State estimation-based distributed model predictive control of large-scale networked systems with communication delays

In this paper, a cooperative distributed model predictive control is proposed for a class of large-scale systems composed of discrete-time linear subsystems which are coupled via states. The subsystems communicate with each other through a network with communication delays and they do not have direct access to their local states. In the proposed scheme, each subsystem is associated with a local MPC unit, a local predictor and a local observer. The local model predictive controllers exchange predicted input sequences via a delayed communication network. The local predictor uses the explicit model of the subsystems and predicts those states which are coupled between subsystems and their actual values are not available due to delays. The states of each subsystem are estimated from local measurements by designing a local observer. Convergence analysis of the proposed estimation methodology is performed and it is shown that the observer and predictor have exponential convergence. Furthermore, the exponential stability of the closed-loop system is existed. Finally, the theoretical results are verified by a simulation study.