Model Order Reduction Method For Electromagnetic Analysis

Finite Element Method (FEM) is quite popular numerical method in computational electromagnetics. Linear equations will be generated after FEM discretization in electromagnetic analysis. When the oriental model is complex, the order of vector equation which is same as linear equation will be very large. Solve high order matrix equation is quite slow, and lead low efficiency in simulation of electromagnetic project. Model order reduction (MOR) techniques may efficiently reduce the order of matrix equation and also maintain the character of oriental model. Using reduced model can reduce computational complexity, shorten time cost of simulation and bring high efficiency.A model order reduction method based krylov subspace theory named Arnoldi method is introduced in this paper for reduction of vector equation by FEM formulation. High order matrix equation calculation is transferred in low order format. During the calculation of orthogonal matrix Q which is used to reduce order of coefficient matrix of oriental model equation, two parameter K andεr are considered in moment match. By reasonable selection of the number of moment for each parameter, error between reduced model and oriental model can be consider small enough to meet the demanded accuracy in practical electromagnetic project. A extend Arnoldi method suit for quadratic equation is used in moment match of parameter K. In addition, design parameter sensitivity is also analyzed in this paper. The design sensitivity about dielectric permittivity dS11/dεr is calculated between reduced model and oriental model. The results of two models also get highly coherence.