Self-Triggered MPC with Decaying Prediction Horizon for Multi-Objective Systems

A computationally efficient Model Predictive Control framework is developed for nonlinear systems characterized by multiple performance metrics and bounded disturbance constraints. The optimization problem in MPC is constructed based on the tube-based method. To reduce the computational complexity of the multiple performance metrics optimization problem, a self-triggered rule is introduced to decrease the frequency of solving the optimization problem. Furthermore, an adjustable prediction horizon reduction parameter is incorporated to mitigate the computational complexity of each optimization problem. Robust stability is ensured by establishing a lower bound on the operational duration for each performance metric, with simulation results demonstrating the efficacy of the proposed approach.