Model predictive control of supercritical heat exchanger based on an accurate low-order dynamic model with kalman filter correction

To improve outlet-temperature regulation of supercritical water heat exchangers in supercritical water thermochemical hydrogen production systems, this study develops an accurate low-order dynamic model with Kalman-filter correction and embeds it into a model predictive control framework (ALOD–KF MPC). Unlike conventional applications of linear MPC in which the prediction model is usually fixed and the observer is mainly used for state correction, the proposed framework emphasizes the construction and online updating of a physics-guided low-order prediction model for condition-dependent supercritical heat-transfer dynamics. By combining a low-order predictive model with Kalman-based state correction and receding-horizon optimization, the proposed method enables accurate dynamic prediction and robust real-time control with manageable computational cost and online adaptability under varying operating conditions. A systematic tuning study of the weighting ratio, prediction horizon, and control horizon clarifies their impacts on tracking aggressiveness, actuator effort, and robustness, providing practical guidelines for controller implementation. Comparative simulations across multiple scenarios show disturbance-tolerant temperature regulation with negligible steady-state error and actuator-friendly control actions. The method maintains stable tracking under sensor measurement noise and mass-flow-rate perturbations, and improves tracking performance by up to 57.7% and 34.3% compared with PID and Dynamic Matrix Control (DMC), respectively.