Optimization of a Railway Line System with a Genetic Algorithm Approach

In the development of a railway line system, the endless set of options and choices that have to be made to get an optimal result is a major obstacle. Customer attractiveness, revenues and costs have to be considered properly. In addition, there is also a very extensive set of constraints such as the existing customer demand, the infrastructure and the available rolling stock. It turns out in practice that making these choices and trade-offs explicit is impossible. For this reason it always remains unclear whether the engineered line system is the best or not. This paper describes a model, the Line Sytemmodel that develops new lines, optimizes and evaluates the developed line system by using a genetic algorithm. The Line System Model optimizes financial result for the carrier. In the model, changes in revenues and costs are both taken into account. The revenues are a result of modelled passenger growth. The costs are a result of changes in the amount of rolling stock and the number of train kilometers, that have to be delivered to execute the timetable. The model generates sets of solutions (of railway line systems) which are then combined and mutated, so that a new generation is formed. The better score a solution has, the greater the chance that it survives. This process continues until the solution does not improve any further. The model is the result of the further development of a prototype. The current model has been applied in practice, but is still further developed.