Genetic algorithm based selective neural network ensemble method to analyse rectangular microstrip antenna

Abstract

Harmony in variety i.e. unity without similarity is a concept inspired from ancient times. Thinkers propose a team approach based on the same concept for problem solving i.e. using a combined group of solvers to resolve a difficult problem. Neural network ensemble (NNE) is a concept based on the same approach. Multiple artificial neural networks (ANNs) are trained for the same dataset to give the appropriate measured resonant frequency from the relative parameters of rectangular microstrip antenna (MSA). The previous experimental works' MSA datasets have been used for training of ANNs. Genetic Algorithm (GA) is employed to compute the optimum subset of ANNs which perform better than rest available to constitute an ensemble. A model of resonant frequency of MSA is established by using this NNE approach and the results have been compared with some previous works.