Bilinear System Robust Fault Diagnosis And Analysis

Fault diagnosis is an important research area in systems and controls. Fault diagnosis has been successfully applied into many engineering fields, such as aviation and spaceflight, nuclear reactor, thermoelectricity plant, mechanical systems and chemical systems. The basic principle of fault diagnosis is evaluating system operation through a serial of resolution, calculation and analysis, which are based on system inputs and outputs. This paper summarizes the development of fault detection and isolation (FDI) in the world and discusses the model-based approach to fault diagnosis. The model-based approach includes parity space method, observer-based method and parameter estimation method etc. All of these methods have fine performance on fault diagnosis of linear system. Since the real systems have character of nonlinear or are linear, nonlinear system fault diagnosis is of important meaning. Fault diagnosis of nonlinear system is more complex than that of linear system. In this dissertation, fault detection of bilinear system is mainly explored. A bilinear fault detection filter (BFDF) is proposed for a bilinear system with unknown inputs. Sufficient conditions are given for the existence of a BFDF and the design procedure of BFDF is presented. At the same time, bilinear fault detection observer method is compared to the method of BFDF. The structure of BFDF is very simple and the design procedure is explicit. Residuals generated by BFDF are robust with respect to the unknown inputs and sensitive to both actuator and component faults and sensor faults. Residuals have different directions for different faults, which are useful for fault isolation. A typical bilinear hydraulic system is used for simulation purpose and the result demonstrates the validity of BFDF.