Application of Generalized Constrained Neural Network with Linear Priors to Design Microstrip Patch Antenna

摘要

This paper describes usage of generalized constrained neural network with linear priors (GCNN-LP) as knowledge based neural network in resonant frequency calculation of various microstrip configurations. Linear priors can be defined as a course of previous evidence that unveils a direct in to the features of benefits, such as variables, free parameters, or their tasks of the models. Recently generalized constraint neural network with linear priors have been suggested by Hu et al. which is a step forward for this proposed work. It takes many known priors like equality, symmetry, ranking, interpolating points, etc., as prior knowledge about the problem. In this paper, GCNN-LP is applied to estimate resonant frequency of rectangular, circular, and elliptical microstrip antenna.