Design of Metamaterial Absorber at X Band Using BP Neural Network

The absorption performance of metamaterial absorber is determined by the configuration parameters of the metallic pattern, the thickness and dielectric constants of the substrate. Conventional parameter sweep method and adjoint- based optimization are time consuming. In this work, a metamaterial absorber at X band is designed by using the BP neural network. In order to determine the geometry parameters for realizing the perfect absorption performance, 81 different simulation results are trained with the BP neural network. The designed structure can obtain the low reflection of -14.09dB at 10GHz. Compared to the full wave simulation, the calculation time with BP neural network is much shorter, which can be widely applied in other artificial electromagnetic structure based on metamaterials.