Abstract
In this paper, an offset compensation method for a continuous-time model predictive control (MPC) is proposed. To achieve offset-free control, additional disturbance states are introduced and the target for MPC is modified based on the estimate of disturbances. The property of steady-state deviation depends on the dimension of the additional disturbance states. An algorithm has already been proposed for discrete-time MPC to achieve zero offset when the model is augmented by disturbance states of the same dimension as tracked variables. We extend this algorithm to continuous-time MPC. A simulation result with the proposed algorithm for a nonlinear tank system is shown.
