In urban rail transit, the existing methods used for arrival time prediction have low accuracy. A high-precision predictive method for train arrival times can reduce the workload of train drivers by providing the predictive terminal arrival time and satisfy the requirement of headway control model. This paper proposes the predictive algorithms for train arrival times using back-propagation (BP) neural network, wavelet neural network and genetic algorithm. These algorithms include two parts: running time prediction for sections between two stations and dwell time prediction for stations. The real data on train operation are used for training and testing the predictive algorithms. The feasibility of these algorithms is validated. A comparison between these algorithms is given. The results prove that these prediction algorithms for train arrival times achieve high accuracy.
Prediction algorithms for train arrival time in urban rail transit
2017-10-01
506968 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Urban rail transit train pre-arrival time calculation method and device and medium
Europäisches Patentamt | 2023
|