Deep Learning Inverse Analysis of Higher Order Modes in Monocone TEM Cell

Abstract

This article proposes a broadband inverse analysis method for the higher order modes inside the resistively loaded monocone transverse electromagnetic (TEM) cell. The analytic solutions of the field pattern for each higher order mode are derived in the frequency domain. The deep learning technique is employed to effectively predict coefficients of these modes over a wide frequency range (1.885–300 MHz) only given the <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$E$ </tex-math></inline-formula> -field amplitude distribution. The predicted results agree well with measurement ones, which validates the accuracy of the deep neural network method, and furthermore, it is analyzed in the time domain with fast Fourier transform (FFT). This work reveals the mechanism of higher order mode-induced field distortion in both the frequency domain of broadband and the time domain, provides guidance for elimination of higher order modes in monocone TEM cells, and can be extended to mode analysis for other kinds of wideband antennas.