Research On GPU Acceleration For Sheet Dynamic Analysis And Topology Optimization

The sheet has the characteristics of two-dimensional structure,is light in weight and can withstand large loads,and is widely used in aerospace,wind power and petrochemical industries.In order to ensure the safe use of the thin plate structure,it is very important to perform bending and vibration analysis.Therefore,Element-Free Galerkin Method(EFG)with high accuracy and fast convergence is used to study the dynamic analysis and topological optimization of sheet in this paper.However,the coefficient matrix based on the EFG method is time-consuming to solve,and the coefficient matrix needs to be solved multiple times in dynamic analysis and topology optimization,seriously affect the computational efficiency of numerical experiments.Therefore,the parallel acceleration of GPU(Graphics Processing Unit)for dynamic analysis and topology optimization of sheet will be studied in this paper.The main contents are as follows:(1)Research on GPU parallel algorithm for static analysis of sheet and its topology optimization.According to the characteristics of sheet theory and GPU parallel computing characteristics,the parallel computing flow of static analysis GPU is designed.Combined with the sheet topology optimization theory,the topology optimization GPU parallel computing flow is designed.The feasibility and effectiveness of the algorithm are verified by numerical examples.Among them,the acceleration ratios of static analysis and topology optimization are about 25 and 28 times respectively,the relationship between node size and acceleration ratio and the influence of the angle of orthotropic material laying on the topology optimization results are discussed.(2)Research on GPU parallel algorithm for sheet modal analysis and its topology optimization.Combining the inverse iterative method theory and the subspace method theory,the corresponding modal analysis GPU parallel computing flow is designed.Based on the inverse iterative method and the sheet modal topology optimization theory,the modal topology optimization GPU parallel computing flow is designed.The feasibility and effectiveness of the algorithm are verified by an example,and the factors affecting the acceleration ratio and computational efficiency are discussed.(3)Research on GPU parallel algorithm for sheet dynamic response analysis.Based on the parallel computing of EFG method and the computational characteristics of Newmark method,the GPU parallel computing flow of dynamic response analysis is designed.The effectiveness of GPU parallel algorithm is verified by numerical examples,the acceleration ratio is about 39 times,and the effects of different factors on acceleration ratio and computational efficiency are discussed.In this paper,the GPU parallel algorithm is used to effectively reduce the computation time of the structure in dynamic analysis and topology optimization.This research content has important theoretical reference value and practical guiding significance for engineering application based on EFG structural dynamic analysis and topology optimization.