Abstract
In this paper, a H-infinity fault detection and diagnosis (FDD) scheme for a class of discrete nonlinear system fault using output probability density estimation is presented. Unlike classical FDD problems, the measured output of the system is viewed as a stochastic process and its square root probability density function (PDF) is modeled with B-spline functions, which leads to a deterministic space-time dynamic model including nonlinearities, uncertainties. A weighting mean value is given as an integral function of the square root PDF along space direction, which leads a function only about time and can be used to construct residual signal. Thus, the classical nonlinear filter approach can be used to detect and diagnose the fault in system. A feasible detection criterion is obtained at first, and a new H-infinity adaptive fault diagnosis algorithm is further investigated to estimate the fault. Simulation example is given to demonstrate the effectiveness of the proposed approaches.
Original language | English |
---|---|
Pages (from-to) | 79-84 |
Number of pages | 6 |
Journal | AIP Conference Proceedings |
Volume | 1107 |
DOIs | |
Publication status | Published - 2009 |
Externally published | Yes |
Event | 2nd Mediterranean Conference on Intelligent Systems and Automation, CISA 2009 - Zarzis, Tunisia Duration: 23 Mar 2009 → 25 Mar 2009 |
Keywords
- B-spline expansions
- Fault detection and diagnosis
- Filter design
- Probability density function
ASJC Scopus subject areas
- General Physics and Astronomy