Microwave circuit models using the structured genetic algorithm

This paper describes an application of the structured genetic algorithm to the construction of microwave circuit neural network models. The algorithm is used to create a neural network model of a high electron mobility transistor (HEMT) power amplifier. Results are compared to other neural network training algorithms such as back-propagation and the Levenberg-Marquardt algorithm. Models of one and two-stage amplifiers created with the ADS circuit simulator give good predictions of measurement results. © 2005 IEEE