Active Disturbance Estimation Embedded Nonlinear MPC Design with Application to Liquid Level Regulation

This study introduces an innovative Disturbance Utilization Nonlinear Model Predictive Control (DUNMPC) framework aimed at enhancing the robustness and efficiency of model predictive control for nonlinear systems subjected to state and input constraints, as well as external disturbances. By incorporating disturbance estimation into the rolling-horizon optimization process, the approach extends traditional feedback control strategies. At the core of this framework is the development of a real-time disturbance observer that accurately estimates disturbances, which are then integrated into the predictive model. Additionally, we enhance the control framework by introducing tightened and terminal constraints based on disturbance estimations. This modification not only accounts for the inherent limitations of the physical system but also ensures robust constraint adherence and sustained feasibility. A comprehensive stability analysis of the closed-loop system demonstrates the DUNMPC framework’s resilience and reliability. To validate our approach, we implement the DUNMPC in a practical nonlinear water-level control scenario. Experimental results indicate significant improvements in control performance, confirming the framework’s effectiveness in managing diverse disturbances and constraints.