Abstract

Metros are the principal means of public transportation in many of the world's cities. Expanding metro infrastructure is costly, however, and at a certain point becomes unsustainable. When this occurs, the only feasible solution is to improve the management system. This can be done using either offline or online approaches. In this work, a distributed predictive control approach for multi-line metro systems is proposed. The metro system is partitioned into sub-systems each of them associated to a local predictive controller in charge of local optimization. Performance comparisons of the proposed distributed scheme with a fully decentralized and a simple control approach are presented, using average passengers wait time and train's occupation as performance criteria. Results show that the distributed model predictive control scheme outperforms decentralized and simple control approaches. © 2011 IEEE.