Development of a method for modelling delay propagation in railway networks using epidemiological SIR models

A method has been developed to simulate propagation of train delays in branched railroad ranges using modified epidemiological SIR models. These models take into account the mutual influence of trains with different priorities in the flow. This makes it possible to study the heterogeneous dynamics in the propagation of delays among trains of different priorities. To consider the propagation of the primary delay in space and time, it is proposed to represent the topology of the railway network in the form of an undirected graph with reference to the edge of the graph in the mathematical system of differential equations of the SIR model. This unifies the process of constructing SIR models for each edge (section) of the network graph and reduces the dimension of the problem. To take into account the influence of the “network effect”, it is proposed to determine the transit coefficient for each station of the section. This coefficient helps calculate the number of delayed trains for adjacent sections. To set SIR models, it is proposed to use empirical data on the propagation of the average delay in the standard traffic schedule in the corresponding section. For the sequential solution of SIR models corresponding to interconnected network sections, an algorithm is applied to turn the network graph into a directed tree the root of which is the station where the delay occurs. Tests on modelling the propagation of train delays in the railway network are carried out taking into account the mutual influence of different categories of trains in the flow and the built-in time reserves for the restoration of movement. The obtained simulation results have confirmed the adequacy of the solutions and helped quantify the influence of primary delays and the amount of time reserve in the schedules of trains of various categories on the reliability of the standard train schedule