Online failure prediction and maintenance planning for railway networks


Staff Mentor:

prof.dr.ir. B. De Schutter (Bart)


Other Mentor(s):

K. Verbert, R. Babuska

Keywords:

Machine learning; Railway networks

Description:

To minimize costs and maximize the availability of the rail infrastructure, an adequate maintenance planning is of great importance. To optimize maintenance planning, insight into the system degradation is needed, so that this insight can be used to organize and prioritize maintenance. For this purpose, various monitoring systems have been developed and installed. An important question is how to use this information to improve maintenance planning. The aim of the research is therefore to develop and validate methods to predict the system degradation behavior and the corresponding moment of functional system failure (i.e. the moment that the system is no longer able to perform its task). For this purpose, we intend to use both system knowledge (quantitative or qualitative models) and monitoring data. The method will be applied to railway track circuits or switches.

MSc research topics:
Ideally, we would like to make the decision regarding required maintenance as early as possible, so that it can be easily incorporated in the maintenance schedule. At the same time, the prediction quality is generally poor at this early stage, so we want to postpone the decision to a later moment when we are able to make a more reliable decision. Therefore, a trade-off has to be made. The idea is to make predictions as early as possible and to update them each time new information comes available. Based on the prediction outcome and the consistency between predictions, the ``optimal'' decision (moment) can be determined. Relevant issues to be addressed are: How to predict the moment of failure? How to update the prediction each time new information comes available? How to represent uncertainty?



Time-to-failure prediction

© Copyright Delft Center for Systems and Control, Delft University of Technology, 2017.