EM Scattered Field Prediction Using Extreme Learning Machine
Akhila Gouda, Nihar Kanta Sahoo, Dhruba Charan Panda, Rabindra Kishore Mishra · 2022 IEEE Wireless Antenna and Microwave Symposium (WAMS) · 2022
This paper introduces extreme learning machine for the prediction of scattered electromagnetic fields using initial FDTD data. It uses the FDTD model to store scattered fields from equivalent PEC objects. It then uses the stored field as input of ELM for the prediction of scattered fields from dielectric objects. The scattered field from ELM prediction, conventional FDTD and commercial package TaraNG follow each other closely over the time steps. This technique can predict with high accuracy while the absolute error percentage remains below 0.04. The ELM prediction is almost instantaneous whereas FDTD and TaraNG require much larger computing time.