An analytic redundancy technique for automotive power train system is developed to detect and isolate faults in engine speed, torque converter turbine speed, and wheel speed sensors. A set of extended Kalman filters, with different observation combinations and specially formulated gains to maximize robustness, is used to generate residuals, which are then used to trigger sensor faults. This method, of gain selection increases the robustness of the residuals to modeling uncertainties, while preserving the sensitivity to failures. This model based failure detection and isolation method is implemented on a research vehicle; experimental results show highly effective and robust failure detection and isolation.
Analytic redundancy for automotive power train sensor diagnostics
Die analytische Redundanz für die Sensordiagnose bei der Kraftübertragung im Automobilbereich
1991
21 Seiten, 20 Bilder, 31 Quellen
Conference paper
English
F-8 DFBW sensor failure identification using analytic redundancy
Tema Archive | 1977
|