An ANN-based analysis model for ACPWs with finite dimensions

Abstract

An analysis model based on artificial neural networks (ANNs) is proposed for computing the effective permittivities and characteristic impedances of asymmetric coplanar waveguides (ACPWs) with finite dielectric thickness and ground-plane width. First, the ACPWs are analyzed with the conformal mapping technique (CMT) to obtain the training data sets. Then, the Levenberg-Marquardt (LM) algorithm is used to train the ANNs for obtaining an accurate analysis model. The models also can be used for symmetric coplanar waveguides. At last, the models are validated by the comparison with the CMT analysis and measured results.