Reduced-Order Observer Based-Fault Estimation for Markovian Jump Systems With Time-Varying Generally Uncertain Transition Rates

This paper addresses such problem for simultaneous estimations of the actuator fault, output disturbance and sensor fault for a type of Markovian jump systems, which takes on time-varying generally uncertain transition rates. In order to perform estimations for faults, a new reduced-order observer is proposed, of which the gain matrix is designed to fully decouple the unknown input for the system. As for the solution to time-varying generally uncertain transition rates, an offline quantization mechanism is developed, which can help to prove, to a large extent, the observer error system to be stable. The proposed method is able to guarantee the exact estimations of the states and faults in spite of the existence of external disturbances and the time-varying stochastic switching. Furthermore, a numerical example is brought to testify the correctness of the designed fault estimation method.