Gain-scheduled robust fault detection on time-delay stochastic nonlinear systems
MetadataShow full item record
This paper studies the problem of continuous gain-scheduled robust fault detection (RFD) on a class of time-delay stochastic nonlinear systems with partially known jump rates. By means of gradient linearization procedure, stochastic linear models and filter-based residual signal generators are constructed in the vicinity of selected operating states. Furthermore, in order to guarantee the sensitivity to faults and robustness against unknown inputs, an RFD filter (RFDF) is designed for such linear models by first designing H 8 filters that minimize the influences of the disturbances and modeling uncertainties and then a new performance index that increases the sensitivity to faults. Subsequently, a sufficient condition on the existence of RFDF is established in terms of linear matrix inequality techniques. Finally, a continuous gain-scheduled approach is employed to design continuous RFDFs on the entire nonlinear jump system. A simulation example is given to illustrate that the proposed RFDF can detect the faults correctly and shortly after the occurrences. © 2011 IEEE.
Showing items related by title, author, creator and subject.
Habibi, Hamed; Howard, Ian ; Habibi, R. (2018)This paper studies sensor fault detection using a game theoretic approach. Sensor fault detection is considered as change point analysis in the coefficients of a regression model. A new method for detecting faults, referred ...
Chen, F.; Yin, YanYan; Liu, F. (2014)A linear full-order robust fault detection observer is designed to solve the robust fault detection problem of Markovian jump systems with partly unknown transition probabilities. Free-connection weighting matrices are ...
Periasamy, Vijay ; Tade, Moses ; Shao, Zongping (2019)© 2019 Elesevier Monitoring and fault diagnosis can play a major role in improving the reliability of Solid Oxide Fuel Cell (SOFC) systems. Due to complex interactions between the variables, quantitative model based methods ...