Data Mining Modeling for Electromagnetic Scattering Computing

摘要

Scattering computing of metal-media complex structure and structure with cavity has been one of the problems in electromagnetic (EM) scattering theoretical calculation field for many years. A novel method, substituting data mining modeling for original theoretical modeling, is proposed creatively in this paper, attempting to solve the problem by machine learning theory. Data mining modeling is to construct "EM scattering training model", applying regression analysis algorithm on measurement data, to achieve the effect superior to that theoretical modeling can have. Given an example of regressive estimation of some inlet backscattering RCS curve, both original least square algorithm and support vector regression are used, so an applicable data mining model is established initially for EM scattering computing.