Tehnički vjesnik, Vol. 20 No. 6, 2013.
Izvorni znanstveni članak
Neural network based estimation of resonant frequency of an equilateral triangular microstrip patch antenna
Kamil Yavuz Kapusuz
; Atilim University, Department of Electrical & Electronics Engineering, Kizilcasar Mahallesi, 06836, Incek Golbasi, Ankara, Turkey
Hakan Tora
; Atilim University, Department of Electrical & Electronics Engineering, Kizilcasar Mahallesi, 06836, Incek Golbasi, Ankara, Turkey
Sultan Can
; Dept. of Electronics Engineering, Ankara University, Tandogan, 06100 Ankara, Turkey
Sažetak
This study proposes an artificial neural network (ANN) model in order to approximate the resonant frequencies of equilateral triangular patch antennas. The neural network structure applied here is trained and tested for both single-layer and double-layer antennas. It is shown upon experiment that the resonant frequencies obtained from the neural network are both more accurate than the calculated frequencies by formula and satisfactorily close to the measured frequencies. Results appear to be promising as per the available literature. This paper also may offer more efficient approach to developing antennas of such nature. While the total absolute error of 7 MHz and the average error of 0,09 % are achieved for single-layer antenna, the total absolute and average errors are 49 MHz and 0,07 % for the double-layered antenna, respectively.
Ključne riječi
microstrip antenna; neural network; resonant frequency
Hrčak ID:
112313
URI
Datum izdavanja:
20.12.2013.
Posjeta: 2.329 *