Research On Fault Diganosis For Time-delay Systems

This dissertation studies the problems on fault diagnosis for systems with delayed mesurement to solve a class of usual actuator and/or sensor faults. At firstly, the history of the development on the fault diagnosis is reviewed. The latest research tendency and the main methods are summarized in the fault diagnosis domain, and the fault diagnosis approaches are introduced for time-delay systems. Then, by using optimal control theory, duality principle, linear matrix inequation, and state observer and so on technology, fault diagnosis approaches are proposed for systems with delayed mesurement. The major results of this dissertation are summarized as follows.1. On a class of linear system with the fault model is described, and the model is established with the concept of "exosystem" and the state space expression for this class of faults with known dynamic characteristics and unknown initial time and initial state. We regard the fault states as the sub-states of an augmented system, the fault diagnosis problem can be converted to state observation. Then according to the different eigenvalues of system actuator failures and sensor failures, as well as two different situations fault model, according to the PBH eigenvector criteria of observability, we obtain the fault can be diagnosed criterion appropriate in different situations.2. The problems of fault diagnosis for systems with delayed measurements are considered. A functional transformation is first presented, which transforms the system with delayed measurements into a system without delay formally. Then the fault states are transformed into the states of an augmented systerm, and design of the observer-based diagnoser is proposed. Finally, by using the Lyapunov stability theory and liner matrix inequality(LMI) technology, the real-time fault diagnosis is realized on-line.3. Studies the problem of fault diagnosis for systems with delayed measurements when there exist sensor and actuator faults. The main contribution consists in the delay-free transformation approach for measurement delays and in the design of a new optimal fault diagnoser. Firstly, we regard the fault states as the sub-states of an augmented system, and a delay-free transformation is proposed to transform the system with delayed measurement vector into a delay-free one. Then a sufficient condition of the observability is proved for the transformed system. By using duality principle, the design problem of the optimal fault diagnoser is transformed into an equivalent optimal control problem for the duality system of the fault diagnosis error equation. Finally, the real-time fault diagnosis is realized by constructing a new reduced-order optimal fault diagnoser.4. The conclusions are made.And the direction for the future study is indicated.