Reflection Phase Analysis based on Multilayer Perceptron Network Model for Unit Element Design of a Dual-Layered Microstrip Reflectarray

In this paper, the most common neural network architecture called as the multilayer perceptron (MLP) Model is presented as convenient interface for obtaining and optimizing the phase characterization of the dual-layered Minkowski unit element. The high priority objective of Reflectarray antenna design is to get the reflection phase characteristic that has small gradient and wider phase range. For this purpose, firstly the Reflectarray unit element was placed at the end of a standard X-band H-wall waveguide simulator and its reflection phase characteristics were obtained by the 3D Computer Simulation Technology Microwave Studio (CST MWS) simulations. Thereafter MLP Network Model is constructed to approximate this nonlinear relationship between the geometrical properties of antenna parameters and the reflection phase characteristics. In comparison with the target data network output data, it is understood that the MLP network structure can be used as a very effective method to estimate the reflection phase characteristics of Reflectarray unit element.