An efficient model nonlinear predictive control algorithm based on stair-like control strategy

By substituting reference trajectory for state in prediction horizon, and using sequential one-step predictions with stair-like control strategy instead of a multi-step prediction, an efficient model predictive control algorithm for affine nonlinear systems was proposed. This algorithm resulted in an analytic model predictive control law with slight computational load. Robustness to model mismatch was also guaranteed by feedback compensation. The efficiency of the algorithm was illustrated by simulative and experimental control results of a water-tank control system.