Deep Learning Model for Capacitance and Inductance Matrices Prediction of Multiconductor Transmission Line

This paper presents a novel deep learning-based approach for estimating the per-unit-length capacitance and inductance matrices ([C], [L]) of a multi-conductor transmission line. These matrices, key-parameters (with per unit length resistance and conductance matrices [R] and [G]) in the transmission line theory, are derived from the geometrical and electrical description of the cross section of the harness. Conventional methods typically involve either analytical models for very few simple configurations or numerical simulations to solve governing equations, as the Laplace equation related to the p.u.l parameters. In contrast, this study utilizes deep learning models to directly estimate these parameters, particularly focusing on the capacitance and inductance matrices. The effectiveness of the proposed method is compared to conventional numerical analysis, and strategies for constructing an efficient neural network for LC matrices calculations are discussed.