Linear Model Predictive Control for Hammerstein System with Unknown Nonlinearities*

A linear model predictive control (MPC) scheme for Hammerstein system with unknown nonlinearities is presented. The Hammerstein model is firstly transformed into a linear system with unknown input, and then a minimum variance unbiased (MVU) filter is used to estimate the state and the unknown input. This approach does not need prior knowledge of the nonlinearities in Hammerstein model, and the nonlinear block is treated as a black box. This enables the design of the MPC system remains in a linear framework. Compared with the existing approaches, the proposed method has the advantage of simple implementation and low online computational cost. It is shown that offset-free control can be guaranteed in the presence of asymptotically constant disturbances. An example is given to demonstrate the effectiveness of the proposed approach.