RBF-ARX model-based two-stage scheduling RPC for dynamic systems with bounded disturbance

With directly considering the unknown and bounded disturbance, a RBF-ARX model-based two-stage scheduling quasi-min–max robust predictive control (RBF-ARX-TRPC) algorithm for output-tracking control is proposed for a class of smooth nonlinear systems with unknown steady-state knowledge. Firstly, from the RBF-ARX model that is identified using input/output data of the system, the two local linearization state-space models that consider the bounded disturbance and a polytopic uncertain LPV state-space model are built to approximate the present and future system’s nonlinear dynamics, respectively. Based on the state-space models, the RBF-ARX-TRPC algorithm is designed without relying on the system steady-state knowledge. In the RBF-ARX-TRPC algorithm, the future nonlinear behavior of the system is forced to vary within the region constructed by the polytopic uncertain LPV state-space model. Closed-loop stability is guaranteed when the algorithm is implemented in a receding horizon fashion by including a Lyapunov constraint in the formulation. The comparative experiments demonstrate the effectiveness of the proposed strategy on a continuously stirred tank reactor (CSTR) simulator.