Abstract

Balancing truncation can be used to reduce the order of system models provided that the system is asymptotically stable. Using this technique the states are transformed to a coordinate system in which the controllability and observability Gramians are equal and diagonal, and the states which contribute weakly to the input-output map can be deleted. This paper discusses the use of this technique for unstable system. We present a method that reduce the order of the unstable system model, based on the fact that translation transformation in frequency domain preserves input-output property. Several case studies are examined to illustrate the proposed method. The approach has been successfully applied to model reduction of the flexible robot systems and the results are discussed.