The Bouncing Ray Method For Electromagnetic Scattering Analysis Of Metal/medium Targets And Its Implementation On GPU Computing Platform

How to efficiently and quickly calculate the Radar cross section(RCS)has been the focus of research on computational electrimagnetics.Since the shooting and bouncing ray(SBR)method is a high frequency approximation method that combines the advantages of the geometric optics method and the physical optics method,it is very suitable for the complex target electromagnetic computing problems.However,as the target size becomes electrically larger,the electromagnetic calculation will be very time-consuming,which greatly limits the application of the SBR method in electromagnetic scattering calculations.In this thesis,the combination of soft acceleration and the GPU-based CUDA parallel computing techniques has greatly improved the computational efficiency of the SBR method.The main work of this thesis is as follows:1.The related knowledge of GPU hardware is introduced,and the GPU software programming model and related optimization methods are studied.2.Then introduces the basic processes and related technologies of the traditional SBR method.The accelerated SBR method based on the adaptive subdivision of the ray tube is added,and the accelerating effect is verified.The parallel processing method of ray tube on GPU is introduced and the parallel strategy to improve parallel computing efficiency is discussed.The parallel computing of SBR method on GPU/CPU heterogeneous platform is implemented.3.The SBR method is introduced into the electromagnetic scattering calculation of the dielectrics/mentals hybrid targets.The wave propagation in the dielectric materials is deduced in detail,and the correctness of the method is verified by numerical examples.The application of SBR in the RCS calculation of coated target is introduced and the solving process of multi-layer coal reflection coefficient is deduced.The characteristics of this method and the hybrid target SBR method are analyzed through numerical examples respectively.Finally,taking into account the long time-consuming when calculating the RCS of the dielectrics/mentals hybrid targets,the GPU-based CUDA parallel techniques have been introduced to implement the parallel computing of dielectrics/mentals hybrid SBR method on GPU/CPU heterogeneous platform.