Data driven defect detection for train wheels

Description

In this project, we support the Swiss Südostbahn (SOB) in the ongoing digitalization of its maintenance program by developing modern data analysis methods for evaluating condition-monitoring data from railway wheelsets. The project focuses on transforming available sensor and operational data from the vehicle fleet into meaningful information that can support maintenance planning and decision-making.

In particular, we develop algorithms for condition monitoring, fault detection, and predictive maintenance. These enable maintenance decisions to be based on the actual health condition of the railway wheelsets rather than on fixed maintenance intervals.

The overall objective of the project is twofold. On the one hand, maintenance costs are to be reduced. This can be achieved by reprofiling wheelsets based on their actual condition, thereby reducing maintenance effort. At the same time, the wheelsets can achieve higher mileage and therefore need to be replaced less frequently. Passengers can also benefit from smoother train operation. By supporting a data-driven maintenance process, the project contributes to improving operational efficiency, passenger comfort, and the long-term cost-effectiveness of railway maintenance.

Key data