An efficient application of machine learning for designing user specific transmissive radome

Abstract

Frequency selective surfaces (FSSs) are widely used for transmissive and absorptive radomes in various frequency ranges. Lack of scientific formulation makes it challenging to determine the dimensions of the FSS radomes corresponding to user specific applications. Therefore, we have proposed a novel methology using machine learning technique for determining the physical parameters of FSS by which the challenge faced due to trial and error technique may be minimised. Simulated data set has been used to train the neural network model producing 8.5 million combinations of physical parameters and their electric responses. These combinations are arranged as look‐up table and optimized physical parameters are obtained using error minimization. The methodology has been implemented on a square loop geometry and validated using high frequency structure simulator. A prototype has been fabricated and measured in free space setup. The measured results are in good agreement with the obtained simulated results.