Robust nonlinear model predictive control via approximate value function

In order to improve the performance of nonlinear model predictive control (NMPC) in the presence of disturbances or model uncertainties, an approximate dynamic programming (ADP) control scheme is proposed. Namely, the Bellman's optimality principle is employed to determine the input based on the approximate value function constructed from the historical operation data. In addition, the support vector data description is also applied in the state space to determine if the ADP control is suitable for the current state. The proposed control strategy is illustrated on a CSTR example to show its effectiveness.