An optimal delay management algorithm from passengers’ viewpoints considering the whole railway network

Abstract

We propose an algorithm for optimal delay management. Delay management means a decision whether or not to keep a connection between trains and is made by dispatchers when trains are delayed. We set dissatisfaction of all passengers in the whole railway network as a criterion and develop an algorithm which seeks for a delay management plan which minimizes passengers’ dissatisfaction. The algorithm is a combination of simulation and optimization. The simulation part consists of a train traffic simulator and a passenger flow simulator which work in parallel. The train traffic simulator forecasts future train diagrams considering the dynamic interaction between trains and passengers. The passenger flow simulator traces behaviour of all the passengers one by one and calculates how many passengers get on/off at each station. This information is given to the train traffic simulator and necessary dwell times are calculated. Passengers’ dissatisfaction is also estimated from the results of the passenger flow simulation. In the optimization part, we use the tabu search algorithm. We will show the details of our algorithm together with numerical results using real world data.