GENERALISATION OF ARTIFICIAL NEURAL NETWORK SUPPORTING PLATFORM TRACK ASSIGNMENT WITHIN RAILWAY STATION SIMULATION

The paper deals with the problem of a decision-making support (realised with the help of an artificial neural network) related to the platform track assignment problem. The mentioned problem is supposed to be solved in simulation models reflecting the railway traffic within passenger stations. The utilisation of two-layered perceptron (as a special kind of an artificial neural network) has been studied during the last period. The standing methodology expects that for each input station direction one trained neural network is available. The presented modified approach pays attention to the construction of just one generalised network, which has a potential to be exploited as a universal decision-making support (related to the platform track assignment) for the trains approaching the station from an arbitrary input station direction.