Multi-adaptive neuro-fuzzy inference system modelling for prediction of band-notched behaviour of slotted-UWB antennas optimised using evolutionary algorithms

Abstract

This study introduces modelling of multi-adaptive neuro-fuzzy inference system (MANFIS) for predicting the bandwidth and notch frequencies of slotted ultra-wideband (UWB) antennas. Rectangular-shaped printed monopole antennas embedded with U-shaped slots are designed to realise triple band-notch characteristics. A MANFIS model is then developed to predict five output parameters (two cut-off frequency points and three notched frequency points) considering 15 geometrical variables of the designed antennas as the inputs of MANFIS model. Extensive simulation has been performed using HFSS software to generate training and testing data patterns. Two optimisation algorithms, genetic algorithm (GA) and particle swarm optimisation (PSO), are implemented to optimally determine the appropriate values of the fuzzy inference system parameters. For GA-optimised MANFIS model, the percentage error is observed between 1 and 2%, whereas, in PSO-trained MANFIS model, it is observed <1%. The comparative analysis establishes that the PSO-trained MANFIS model precisely predicts the antenna performances. For validating the proposed modelling technique, an optimised configuration of the slotted UWB antenna prototype has been fabricated and characterised.