A Shift-Daily Planning of Freight Trains Composition and Transit Using Maximum Network Flow Approach

Abstract

Automation of shift-daily planning for the formation and passage of trains is carried out by searching for the optimal flow on a specially designed network graph. The nodes are events on the train schedule plane – in distance-time coordinates. The resource flow units in the graph are cars. The graph edges are the phases of the resource life cycle: such as carriage idling at a station, movement of a carriage as part of a train, etc. The optimization is carried out by the Goldberg preflow-push method. The algorithm is modified to work with distinguishable flow units, adapted for the task of forming and disbanding trains from cars at stations. Various scenarios of the local train environment of trains and methods of their processing are considered. Successful testing of the algorithm on the model of the Trans-Baikal road is performed.