The Elevator dispatching Problem: Hybrid System Modeling and receding horizon control

Abstract Elevator dispatching is a combinatorially hard problem with many control constraints, time-varying traffic patterns, partial state information, and random effects. We develop a hybrid system model of a building with multiple elevator cars and apply a receding horizon control approach, bypassing some of the problem's complexities. Thus, we obtain a “universal” dispatcher which is robust with respect to changing traffic patterns and avoids the problem of having to switch among different controllers when these patterns change over the course of a day (as currently done). Moreover, simulation results show that the performance of this approach improves upon that of state-of-the-art dispatchers.

[1]  Matthew Brand,et al.  Optimal parking in group elevator control , 2004, IEEE International Conference on Robotics and Automation, 2004. Proceedings. ICRA '04. 2004.

[2]  D. Mayne,et al.  Receding horizon control of nonlinear systems , 1990 .

[3]  Andrew G. Barto,et al.  Improving Elevator Performance Using Reinforcement Learning , 1995, NIPS.

[4]  Christos G. Cassandras,et al.  Optimal dispatching control for elevator systems during uppeak traffic , 1997, IEEE Trans. Control. Syst. Technol..

[5]  Matthew Brand,et al.  Marginalizing Out Future Passengers in Group Elevator Control , 2003, UAI.

[6]  Christos G. Cassandras,et al.  A Cooperative receding horizon controller for multivehicle uncertain environments , 2006, IEEE Transactions on Automatic Control.