Parallel Track Models from Road Measurements

Abstract

In this paper, coherence function estimation for parallel tracks from road measurements is studied. A two-stage scheme is proposed. In the first stage, the measurements are de-trended and periodic components are removed. An auto regressive moving-average (ARMA) model is fitted in the second stage to the vector-valued residuals derived from the left and right track measurements by using a recently developed spectral estimation algorithm. The ARMA model yields not only the power spectral density (PSD) estimates of the individual tracks, but also the coherence function estimate of the parallel tracks. The proposed model development procedure is tested on the real data from the Long Term Pavement Performance (LTPP) database.