PREDICTION OF THE ELECTROMAGNETIC FIELD IN METALLIC ENCLOSURES USING ARTIFICIAL NEURAL NETWORKS

Abstract

In complex electromagnetic (EM) environment, EM fleld distribution inside a metallic enclosure is determined by the external EM radiation and emissions from internal contents. In the design of an electronic system, we usually need to estimate the EM fleld level in a concerned region inside the enclosure under various EM environments. In this paper, we use artiflcial neural network (ANN), rather than full wave analysis, combined with the numbered measurements to predict the EM fleld in the concerned region inside a metallic enclosure. To verify this method, a rectangular metallic enclosure with a printed circuit board (PCB) is illuminated by external incident wave. The measured electric flelds inside the enclosure combined with ANN model based on back propagation (BP) training algorithm are used to estimate the values of electric fleld. The calculation is fast and predictions reveal good agreement with the measurements that validate this method.