Stochastic Modeling and Analysis of Automotive Wire Harness Based on Machine Learning and Polynomial Chaos Method

Abstract

This paper proposes a method based on machine learning technique and polynomial chaos (PC) method to model and analyze the stochastic behavior of an automotive wire harness. In this research, we assume that the automotive wire harness is a bundle of wires above a conductor plane, and its behavior can be represented by stochastic transmission line equations. First, the proposed method constructs the regression models related to per-unit-length (p.u.l.) parameters by means of a machine learning technique. Then, the stochastic transmission line equations including the regression models are approximated using orthonormal polynomials through a PC formulation. Since the regression models correlate the geometric and shape parameters of the wires and the p.u.l. parameters, PC expansion coefficients can efficiently be calculated. We adopt three types of regression models and compare them to investigate the performance of the proposed method.