A Deep Learning Based Design Method for Circuit Analogue Absorbers

A deep learning-based method for the inverse design of circuit analogue absorbers (CAA) is proposed. Four regular and simple shaped mental with lumped resistances are selected as the basic structures of circuit analogue absorbers. By integrating a classification network (CN) and a geometric parameters prediction network (GPPN), the proposed method enables to predict the class of base-shaped CAA according to the sample data extracted from the input electromagnetic responses. The CN categorizes the electromagnetic responses, while the GPPN predicts the corresponding geometric parameters. Simulation results demonstrate the effectiveness of the method in the design of circuit analogue absorbers.