Preparing Track Geometry Data for Automated Maintenance Planning

Abstract

Research on track quality behavior has been extensively published, and many different approaches have been presented to describe the process of track quality. The research goal of this paper is to form a basis for a data-driven tamping prediction based on track quality analyses over time. A research database containing asset information, executed maintenance tasks, and measuring data of some 4,400 km of track of the Austrian rail network in a time sequence of 16 years is available. The modified standard deviation of vertical track geometry is identified as an ideal track quality indicator for planning and predicting tamping tasks in Austria in the context of this research. Further analyses show that a linear regression function is best suited for describing track quality between two tamping tasks and shows the best accuracy for predicting track quality in the future. An algorithm was developed by means of the linear regression function that enables analyses of track quality behavior over time for long time series and the whole network. This includes track quality before and after tamping tasks as well as deterioration rates. In the future, these basics have to be combined with further technical evaluations to detect an optimal intervention limit. Furthermore, economical and operational considerations must be incorporated to find the optimal tamping strategy under the given conditions.