Recursive Algorithm of Subspace Model Identification for a Class of Closed-loop Systems with Slowly Time-varying Parameters

Abstract

This paper concerns subspace model identification of slowly time-varying systems operating in closed-loop. A recursive MOESP-type closed-loop subspace model identification algorithm is developed. The key technique of derivation of the proposed algorithm is the Hessenberg QR algorithm. The exponential forgetting mechanism is introduced into the recursive algorithm in order to track the time variation of the parameters of the systems. A numerical study on a closed-loop identification problem shows the effectiveness of the proposed algorithm.