Fault Detection and Estimation based on Closed-loop Subspace Identification for Linear Parameter Varying Systems

This paper presents a data driven solution of the Fault Identification approach Connected to Subspace Identification (FICSI) for Linear Parameter Varying (LPV) systems. The proposed solution links system identification to fault detection and estimation in affine LPV systems. As an extension of the model-based FICSI-LPV, the data driven solution is also formulated based on the affine LPV model structure. Recently developed closed-loop LPV subspace identification method can be applied in estimating the Markov parameters required by the data-driven FICSILPV. The batchwise identified Markov parameters can then be used to create the output observer of an LPV system, instead of using the Markov parameters computed from its statespace realization. Furthermore, a contribution of this paper is the verification of the elaborated method on a real-life setup, a DC motor with an unbalanced disc, in pseudo realtime environment.