Neural Network Characterization of Microstrip Patches for Reflectarray Optimization

The increasing interest among the scientific community of reflectarray antennas in recent years leads to develop effective ad-hoc ways to optimize antenna performances, often by means of evolutionary iterative algorithms. To enhance the speed of the optimization task, in this work a Neural Network model of an annular patch antenna is presented as convenient interface between antenna design and global optimization algorithms. In particular, an artificial neural network (ANN) is proposed to predict the phase behaviour of a broadband patch radiator as a function of its geometric parameters. The characterization of the antenna is first obtained by numerical simulations, and the ANN is constructed to approximate the nonlinear relationship between the antenna geometry and the phase behaviour. Some preliminary results are provided to demonstrate the effectiveness of this method in providing a fast interface between the antenna model and the optimisation tool.