Heuristic Based Real-Time Train Rescheduling System

This article deals with the problem of train rescheduling on a railway network. Starting from a defined network topology and initial timetable, the article considers a dynamic train rescheduling in response to disturbances that have occurred. The train rescheduling problem was mapped into a special case of the job shop scheduling problem and solved by applying a constraint programming approach. To improve the time performance of available constraint programming tool and to satisfy a selected objective function, a combination of three classes of heuristics are proposed: bound heuristics, separation heuristics, and search heuristics. Experimental evaluation of the implemented software in Belgrade railway dispatching area indicates that the proposed approach is capable of providing the support to a real‐life operational railway control. In our solution, the dispatcher has the possibility of choosing the most suitable optimization criterion from the set of seven available ones.