This article proposes an integrated model predictive control (MPC) framework with disturbance preview information for nonlinear systems. It is assumed that the disturbance can be previewed within the prediction horizon but unknown outside the horizon. First an integrated terminal control law consisting of both feedback and feedforward is considered. Based on that, a procedure is presented to calculate the associated terminal constraints and terminal cost. A new MPC formulation is then presented with these terminal elements and it is shown that stability and recursive feasibility can be guaranteed under the proposed design using the input-to-state stability tool. Another distinctive feature of the proposed MPC scheme is that the disturbance and the reference information in the horizon is integrated in online optimization, rather than treating disturbance rejection and trajectory following separately, which makes it possible to make full use of the predictable disturbance if it is beneficial to the control task. Numerical examples show that this integrated MPC yields a larger stability region and better performance under prescribed disturbance in comparison with the existing MPC algorithms with disturbance preview.
K E Y W O R D Sdisturbance preview, feed forward and feedback, input-to-state stability, model predictive control, recursive feasibility, terminal constraint
INTRODUCTIONAmong advanced control strategies, model predictive control (MPC) has been receiving much attention in the last decades. One distinctive feature of MPC is the capability to deal with input and state constraints explicitly. [1][2][3] By taking advantage of ever increasingly available computing power, MPC attempts to achieve optimal performance through solving an online finite-horizon optimization problem at each sampling instant. Nevertheless, the presence of disturbance may significantly degrade its control performance, and even render the MPC algorithm infeasible and unstable. There are specific concerns about the presence of disturbance/uncertainty in the MPC setting including constraint satisfaction and recursive feasibility of the on-line optimization. This triggers extensive research on MPC under disturbance and generates rich results in this field; for example, inheritThis is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.