Application of the ARTMAP neural network in the design of cascaded gratings and frequency selective surfaces

Currently, there is no closed form solution that can directly relate a desired frequency response to a corresponding frequency selective surface (FSS). Trial and error procedures are used until a frequency selective surface matches the desired criteria. One way of avoiding this laborious process and obtain a synthesis procedure is to utilize the training capabilities of neural networks. A neural network can be trained to keep changing the dimensions of the metallic strips or patches, their distance of separation, their shape, and the number of layers required in a multilayer structure until the frequency response matches the desired one. In the past, to achieve this goal, the back propagation learning algorithm was used in conjunction with an inversion algorithm. Unfortunately, both the back-prop algorithm and the inversion procedure are slow to converge. Others used "genetic algorithms" to solve the same problem. In this work the Fuzzy ARTMAP neural network is utilized. The Fuzzy ARTMAP is faster to train than the back-prop and it does not require an inversion algorithm to solve the FSS problem. Several results (frequency responses) from cascaded gratings for various angles of wave incidence, layer separation, width strips, and interstrip separation are presented and discussed.<>