AI-based estimator for computational discovery and synthesis of customized microwave absorbing materials

Abstract

This article investigates the viability of a deep neural network (DNN) for the computational discovery and synthesis of efficient microwave-absorbing materials and structures. A DNN is trained to tackle specific objectives in a constrained environment by utilizing the conventional forward and reverse approaches. In the forward approach, the DNN predicts various topologies of the absorbers and it is found to be effective in determining the stacking sequence of microwave-absorbing materials and their associated thicknesses. Designing a microwave absorber is observed to be exceptionally cumbersome utilizing a DNN if the material database increases unexpectedly. Following that, the solution is offered by addressing the reverse approach, in which a DNN is utilized to forecast the electromagnetic (EM) properties based on user-defined specifications. It is a convincing and simple method of designing thin and wideband customized absorbers. DNN prediction is authenticated by fabricating two distinct absorbers based on the frequency-dependent EM properties. Furthermore, the synthesized model is tested and validated with the response of the EM mixing model and microwave measurements. The suggested DNN strategy can effectively fix the issues in designing thin and broadband absorbers.