Recursive subspace system identification for parametric fault detection in nonlinear systems

Authors

Abstract

tThis work addresses the problem of detecting parametric faults in nonlinear dynamic systems by extend-ing an eigenstructure based technique to a nonlinear context. Two local state-space models are updatedonline based on a recursive subspace system identification technique. One of the models relies oninput–output real-time data collected from the plant, while the other is updated using data generated bya neural network predictor, describing the nonlinear plant behaviour in fault-free conditions. Parametricfaults symptoms are generated based on eigenvalues residuals associated with two linear state-spacemodel approximators. The feasibility and effectiveness of the proposed framework are demonstratedthrough two case studies.

Related Project

iCIS - Intelligent Computing in the Internet of Services

Journal

Applied Soft Computing, Vol. 37, pp. 444-455, Elsevier 2015

DOI