The train delay time prediction is conducive to assisting dispatchers to make the regulation of train diagram, optimize passenger trip-chain, and improve railway network quality and safety. According to the characteristics of delay time distribution, the law of delay propagation and the time limit relationship between adjacent trains, the independent and dependent variables of the train delay prediction model are determined, and the delay prediction model is constructed. The results show that the prediction accuracy of the adjacent arrival and departure delay time prediction model based on Logistic regression can reach more than 59% under 1 minute allowable error and 80% under 2-minute allowable error. By testing the data of the test set, the predicted value is in good agreement with the size, trend and range of the real value, and has a good prediction effect.

High-speed railway;Delay propagation;Delay prediction;Generalized regression;Delay distribution;Principal component analysis