Präzise Vorhersage von Weichenstörungen

Abstract

By an accurate forecast of the remaining useful life (RUL) appropriate maintenance actions can be carried out before a turnout breaks down and produces expensive delays. Based on the current research activities at the DLR (German Aerospace Center) the present article shows how a data driven forecast in the case of turnouts is carried out. It outlines how control current measurements of the drive, climate data and design characteristics of turnouts are processed to provide an RUL with the help of data mining procedures. Apart from numerous correct forecasts, many less accurate forecasts were produced. This shows that the approach is promising but that it still has to mature significantly. It is above all the quality of data as well as the extracted information contents of the measurement that must be increased in order to improve the forecasts.