Data mining models for computing the characteristic impedances of elliptic and circular shaped microshield lines

Abstract

In this work, data mining techniques (DMT) are proposed to be used in regression, in the other words in modeling. As an application example, the commonly used data mining techniques such as Multilayer Perceptron (MLP), M5 model tree, and Reptree are employed to model the characteristic impedances of elliptic and circular-shaped microshield lines as a function of the geometry and substrate parameters. The training and validation data are provided from the analytical formulae. Thus the black- box models are achieved for these types of wave guiding systems where their characteristic empedances are evaluated through data mining techniques to be used in the analysis and design of the microwave circuits. It has been shown that DMT have a high potential of faithful and efficient lines modeling.