Research On Fault Detection For A Class Of Generalized Dynamical Systems

Fault detection and isolation (FDI) technique to control systems has received more and more attention in the recent years. Fault detection and isolation is the basis of fault-tolerant control systems. For a complex system, the high requirement in reliability is necessary. There are many approaches of fault diagnoses, including diagnosis filters, parity space methods, parameter identification methods, and artificial intelligent approaches etc. In this paper, observer-based (filter-based) fault detection and diagnosis is investigated for systems with modeling errors and input disturbances. The robust fault diagnosis problem is considered. The fault detection to generalized dynamical systems is discussed and the residual generator is constructed. The causal observer is designed for generalized continuous systems with unknown input uncertainties, and the fault signal can be estimated directly. Furthermore, the discrete causal observer is studied for discrete generalized dynamical systems. The mathematical proofs are given in detail. Finally, some numerical examples or practical examples are given to illustrate the design procedures and applications. The simulation results show the satisfactory tracking performance and detecting performance. The proposed causal observers can determine the existence and the place of unknown input faults. Therefore the present fault diagnosis approach has potentials in practical applications.