Data-driven model predictive control of transcritical CO₂ systems for cabin thermal management in cooling mode

The transcritical CO₂ cabin thermal management system has gained significant attention in the field of electric vehicles due to its outstanding heating performance and environmental advantages. However, ensuring its optimal operation in real-time during vehicle operation poses a challenge. Amongst these challenges, controlling the optimal discharge pressure is particularly difficult. In this paper, we propose a novel model predictive controller that focuses on the cabin cooling mode. The controller utilizes a high-fidelity data-driven dynamic model of the transcritical CO₂ system, coupled with a dynamic thermal model of the cabin. By simultaneously controlling the compressor, electronic expansion valve, and indoor fan, the proposed controller enables the cabin thermal management system to operate in real-time at the optimal discharge pressure while ensuring passenger comfort, thereby minimizing the total power consumption of the system. Additionally, two model predictive control strategies, focused on comfort and energy-saving, respectively, are introduced. Through simulations under various conditions over a 6-hour period, comparing the PI controller, the comfort priority model predictive controller reduces energy consumption by 13.33%, and the energy-saving priority model predictive controller achieves a 20.27% reduction. The proposed novel model predictive controller exhibits energy-saving advantages.