“…For general response data, the Observer/Kalman "lter identi"cation algorithm (OKID) can be used and it consists of three steps, i.e., computation of the observer Markov parameters, recovery of the system and observer gain Markov parameters, and realization of a state-space model [1,5]. The Observability Range Space Extraction (ORSE) algorithm [6] is developed by generalizing the Q-Markov Covariance Equivalent Realization (Q-Markov Cover) [7] and the ERA. The ORSE algorithm can obtain a state-space model directly from general input/output data.…”