:   .., .., .., ..
:   :
:  119
:  
:  2026
:   .., .., .., : // . - 2026. - . 119. - .298-329.
:   , , , ,
(.):  train traffic, delay prediction, Markov chain, machine learning, data class imbalance
:   . , . , . , , . ~, . , . . , . .
(.):  Schedule deviations predicting is an integral part of the operational management of train traffic. It is especially important to forecast the development of the situation when a large delay occurs, when it is necessary to take effective corrective actions. However, since large delays occur relatively rarely, the delay data array has the property of imbalance. The latter negatively affects the accuracy of the forecast, which can be observed when the prediction is built using machine learning theory. This paper is devoted to an approach to predicting that is not associated with machine learning. The article proposes a method for preprocessing train traffic data arrays aimed at improving the accuracy and speeding up the forecasting algorithms. Calculations of predicted delays are carried out using a Markov model for a delay chain. In this case, databases of commuter trains on Russian Railways and high-speed trains on Chinese railways are used. The study confirms the prospects of using the proposed method for calculating delays in the dispatcher decision support system.

PDF

: 45, : 18, : 8.


© 2007.