Artificial Neural Networks for Microwave Behavioral Modeling

This talk describes artificial neural network techniques for RF/microwave behavioral modeling. Neural networks are trained to learn the behavior of RF/microwave passive and active devices and circuits. The trained neural networks become fast and compact behavioral models for high-level circuits and systems design. Knowledge-based approaches combining microwave knowledge with neural networks allow us to use machine learning approaches to build new behavioral models by modifying existing ones. Neural networks can be trained to learn inverse relationships providing direct solutions to microwave problems where no direct formula is available. Applications of neural networks for behavioral modeling of passive and active microwave circuits including electromagnetic structures, nonlinear microwave devices and power amplifiers will be demonstrated.