Actuator Fault Detection based on Combination of Reduced-order Observer and Zonotope Method

Abstract:

For the discrete-time system with actuator fault， and the state equation and output equation are affected by external disturbance， the reduced-order observer design method and interval estimation method based on the zonotope theory are studied， and an interval-observer-based actuator fault detection method is proposed. First， the original system is decomposed into two reduced-order subsystems by an appropriate state equivalent transformation， so that one subsystem can directly decouple the disturbance. Then， under the assumption that system has no actuator fault， a reduced-order observer design method is proposed for the subsystem without disturbance to obtain the asymptotic convergence state estimation of the original system. Next， a Luenberger observer is designed for another subsystem， and by combining it with the zonotope theory， an interval estimation method for measurable output is proposed. Afterwards， based on the interval estimation of the measurable output， a new actuator fault detection method is proposed by constructing a new residual. Finally， a simulation example is given and analyzed to verify the effectiveness of the method.

Key words:actuator fault;fault detection;reduced-order observer;interval estimation;zonotope method;residual construction