On fuzzy modelling of dynamic track behaviour

Abstract Rolling noise is an important source of railway noise and depends also on the dynamic behaviour of a railway track. This is characterized by the point or transfer mobility and the track decay rate, which depend on a number of track parameters. One possible reason for deviations between simulated and measured results for the dynamic track behaviour is the uncertainty of the value of some track parameters used as input for the simulation. This in turn results in an uncertainty in the simulation results. In this contribution, it is proposed to use the general transformation method to assess a uncertainty band for the results. Most relevant input parameters for determining the point input mobility and the track decay rate for a ballasted track are analysed with regard to the uncertainties and for the value of each an interval is determined. Then, the general transformation method is applied to four different simulation methods, working both in the frequency and time domains. For one example track, the resulting uncertainty bands are compared to one dataset with measurements for the point mobility and the track decay rate. In addition, a sensitivity analysis is performed to determine the parameters that significantly influence the overall result. While all four simulation methods produce broad uncertainty bands for the results, none did match the measured results for the point mobility and the track decay rate over the entire frequency range considered. Besides the large influence of the uncertain pad stiffness, it turned out that the rail wear is also a significant source of uncertainty of the results. Overall, it is demonstrated that the proposed approach allows assessing the influence of uncertain input parameters in detail.