Artificial Neural Network Based Microstrip Reflectarray Unit Element Design

Authors : Mehmet BEREKET, Aysu BELEN, Mehmet Ali BELEN

Pages : 160-164

View : 19 | Download : 14

Publication Date : 2021-12-30

Article Type : Research Paper

Abstract :Microstrip reflectarray antennas insert ignore into journalissuearticles values(RAs); are designs that can achieve equivalent performance of parabolic reflector, but with simple and light electromagnetic and mechanical structures. The challenging problem in design of RA is the fast and accurate modelling of the unit element for the array optimization. 3D EM simulators are computationally very ineffective, thus in this study artificial neural network based unit element modelling for characterization of the reflection phase of the unit element in terms of its geometry, and operation frequency is studied. For this mean, a Malta Cross shaped design for X-band applications is taken into the consideration using Multilayer Perceptron insert ignore into journalissuearticles values(MLP); neural network trained the 3D CST microwave Studio simulator data. Validation of the MLP model is also worked out successfully with the 3D CST data. By this mean, a continuous function is obtained for the reflection phase of the unit element with respect to the variation of geometrical design parameters and operation frequency had been achieved which can be used for a design optimization process fast as analytical approach design while being accurate as 3D EM simulator tools.
Keywords : Artificial neural network, multi layer perceptron, reflectarray antenna, X band