Fault-Tolerant Tube-Based Robust Nonlinear Model Predictive Control

This paper proposes a nonlinear model predictive control algorithm that is robust to bounded disturbances and tolerant to a finite number of faults. The approach ensures that there are feasible trajectories to a safe state given any fault occurring at any future time. The controller chooses the safe state, which enlarges the feasible region relative to prescribing fixed safe regions as a part of the control design. A numerical example demonstrates the efficacy of the proposed control approach and shows the benefits of the safe state being a decision variable for every possible time of fault occurrence.