A Two-Level Rolling Horizon Procedure for Railway Predictive Maintenance Scheduling

In this paper, a two-level model for railway maintenance is proposed. The first level consists of an off-line decision support system to schedule railway predictive maintenance activities, in order to intervene when a track segment has reached a certain degradation state, and thus preventing failures. The second level consists of an on-line recovery methodology aiming at scheduling fault diagnosis and fault management activities when a failure occurs, or when the off-line scheduling cannot be met. Therefore, the set of mitigation actions to be executed and the time in which they have to be taken, are determined. The whole approach is based on the risk minimization framework, according to ISO 55000 guidelines, which allows in particular to evaluate the maintenance activities priorities. In this paper, after providing a formal methodology description, some experimental results are reported and some indications about its future developments are given.