Application of Independent Component Analysis for Road Profile Estimation

Abstract

An application method of Independent Component Analysis (ICA) on vibration of an automobile considering dynamical properties is introduced and its performance is examined through a numerical simulation using a half car model. ICA is categorized in blind source separation method, and originally invented to separate significant signals buried in mixed signals. The proposed method separates road irregularity profile from mixed signals; data set of measured accelerations and calculated double integral of accelerations (displacements), integral of accelerations (velocities), differential accelerations (jerk) and double differential of accelerations. The advantage of proposed method is considering dynamical properties, while original ICA assumes a static mixture. So, it is effective on analysis of source separation in mechanical system including springs and dampers. Also, methods for fixing order of independent components and amplitude are proposed for the estimation of road irregularity. By applying the proposed method, road irregularity with information of amplitude is estimated without knowing parameters such as mass and spring constant.