Design and analysis of a novel UWB bandpass filter using 3-D EM simulation-based neural network model with HSA

This paper presents a new systematic analysis and design of the ultrawide-band bandpass filter by using the 3-D electromagnetic simulation-based multilayer perceptron neural network (MLP NN) model of unit elements. This MLP NN model is utilized efficiently as a fast and accurate model within a harmony search algorithm (HSA) procedure to determine the resultant optimum microstrip structure geometry. Moreover, the validity and efficiency of the HSA is manifested by comparing it with those of the standard metaheuristics, which are genetic and particle swarm algorithms. The filter that shows the best performance is designed and realized. Measurements taken from the realized filter demonstrate the success of this approximation over the band range of 3.1 GHz to 10.6 GHz, with a flat group delay performance within that range.