Fault detection and isolation in nonlinear systems: observer and energy-balance based approaches

Due to the increasing complexity of modern technical processes, the most critical issues in the design of an automated system nowadays are safety/reliability, higher performance and cost efficiency. Faults in process components can lead to a considerable reduce of the efficiency of the process, quality of the product and in some cases even result in fatalities. In order to avert these losses, an efficient diagnosis of the faults plays a central role. Therefore, fault diagnosis is becoming an essential part of modern control systems. Fault diagnosis of linear dynamical systems has been extensively studied since decades and well-established techniques exist in the literature. However, fault diagnosis for nonlinear dynamical systems is yet an active field of research. Since most of real systems are nonlinear in nature, classically, linear fault diagnosis techniques have been applied to nonlinear systems based on the linearized system model around an operating point. The drawback of this approach is the limited fault diagnosis performance. In order to fulfill the increasing demand of more effective fault diagnosis systems for nonlinear processes, a lot of attention has been paid to nonlinear fault diagnosis techniques, which is the major topic of this thesis. Different from linear systems, there is no uniform solution for the fault diagnosis of general nonlinear systems. Various schemes have been proposed for nonlinear systems with special structures. Among them, Lipschitz nonlinear systems have been intensively studied, since on one hand more general nonlinear systems can be transformed into Lipschitz nonlinear systems, and on the other hand, many linear fault diagnosis approaches can be extended to this kind of nonlinear systems. For Lipschitz nonlinear systems, observer-based fault detection approach has been mostly applied, which consists of an observer-based residual generator and a residual evaluator. Classically, residual generator and residual evaluator are designed separately. Since the performance of fault detection system is decided by residual generator and evaluator together, it can be expected that, higher fault detection performance can be achieved by designing these two units in an integrated manner instead of separate handling of them. Motivated by this fact, an integrated design approach of observer-based residual generator and evaluator is proposed for Lipschitz nonlinear systems. Besides the schemes extended from linear methods (i.e. observer-based approach, parity space approach etc.), new nonlinear fault diagnosis techniques have also been studied recently, which can be effectively applied to complex nonlinear systems i.e. switched nonlinear systems, hybrid nonlinear systems etc. Among them, new fault diagnosis schemes based on passivity and energy-balance which are closely related to system “energy” have