Analysis Of Electromagnetic Scattering Characteristics Of Target Separation Process Based On GPU Parallel Acceleration

Solving electromagnetic scattering problems of complex targets efficiently is very important in radar system design,especially the correct handling of the joint surface during the process of targets' separation process and the efficient reuse of commom information.In recent years,original characteristic basis function method(original CBFM)and improved characteristic basis function method(CBFM)with integral equation discontinuous Galerkin(IEDG)methods and its derived method multilevel characteristic basis function method(MLCBFM),as new accurate and efficient methods for solving electromagnetic scattering problems,have attracted people's widely attention.Constrained by the traditional serial structure,the method of improving the speed of operation by increasing the main frequency of CPU has encountered problems and the effect is limited.In contrast,although GPU is designed for graphic and image processing functions,it is also emerging in the field of large-scale high-performance parallel computing,and is rising rapidly.GPU large-scale high-performance parallel computing has made many great breakthroughs in many fields,such as electromagnetics,astronomical computing,medical images and so on.At present,the cores of all kinds of electromagnetic simulation softwares on the market are mostly written by CPU code,which is slow in computation and inefficient in use,and does not possess the function of dynamic simulation or needs complex tools such as script files to realize it.The paper aiming at electrically large object model,setting in frequency domain analysis of electromagnetic scattering modeling of stages in complex targets'separation process,researching CBFM and MLCBFM with IEDG,and using the current popular GPU parallel technology to optimize the paralleling modular of the algorithm,stability simulates the Radar Cross Section(RCS)of electrically large targets and stages in complex targets' separation process.After optimization,the GPU version software is not only much faster than the CPU version,but also can adapt to many complex simulations and performs very well in engineering applications,so it has very high practical value.