Rail-Freight Crew Scheduling with a Genetic Algorithm

Abstract

AbstractThis article presents a novel genetic algorithm designed for the solution of the Crew Scheduling Problem (CSP) in the rail-freight industry. CSP is the task of assigning drivers to a sequence of train trips while ensuring that no driver’s schedule exceeds the permitted working hours, that each driver starts and finishes their day’s work at the same location, and that no train routes are left without a driver. Real-life CSPs are extremely complex due to the large number of trips, opportunities to use other means of transportation, and numerous government regulations and trade union agreements. CSP is usually modelled as a set-covering problem and solved with linear programming methods. However, the sheer volume of data makes the application of conventional techniques computationally expensive, while existing genetic algorithms often struggle to handle the large number of constraints. A genetic algorithm is presented that overcomes these challenges by using an indirect chromosome representation and decoding procedure. Experiments using real schedules on the UK national rail network show that the algorithm provides an effective solution within a faster timeframe than alternative approaches.KeywordsColumn GenerationLinear Programming RelaxationCrew ScheduleChromosome RepresentationRail NetworkThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.