputting the pre-processed geometric model to be meshed into a mesh partitioning prediction model obtained through machine learning, wherein the mesh partitioning prediction model obtained through machine learning performs mesh partitioning on the geometric model to be meshed, and obtaining the mesh partitioning result of the geometric model: Performing a quality inspection and / or a simulation calculation verification on the meshing result of the geometric model to obtain a quality inspection and / or a simulation calculation verification result of the meshing result of the geometric model; According to the quality inspection and / or simulation calculation verification result of the mesh division result of the geometric model, the mesh division result of the geometric model is optimized to obtain the optimized mesh division result of the geometric model.
3. The geometric model meshing method based on machine learning according to claim 1, characterized in that: The method for constructing the grid partition prediction model obtained through machine learning comprises the following steps: Acquire a geometric model training sample, wherein a mesh division result of the geometric model training sample is known; Inputting the geometric model training samples into a set machine learning model so that the set machine learning model obtains a mesh division result corresponding to the geometric model training samples one by one; The parameters of the set machine learning model are continuously adjusted so that the mesh division results of the geometric model obtained through the set machine learning model correspond one-to-one with the training samples of the geometric model, thereby obtaining a mesh division prediction model obtained through machine learning.
4. The method for meshing a geometric model based on machine learning according to claim 3, characterized in that: The following steps are also included: Each geometric model of the grid to be divided and the grid division result corresponding to the geometric model of the grid to be divided are added to the training samples.
5. The method for meshing a geometric model based on machine learning according to claim 1, characterized in that: In the step of preprocessing the geometric model of the grid to be divided to obtain the preprocessed geometric model of the grid to be divided, the preprocessing includes at least one of feature extraction, data cleaning and normalization. in, The features extracted by the feature extraction process include at least any one of curvature, boundary conditions, and stress concentration areas; The data cleaning and standardization processing includes cleaning and standardizing the parameters of the features extracted by the feature extraction process, so that the noise and abnormal values of the features extracted by the feature extraction process are eliminated, thereby ensuring the consistency and comparability of the data.
6. The method for meshing a geometric model based on machine learning according to claim 1 or 3, characterized in that: The algorithm of the grid partition prediction model obtained through machine learning is selected from any one of a convolutional neural network algorithm, a random forest algorithm or a support vector machine algorithm.
7. The method for meshing a geometric model based on machine learning according to claim 2, characterized in that: According to the quality inspection and / or simulation calculation verification result of the mesh division result of the geometric model, the mesh division result of the geometric model is optimized to obtain the optimized mesh division result of the geometric model, which specifically includes the following steps: Predict the optimal meshing strategy, where a mesh uniformity metric might be defined as the variance of all element sizes in the mesh: Among them, S i is the size of the ith grid cell, is the average size of all grid cells, and N is the total number of grid cells; Dynamically adjust the grid density, where the grid density adjustment algorithm is: NewDensity=OldDensity× ( 1-α×Gradient Density ) Among them, NewDensity is the adjusted grid density, OldDensity is the current grid density, α is the learning rate, and Gradient Density is the gradient of the optimization objective function with respect to the grid density. Select a mesh type that is suitable for the geometric model to be meshed.
8. A geometric model meshing device based on machine learning, characterized in that: include: A geometric model acquisition module to be divided is used to obtain a geometric model of a grid to be divided; A data preprocessing module, used for preprocessing the geometric model of the to-be-meshed grid to obtain a preprocessed geometric model of the to-be-meshed grid; The mesh division prediction model obtained through machine learning is used to input the preprocessed geometric model of the mesh to be divided into it, and it meshes the geometric model of the mesh to be divided to obtain the mesh division result of the geometric model.
9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a geometric model mesh division program based on machine learning. When the geometric model mesh division program based on machine learning is executed by the processor, the steps of the geometric model mesh division method based on machine learning described in any one of claims 1-7 are implemented.
10. An electronic device, characterized in that: It includes a memory and a processor, wherein the memory stores a geometric model mesh division program based on machine learning, and when the geometric model mesh division program based on machine learning is executed by the processor, the steps of the geometric model mesh division method based on machine learning described in any one of claims 1-7 are implemented.