Model Predictive Control Strategy for CO2 Heat Pump Systems Based on State-Space Modeling

The transcritical CO₂ heat pump system is regarded as a promising technical solution for the integrated thermal management of passenger cabins and power batteries in electric vehicles owing to its high efficiency and environmental friendliness. An efficient system control strategy not only significantly affects the vehicle's driving range， but also enhances cabin thermal comfort. This paper proposed a model predictive control （MPC）-based solution， in which white noise was applied as an excitation signal to a one-dimensional simulation model of a heat pump system to efficiently construct an explicit system model suitable for MPC implementation. Furthermore， a dynamic-feedback-correction-based MPC algorithm （DFC-MPC） was introduced. The proposed method demonstrated significant improvements in system performance： compared to conventional PI control， overshoot was reduced by 76.8% in cooling mode and 80.1% in heating mode， while the system COP （coefficient of performance） was improved by 8.33% and 1.07%， respectively. Moreover， this method enhanced the real-time protection of the compressor， thereby providing a reference for the implementation of intelligent control in real-vehicle thermal management systems.