Sliding mode based robust disturbance observer design

Abstract

The presence of external disturbances, model uncertainties and nonlinear effects are inevitable in almost all control system. The unexpected effects such as undesired fault or model uncertainty may destroy and corrupt the automated system. For such reasons, a design of robust disturbance observer has received considerable attention and various schemes have been suggested. The robust disturbance observer methods may not only satisfy increasing demands for highly reliable control system but also apply to FDI (fault detection and isolation) or parameter identification. This dissertation mainly deals with i) a simultaneous state-disturbance observation method for the system with uncertainties and disturbances based on adaptive sliding mode observer, and ii) an input/ output disturbance and state observation method for the nonlinear uncertain system with input and output disturbances based on sliding mode descriptor observer, and iii) a parameter identification method for the system with system parameter variation and external disturbance based on sliding mode scheme and recursive least squares method. At first, novel analysis results are introduced for two phase sequential attractions phenomenon in the reaching phase of the robust adaptive sliding mode observer. It is shown that without the knowledge on the uncertain parameters, the observer has two sequential phase attractions to reach the sliding surface, which are estimation phase and reaching phase, even if the sliding function is bounded regardless of the existence of the estimation phase. And then, the robust sliding mode observer for linear system with structured uncertainty and external disturbance is constructed. The proposed method, which consists of a sliding mode observer and a robust adaptive observer, is designed to estimate the system state and external disturbance simultaneously. The effects of structured uncertainty of the system, especially matched parts of the uncertainty, on the estimation erro...