Selçuklu Yıldızı Şekilli Mikroşerit Anten İçin Ateş Böceği Algoritmalı Yapay Sinir Ağı Modeli

Abstract

In this study, Seljuk Star microstrip antenna (SSMA) design based on the hybrid Artificial Neural Network model for frequency values in the range of 0.5-3.5 GHz has been performed. In the present study, a new algorithm is developed for neural network training by combining a back propagation (BP) and a meta-heuristic algorithm. The major disadvantage of back propagation in finding solutions is that it stuck local minima rather than global one. In this new hybrid training algorithm, local and global search made simultaneously. Initially, Firefly Algorithm (FA) was utilized to obtain weights of neural network due to the lower probability of entrapment into local minima thanks to long jump. Then, this algorithm is combined with back propagation (BP) to use the advantages of enhanced global search ability of Firefly Algorithm and local search ability of BP algorithm in training neural network. Levenberg-Marquardt back propagation algorithm was used in the training phase of the Artificial Neural Network. In this paper, Seljuk Star microstrip antenna has been designed on DE104, double faced with 1.55mm dielectric and 35um conductor thickness, which has an electrical conductivity of 4.37 and a loss tangent of 0.002. HFSS antenna simulation program was used to design for 272 microstrip antennas. 90% of the data set was used as training and 10% as test data. The ANN with Firefly Algorithm results are more in agreement with the simulating results.