Characterization of S-Parameters for Nonuniform Microstrip Lines With Tabbed Routing Using Analytical-Numerical Method and Machine Learning

Abstract

In this article, a model for characterizing the S-parameters of periodic nonuniform microstrip lines with tabbed routing is proposed, which can calculate the corresponding scattering parameters quickly and accurately from the physical parameters. The model employs an equation-based analytical (EBA) solution, utilizing a piecewise cascade methodology and integrating numerical analysis from the finite difference time domain (FDTD) method within symmetrical repeated regions. This approach mitigates errors resulting from inadequate quasistatic conditions, thereby improving accuracy. Furthermore, the model integrates partial least squares (PLS) machine learning (ML) for correction. It leverages discrepancies between calculated and electromagnetic simulation results as learning inputs to predict model deviations under new structural parameters, thereby enhancing characterization precision. Validation against full-wave simulations confirms the model’s strong alignment and superior accuracy over the EBA solution, without compromising computational efficiency during numerical solution and model training. Moreover, the model’s accuracy is confirmed through comprehensive board fabrication and measurement experiments.