Analysis and design of X-band Reflectarray antenna using 3-D EM-based Artificial Neural Network model

This paper presents analysis and design of the X-band Reflectarray using a three-dimensional EM simulation based- Multilayer Perceptron Neural Network (MLP NN) model. For this purpose, firstly training and validation data for the reflection phase characteristics of a unit element is obtained by the 3-D Computer Simulation Technology Microwave Studio (CST MWS) depending on the geometrical parameters, substrate and frequency using the H-wall waveguide simulator. Thereafter, the MLP NN is validated utilizing the CST MWS test data. Comparison of the reconstructed data with the target data shows that MLP can be used as a very efficient technique to predict the reflection phases of reflectarray. The MLP NN modeling is applied to the minkowski and circular ring shaped radiators and their reflection characteristics are obtained within the continuous domain of the geometrical parameters, substrate and frequency. Some typical reflection phase characteristics are given. Furthermore, a minkowski reflectarray of 5 × 5 on Taconic RF-35 with ε_r =3.5 and tan δ =0.0018 is designed and its radiation patterns are simulated by using CST MWS. The bandwidth characteristics of the array are also observed within the X-band frequency range.