An electromagnetic parameter retrieval method based on deep learning

Abstract

The electromagnetic parameter retrieval method based on scattering (S) parameters is widely used because S parameters are easy to obtain. As a classic S parameters retrieval method, the Nicolson–Ross–Weir (NRW) algorithm has problems such as half-wave resonance, phase angle jump, and multivaluedness. In this work, an electromagnetic parameter retrieval method based on deep learning is proposed, which aims to solve the multivaluedness problem in the traditional NRW method. The method we proposed is suitable for retrieval of inhomogeneous medium that satisfies the theory of effective medium. The simulation shows that the electromagnetic parameter retrieval method based on deep learning has high calculation accuracy without the multivaluedness problem.