A distributed predictive control strategy for a two-route public transport system with a transfer station

Abstract

In the context of transit systems’ management, uncertainty is present in different inputs required for the operator to make dispatch decisions properly. At the operational level, traffic congestion, passenger demand, non-recurrent congestion due to incidents, among multiple possible facts, make these systems very difficult to control without efficient tools or systematic rules to ameliorate the effect on the level of service caused by unexpected randomness in these features of the real timeoperation. One efficient way to deal with such situations is to impose real-time control strategies to set the service frequencies trying to minimize the effects of the different disturbances in the operation of such transit systems. In this sense, the literature is extensive in developing flexible control strategies, depending on the specific features of the problem. The most studied strategy is holding (Eberlein, 2001; Yu and Yang, 2007), in which vehicles are held at specific stations for a certain time, mostly trying to keep the headway between successive buses as close as possible to a predefined value. Moreover, and as a complement to holding-type strategies, rules based on expressing or stop-skipping are formulated in order to somehow speed up vehicles instead of delay them as in holding (Fu et al., 2003). In addition, if the dynamic control of the systems involves synchronization of routes at specific stations in order to allow travelers to transfer from one line to another, the resulting optimization problem becomes more complicated and hard to implement in a real setting (Nesheli and Ceder, 2015; Kim and Schonfeld, 2014).