User-friendly cross-directional MPC tuning for uncertain multiple-array paper-making processes

摘要

The cross-directional model predictive control (CD-MPC) tuning problem is investigated for multiple-array paper-making process with parametric uncertainty. A systematic spatial tuning method is proposed to tune the penalty matrices such that the harmful high frequency elements in each actuator profile can be eliminated. A multivariable temporal filter is employed to adjust the output references, and a temporal tuning algorithm is developed for the filter parameters to achieve the pre-specified 2 sigma performance indices. The computational time of the proposed methods is reduced via event-based sampling strategy. The effectiveness of the techniques is shown via a CD system from paper-making industry.