Event-Triggered Model Predictive Control for Perturbed Nonlinear Systems via a Non-Squared Integrand Objective Function*

This paper studies a dual-mode predictive control scheme based on an event-triggered approach. Firstly, a non-squared integrand objective function is adopted to construct the finite-horizon optimization problem. By analyzing the recursive feasibility on this type of optimization problems and its corresponding system stability, a triggering condition including three criteria is introduced into the considered system and constructed to reduce the solving times on optimization problems. Moreover, prediction horizon is not constant and computed at each triggering instant such that prediction horizon decays as the system operates. By this means, the solving complexity of optimization problems is reduced further. By considering the effect of disturbances, the maximum upper deviation on the actual and predicted trajectories under same control input signals is estimated and Zeno behavior of this event-triggered strategy is excluded. Then, both the feasibility and stability of this control mechanism are verified where disturbance is the only factor that matters. Finally, the correctness and superiority of this control strategy are verified based on a vehicle-damp-spring system.