A comparative study of performance of different back-propagation neural network methods for prediction of resonant frequency of a slot-loaded double-layer frequency-selective surface

A comparative study of back-propagation neural network and method of moment-based result is articulated to investigate the performance criteria of a slot-loaded double-layered frequency-selected surface. Combining the conventional merits of the artificial neural network and considering the simplicity and robustness of the analysis, only three artificial neural network techniques—the gradient descent back-propagation, the gradient descent with momentum and the conjugate gradient back-propagation—are considered to be the most efficient candidates for artificial neural network analysis so far. The resonant frequencies are obtained from trained artificial neural network by exploiting various properties of double-layered frequency-selected surface and are compared with some test bed consequences against the method of moment-based outcomes. This analysis is undoubtedly imperative as it is based on rigorous soft computations. Moreover, it creates greater impact in impending study of slot- and slit-loaded frequency-selected surface as the results achieved so far are in good agreement.