Author(s)

Anqi Shi, Baotian Dong, Fangcan Zhao, Yang Wang

Corresponding Author

Baotian Dong

ABSTRACT

Railway transportation is an important part of China's transportation system. Due to the original business model of railway, its punctuality is poor. In addition, the current calculation method of dwell time of railway freight cars is still be the general old-fashioned algorithm. In order to improve the punctuality of dwell time of cars at the station, this paper takes the terminal as an example, and puts forward a prediction model based on Gradient Boosting Regression Tree. The influencing factors of cars’ dwell time in terminal are characterized. Six influencing factors are selected, i.e. car number, goods type, car type, start station, end station and train number. The six discrete characteristics are quantified. The six characteristics are combined into input vectors, and the data in 2016 and July 2017 are selected for prediction. The experimental results show that the model has a good prediction effect.

KEYWORDS

Gradient Boosting Regression Tree, Railway Freight, Dwell Time