Study Of Key Parallel Computing Technology For Method Of Moments And Finite Element Method

The electromagnetic related issues become more and more complex,as the development of science and technology and the progress of society.Electrically large size,complex structure,high integration and systematization are main features of current electromagnetic targets,and this leading to the traditional electromagnetic computing methods often face the problem of lack of computing resources or time consume is horrible.Though high-performance computing technology has been gradually applied to the electromagnetic computing field in recent years and has solved a number of challenging issues,its theoretical system and application scope is still immature compared to other areas,there are still many worth place of study.Focusing on the practical engineering application requirements the parallel computing technology of the method of moments(MoM)and finite element method(FEM)are studied in this thesis to use current distributed cluster's computing power and storage capacity improve their capacity and efficiency of electromagnetic simulation.Aiming at the trend of heterogeneous development of current computer,a heterogeneous parallel technology of MoM is studied.By utilizing the CONTEXT technology in CUDA which is the GPU common programming criterions,and the concept of environment variables in MIC,a general heterogeneous parallel programming model for CPU/GPU and CPU/MIC is proposed,which meets the static load-balancing for the across-nodes heterogeneous parallel MoM.Numerical results show that the heterogeneous parallel MoM codes based on the proposed parallel programming model can obtain ideal speedup and good scalability.It is a very common scenario that using the integral equation to calculate the near or far field in practical engineering applications.When the number of near-field or far-field points is very large this computing process is very time consuming.In this thesis,using parallel multilevel fast multipole algorithm accelerating this process is studied,and a detailed accelerating theory and parallel strategy are summarized.Through the numerical examples,the speedup of parallel multilevel fast multipole algorithm is verified compared with the traditional integral equation method.Finite element iterative integral equation method that a novelty mesh truncation technology currently in the field of domestic finite element research,and that can reduce the unknowns of the finite element absorbing boundary region while the original sparse and banded structure of the FEM matrices is retained.Focusing on the problem of high memory complexity and too much time are needed when use FEM analyze complex and electrically large problems,the parallelization based on MPI and OpenMP hybrid programming model is studied.Large base station antenna(64 elements)equipped with real feeding network which hard to efficient solved by using commercial software-ANSYS HFSS is simulated use 1 152 CPU cores in this thesis,finally.In summary,heterogeneous parallel method of moments,multilevel fast multipole algorithm accelerating field calculation and parallel finite element iterative integral equation method are realized in this thesis,and all achieved good parallel speedup.These works provide some available technical means to realize efficient electromagnetic simulation.