A new fault detection observer scheme for T-S fuzzy systems with unmeasurable variables

Abstract

This paper deals with the sensor fault detection problem for T-S fuzzy systems with unmeasurable premise variables. A new observer is constructed to detect the faults, where measurable premise variables and the estimations of unmeasurable premise variables are used as the premise variables of the observer simultaneously. Different from the existing results, where all premise variables of the observer are obtained by the estimations of the premise variables in the system, the proposed scheme can exploit the partly measurable premise variables better for a good fault detection performance. The residual, generated by the proposed observer, can be robust to disturbance and sensitive to the faults by the design with a mixed H ∞ /H − performance index, and the sufficient conditions for existence of such observer are expressed in term of LMIs. An example is given to illustrate the effectiveness of the proposed method.