Finite element mesh optimization method and device based on deep reinforcement learning and medium

A technology of reinforcement learning and finite element, applied in design optimization/simulation, character and pattern recognition, biological neural network model, etc., can solve the problem that the grid cannot fully meet the analysis and calculation, and achieve the effect of improving quality and efficiency
CN113221403AActive Publication Date: 2021-08-06中汽数据(天津)有限公司
6 Cites 4 Cited by

Patent Information

Authority / Receiving Office
CN · China
Current Assignee / Owner
中汽数据(天津)有限公司
Publication Date
2021-08-06

Smart Images

  • Figure 1
    Figure 1
  • Figure 2
    Figure 2
  • Figure 3
    Figure 3
Patent Text Reader

Abstract

The invention relates to the field of grid division, in particular to a finite element grid optimization method and device based on deep reinforcement learning and a medium, and the method comprises the steps: obtaining initial finite element grid data of a geometric model; generating a three-dimensional model of the geometric model according to the initial finite element grid data; performing feature recognition and classification on the three-dimensional model by adopting a detection network to obtain local features of the three-dimensional model and a classification result of the local features; inputting the unreasonable local features into an optimization network so as to enable the grid quality of the adjusted finite element grid data to meet a set requirement; capturing a readjustment strategy of a user on the adjusted finite element grid data, and performing reinforcement learning on the optimization network by taking the readjustment strategy as forward excitation so as to update the optimization network; and continuing to optimize the adjusted finite element grid data by adopting the updated optimization network. According to the embodiment of the invention, the accuracy of grid division can be improved.
Need to check novelty before this filing date? Find Prior Art

Description

technical field

[0001] The present invention relates to the field of grid division, in particular to a finite element grid optimization method, device and medium based on deep reinforcement learning. Background technique

[0002] The common methods of finite element mesh division include geometry coded topology, graph coded topology, similarity heuristic topology and character coded topology. In mainstream computer-aided engineering (Computer Aided Engineering, CAE) simulation software, the bottom layer of mesh division is realized by geometric coding topology.

[0003] CAE is a very important link in the process of automobile research and development, among which mesh division is the most basic and important pre-work. Grid work accounts for about 50% of the workload of automotive R&D simulation, and the quality of the grid also has a greater impact on the simulation results. Therefore, how to improve the efficiency of grid division and ensure the quality of the grid is a ...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More