Neural Networks in Antennas and Microwaves: A Practical Approach

Abstract

Neural networks are electronic systems which can be trained toremember behavior of a modeled structure in given operational points,and which can be used to approximate behavior of the structure out ofthe training points. These approximation abilities of neural nets aredemonstrated on modeling a frequency-selective surface, a microstriptransmission line and a microstrip dipole. Attention is turned to theaccuracy and to the efficiency of neural models. The association ofneural models and genetic algorithms, which can provide a global designtool, is discussed.