Closed-loop Subspace Identification

Contents

The problem of closed-loop identification has been investigated for over 30 years. Important issues such as identifiability under closed-loop conditions have received attention by many researchers [75, 76, 77, 10]. A number of identification strategies have been developed [11, 10].

Closed-loop identification refers to the identification of process models using the data sampled under feedback control. Correlation between the disturbances entering the process and the input offers a fundamental limitation [28, 29, 11, 30, 10] for utilizing the standard open-loop identification methods with closed-loop data. Several closed-loop parametric model identification methods have been suggested in the literature which require either certain assumptions about the model structure or knowledge of the controller model. The closed-loop identification methods found in the literature are broadly classified into direct, indirect and joint input/output identification methods [29]. See Chapter 2 and references in [78, 29, 79] for a review of the features and limitations of different classical closed-loop identification methods.

Section 4.3 is reprinted from Journal of Process Control, vol 15, Huang, B., S.X. Ding, J. Qin, “Closed-loop Subspace Identification: an Orthogonal Projection Approach”, 53-66, © 2005 Elsevier Ltd., with permission from Elsevier.