Large Complex Stealth Targets, Accurate Modeling And Fast Algorithm

With the rapid development of microwave technology, higher frequencies are applied in wireless communication and radar systems. Therefore, it is urgent to develop fast algorithms for the analysis of the electromagnetic characteristics of large-scale and extremely large-scale targets. Comparing with other numerical methods for solving scattering and radiation problems, Multilevel Fast Multipole Method (MLFMM) has become popular because of its accuracy and low computational complexity, which enables itself can solve scattering and radiation problems of electrically-large objects. In this thesis, first of all, the integral equations for the electromagnetic scattering problems are introduced. These equations are solved by the Method of Moments (MOM) in detail in order to demonstrate the specific steps for the solution of three dimensional scattering problems by numerical methods.Then the mathematical theory, realization and parallelization of MLFMM are discussed in Chapter 3. Several optimization methods are adopted here to improve efficiency and reduce the computational requirement. In particular, a novel method for reducing the truncation number of MLFMM is proposed in order to economize the memory consumption as well as the computational time without any loss of accuracy.Moreover, Sparse Approximate Inverse Preconditioner (SAI) is adopted in the context of parallel MLFMM for the efficient solution of large-scale problems. Besides, several Krylov subspace based iteration methods, including GMRES, Bi-CGSTAB and Flexible GMRES, are introduced. Performances of these methods are compared for solving large-scale scattering problems of both open and closed objects. Suggestions are provided for the purpose of choosing appropriate methods for different kinds of problems.In Chapter 5, the Impedance Boundary Condition (IBC) is studied for the analysis of scattering problems of PEC objects coated by thin dielectric materials. The valid region of IBC approximation is demonstrated, and the integral equations arising from IBC are solved by MLFMM. The accuracy of IBC-MLFMM is proved by comparing the numerical results with analytical or measurement data. In Chapter 6, the high quality meshing for extremely large objects is discussed, which is the fundament of the accurate numerical simulation using integral equation. By employing the geometric symmetry of the target and the refinement of basic meshes generated by Hypermesh, the high quality meshes can be obtained. Then, the accuracy and capability of our code are demonstrated by computing several scattering problems of extremely large-scale targets. The bi-static RCS of a PEC sphere with diameter of 300 wavelengths, involving more than 0.11 billion unknowns, is computed. The root mean square (RMS) is only 0.527 dB comparing with analytical solution. Furthermore, the mono-static and bi-static RCS of two stealth aircrafts, F117 and vfy-218, are calculated respectively, in order to exhibit the capability of our code for solving the complex real-life targets. The largest electrical size is more than 660 wavelengths, containing around 47 millions unknowns.