RBF model of microwave filter using PDGS with defected rectangles

Abstract

Microwave filter is the kind of device which can separate different signals within a certain band of microwave frequency, widely used in microwave communication system. Periodic defected ground structures (PDGS) with defected rectangles have excellent filter properties when periodic unit amounts and structure sizes meet appropriate conditions. According to artificial neural network (ANN) theory, radial basis function (RBF) model of PDGS with etched rectangles is developed for the first time on the basis of FDTD analysis in this paper. Periodic unit amounts, the structure sizes of PDGS and the frequency are defined as the input samples of the RBF, transmission coefficient (S21) are defined as the output samples. Transmission coefficient of PDGS at any arbitrary parameters including periodic unit amounts, structure sizes and frequency within training values range can be obtained quickly from RBF model after the RBF has been successfully trained with improved Gaussian algorithm. Finally, RBF model has been approved by FDTD results. It is also showed that RBF model is very effective, which will provide powerful approach for the precise analysis and quick design of microwave filter using PDGS with defected rectangles.