Observer Design And Fault Detection For Generalized Dynamical Systems

In this paper, the design of observers and their applications in fault detection andisolation (FDI) are investigated in detail. Firstly, a variable-structure observer isdiscussed and a FDI method is illustrated for generalized dynamical systems. Then, aclass of function observers is characterized and its applications in observer-basedcontrol and fault detection are analyzed. How to detect and isolate the system faults has been of considerable interestduring the past two decades. The research is still under the way in developing moreeffective solutions to FDI in automatic control systems. The purpose of the faultdetection is to determine if a fault has occurred in the system, whereas the faultisolation is to determine the location of the detected fault. Based on the analyses insystem structure and input and output information, it is important to detect and isolatethe fault signals for fault-tolerant control decision. Observer-based FDI approach is one of the most important methods in systemdesign. The error between the actual output and its estimation is utilized as the faultindicator. The sliding mode observer is insensitive to unknown parameter variationand noises. Thus the sliding mode observer-based FDI has nice robustness and hasbecome a kind of powerful design approaches. The function observerparameterization can be utilized to solve optimal H2 H∞ filtering problems, robustobserver design problem and FDI problem. By selecting the free parameterQ(s)∈RH∞ , the fault signal can be exhibited properly via the residual signal when afault occurs in a system. Numerical examples are given to illustrate the designprocedure. The simulation results show satisfactory performance.