This paper discusses some of the more practical aspects of combining subspace identification and model predictive control into a combined identification and control scheme. Algorithms for using subspace identified models directly in MPC are presented for the fixed model and adaptive mode cases. Details for implementing MPC with subspace models are given for both stable and integrating linear plants. Finally, MPC results are presented for a non-square integrating plant using a fixed model identified with subspace identification from noiseless data.
|Name||Proceedings of the American Control Conference|
|Publisher||Institute of Electrical and Electronics Engineers|
|Conference||2002 American Control Conference (ACC)|
|Period||8/05/02 → 10/05/02|