Advances in Artificial Neural Networks for Electromagnetic Parameterized Modeling

摘要

Electromagnetic (EM) parameterized modeling is important for EM repetitive analysis, such as EM optimization, what if analysis, and yield optimization. An overview of advances in artificial neural networks (ANNs) for EM parameterized modeling is presented in this paper, covering forward/inverse modeling, deep neural networks, knowledge-based neural networks, neuro-transfer functions, and applications for fast EM modeling with varying values in geometrical parameters.