Phase Characterization of X-band Minkowski Reflectarray Antennas Using 3-D CST Microwave Studio-based Neural Network Model Included Dielectric Properties

Abstract

This paper presents a complete MLP NN model including the dielectric property epsilon(r) of the substrate is built to be used in both the design and performance analysis of the Minkowski RAs for the X-band applications. For this purpose, 3-D simulations by Computer Simulation Technology Microwave Studio (CST) are generated from Minkowski radiating element placed at the end of a standard X-band waveguide, by varying the geometrical parameters of the patch for each substrate (epsilon(r) - h) at each operation frequency f within the defined input domain. Thus, a data set is composed. After the training and validation data sets are determined, the MLP NN with the Levenberg-Marquardt algorithm is applied to the actual data. In the final stage, the reflection phase characterization of the MLP ANN model for a Minkowski patch is presented as the numerous graphics depicting the some variations of the geometry parameters and substrate parameters (epsilon(r) - h) as compared with analysis of the 3-D EM simulator CST and interpreted in terms of design parameters.