Model predictive control for the ice-storage air-conditioning system coupled with multi-objective optimization

Abstract

Optimal operation of ice-storage air conditioning (IAC) system is beneficial to balance the power grid pressure, enhance load flexibility and reduce system operating costs. Conventional control methods, like fixed scheduling and storage priority, are insufficient for dynamically regulating the IAC system in response to real-time variations in building load. This results in low system operation efficiency and the inability to fully utilize the ice storage device to achieve economic and energy-saving benefits. This study proposes a novel method coupling multi-objective optimization with model predictive control (MPC), which can effectively determine the optimal operation strategy and achieve precise control of IAC system. At the strategy level, a multi-objective global optimization model is established, considering both cost and energy efficiency. This model enables the determination of the optimal load distribution of the chiller and ice storage, which serves as the reference trajectory for MPC. At the control level, an MPC controller is established to achieve accurate tracking of the reference trajectory. This controller uses an artificial neural network as the prediction model and employs the particle swarm algorithm as the optimization solver. It also incorporates continuous feedback correction based on load changes. The results indicate that the optimal operation strategy obtained using multi-objective optimization can theoretically reduce energy consumption by 25% and operating cost by 20.9% compared with the storage priority control. The MPC can exactly track the reference trajectory of the ideal operation strategy, with an energy consumption error of 0.2% and an operating cost error of 0.4%, which precisely regulates the optimal operation strategy and generally improves the economic and energy saving benefits of the IAC system.