Machine Learning Based Computational Electromagnetic Analysis for Electromagnetic Compatibility

Abstract

While machine learning is becoming a demanding request in every corner of modern technology development, we are trying to see if we could make computational electromagnetic algorithms compatible to machine learning methods. In this paper, we introduce two efforts in line with this direction: solving method of moments (MoM) can be seen as a training training process. Consequently, the artificial neural network (ANN) could be used to solve MoM naturally through training. Amazon Web Service (AWS) can be used as the computations platform to utilize the existing hardware and software resources for machine learning. Another effort regarding to the nonlinear IO of ICs can be modeled through ANN. Hence, a behavior model with growing accuracy can be obtained for the signal integrity and power integrity analysis. It can be further hybridized into discontinuous Galerkin time domain (DGTD) method for CEM characterizations. Benchmarks are provided to demonstrate the feasibility of the proposed methods.