Computationally Efficient Nonlinear MPC for Discrete System with Disturbances

Nonlinear Model Predictive Controller (NMPC) is intensive in online computation. We propose an efficient formulation for reducing its computational requirements. The proposed algorithm avoids stability-related terminal costs, constraints, and varies the prediction horizon after a simple check. Further, we use a condition based on negative contraction to handle undesirable effects of disturbance on the algorithm. The stability analysis for the proposed algorithm in a Monotonically weighted NMPC framework without stability related constraints is derived. Simulation and experimental validation on benchmark systems illustrate a significant reduction in the average computation time compared to the Monotonically Weighted NMPC without much loss in performance.