System identification and model reduction for a single-link flexible manipulator

A new model reduction and updating technique is proposed and applied in modelling of a single-link flexible manipulator. The Observability Range Space Extraction algorithm is used to generate an initial overparameterized state-space model. The identified model is transformed into modal realization. The modal responses of individual modes are evaluated. A new measure is proposed to quantify the contribution of individual modes to the total responses. Using the proposed measure, a reduced order model is obtained by retaining the most significant modes. To improve model accuracy, either the reduced input or output matrix can be recalculated by a proposed method. Several critical issues related to the experimental identification are addressed. Experimental identification results are presented to illustrate the proposed technique. (C) 2001 Academic Press.