Genetic and memetic algorithms for scheduling railway maintenance activities

Nowadays railway companies are confronted with high infrastructure maintenance costs. Therefore good strategies are needed to carry out these maintenance activities in a most cost effective way. In this paper we solve the preventive maintenance scheduling problem (PMSP) using genetic algorithms, memetic algorithms and a two-phase heuristic based on opportunities. The aim of the PMSP is to schedule the (short) routine activities and (long) unique projects for one link in the rail network for a certain planning period such that the overall cost is minimized. To reduce costs and inconvenience for the travellers and operators, these maintenance works are clustered as much as possible in the same time period. The performance of the algorithms presented in this paper are compared with the performance of the methods from an earlier work, Budai et al. (2006), using some randomly generated instances.