Fuzzy-Tuned Model Predictive Control with Extended State Observer for Refrigeration Systems: A Hardware-in-the-Loop Approach

Optimizing the trade-off between temperature-tracking precision and energy efficiency remains a significant challenge in industrial refrigeration systems. To address this, this paper presents a novel hierarchical control architecture combining Model Predictive Control (MPC), Fuzzy Logic Controller (FLC), and an Extended State Observer (ESO). Specifically, the MPC manages the system’s physical constraints, while the FLC dynamically tunes the objective function weights online, ensuring an optimal balance between performance and energy savings. Furthermore, the ESO is employed to estimate and actively compensate for exogenous heat load disturbances and model uncertainties. Comparative results confirm that the proposed strategy not only reduces energy consumption by 10.93% but also achieves highly disturbance rejection when compared to conventional PI control. The practical feasibility of the proposed algorithm is rigorously validated via hardware-in-the-loop (HIL) simulations utilizing an STM32F767ZI microcontroller. The successful hardware-in-the-loop validation on an embedded microcontroller demonstrates the industrial viability of the proposed architecture, proving it to be a highly deployable and cost-effective solution for refrigeration system.