COMPUTING RESONANT FREQUENCY OF H-SHAPED COMPACT MICROSTRIP ANTENNAS OPERATING AT UHF BAND BY USING ARTIFICIAL NEURAL NETWORKS

In this study, resonant frequency of H-shaped compact microstrip antennas (HCMA) operating at UHF band is computed by artificial neural network (ANN) based on the multilayer perceptrons (MLP). The resonant frequencies of 216 HCMA having different electrical and physical parameters are determined by simulating with the use of electromagnetic packaged software based on finite-difference time domain method (FDTD). 196 HCMA are employed for training while 20 HCMA are used for testing ANN. For training and testing, average percentage error values are obtained as 0.68 % and 0.84 %, respectively. Furthermore to point out the validity and accuracy of ANN, simulation and measurement resonant frequency results given in the studies presented in literature, and measurement results of the HCMA designed in this study are also utilized and the average percentage error is obtained as 1.18 % and 1.20 %, respectively. The results achieved in this work illustrate that ANN based on MLP can be successfully used to determine the resonant frequency of HCMA.