Study on the distributed model predictive control for multi-zone buildings in personalized heating

Abstract

To meet demands of smart heating development, the novel dynamic regulation methods of heating systems need to be pursued. This paper presents a model predictive control (MPC) approach for room temperature regulation in multi-zone buildings. Among them, a) For heating systems with nonlinearity and large time delay, a hybrid optimization algorithm consisting of the improved particle swarm optimization (IPSO) algorithm and Newton-Raphson (NR) method in online optimization of MPC for a single zone building is proposed. Compared with the quadratic programming method, it can identify successfully the optimal control vector. b) A distributed MPC approach is presented for multi-zone buildings existing hydraulic coupling. Firstly, Nash-optimization acts on searching the global equilibrium point in the distributed structure, then the hybrid optimization algorithm works in the online optimization of these local MPCs. Compared with the decentralized MPC, the control performance is improved through co-operating and communicating between subsystems via network to share information related to their future behavior. Besides, the global performance index in the distributed MPC is redefined as the weighted sum of energy consumption and thermal comfort, which provides the possibility to eliminate quickly the most unfavorable thermodynamic loop in heating systems and reduce energy consumption.