Rail Track Modelling by Using Identification and Curve Fitting Techniques

In this paper, a seventeen-degree-of-freedom full-car model of a high speed railway vehicle excited by real track profile is going to be studied. The field measurements are collected by Turkish State Railways (TCDD) on a pre-specified pilot section at a constant forward train speed. First, empirical auto-power spectral densities of the left and the right tracks are estimated on the uniform frequency range by using the Welch-method. The estimated power spectra is matched by generally preferred mathematical track spectrum in the Federal Railroad Administration Standard (FRA). Then, curve fitting techniques in the frequency domain as the two-slope and three-slope approximations are performed on the Welch estimated track spectra. Next, subspace-based identification algorithms are applied to shape to the Welch's rail spectrum for the right and left tracks. Based on all these parametric and non-parametric studies, the simulation studies showed that the effect of the rail track modelling on the performance enhancement of the vehicle is quite large.