Subspace Identification Methods for Closed Loop Input-Output Data

So called subspace methods for direct identication of linear mod els in state space form have drawn considerable interest recently They have been found to work well in many cases but have one drawback they do not yield consistent estimates for data collected under out put feedback This contribution points to the reasons for this and also shows how to modify the basic algorithm to handle closed loop data We stress how the basic idea is to focus on the estimation of the statevariable candidates the kstep ahead output predictors By re computing these from a nonparametric or rather high order ARX onestep ahead predictor model closed loop data can be handle