Automatic decomposition method for two-dimensional structure grid of multi-inner-hole part based on neural network

Abstract

The invention discloses an automatic decomposition method for a two-dimensional structure grid of a multi-inner-hole part based on a neural network. The introduction of a manual link in the existing block structured grid generation greatly reduces the grid generation efficiency. The method comprises the following steps: firstly, making a sample set for neural network learning two-dimensional region decomposition, and training a neural network model through position information of each grid point in the sample set and frame vector labeling information of each frame; performing neural network prediction on the prediction sample data of the multi-inner-hole part of which the structure grids are to be divided, and processing the prediction sample data by utilizing frame vector labeling information predicted by the neural network to obtain final regional decomposition data; and finally, generating a quadrilateral grid of the multi-inner-hole part of the to-be-divided structure grid by utilizing a mapping method. The method can achieve the quick automatic region decomposition of a new model, and is of great significance to the quick and accurate simulation analysis of a multi-inner-holepart.

Description

technical field [0001] The invention relates to the area decomposition process of block structured quadrilateral grid generation in the preprocessing of the field of numerical simulation, in particular to a neural network-based automatic two-dimensional structure grid decomposition method for multi-hole parts. Background technique [0002] The grid types commonly used in numerical simulation analysis include structured grid, unstructured grid, nested grid, right-angle grid and mixed grid, etc., and each type of grid has its own advantages and disadvantages. In two-dimensional problems, structural grids are composed of regularly arranged quadrilateral elements, which are favored in various simulation analyzes due to their high solution accuracy and small number of required elements. However, compared with other types of grids, structured grids have strong topological constraints; for complex geometric models, generating a set of high-quality structured grids is a very time-co...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to provide a neural network-based automatic decomposition method for two-dimensional structural grids of multi-hole parts for the shortcomings of the need to introduce artificial domain decomposition in the generation of existing block structured grids. By analyzing a large number of existing It is suitable for learning the region decomposition data of multi-hole parts generated by block-structured quadrilateral mesh, training the neural grid to recognize the structural features inside the region of multi-hole parts, and realizing the fast automatic region for multi-hole parts with similar topology break down

Images