Research on Dynamic Calibration Technology of Key Parameters of Railway Empty Container Allocation Driven by Data

To realize the accurate allocation of empty railroad containers, it is necessary to know the dynamic calibration of key parameters such as the stacking capacity of container handling stations. This study is based on large-scale container transportation historical data, the use of machine learning methods, the establishment of data-driven prediction model of key parameters of empty railroad container deployment, seasonal decomposition of the number of stacked containers at the handling station, and other information and Prophet model fitting. The trend of the data related to the transfer of empty containers between the stations and the seasonal pattern is derived, so as to achieve the forecast of the number of arrivals at the station in the next five days and dynamically calibrate the capacity parameters such as return capacity within a certain period of the empty containers.