Dynamic Safety Margin in Fault diagnosis and Performance Recovery
The complexity of modern industrial processes makes high dependability an essential demand for reducing production loss, avoiding equipment damage, and increasing human safety. A more dependable system is a system that has the ability to: 1) detect faults as fast as possible 2) diagnose them accurately 3) recover the system to the nominal performance as much as possible. Therefore, a robust Fault Detection and Isolation (FDI) and a Fault Tolerant Control (FTC) system design have attained increased attention during the last decades. This thesis focuses on the design of a robust model-based FDI system and a performance recovery controller based on a new performance index called Dynamic Safety Margin (DSM).
The DSM index is used to measure the distance between a predefined safety boundary in the state space and the system state trajectory as it evolves. The DSM concept, its computation methods, and its relationship to the state constraints are addressed. The DSM can be used in different control system applications some of them are highlighted in this work.
Controller design based on DSM is especially useful for safety-critical systems to maintain a predefined margin of safety during the transient and in the presence of large disturbances. As a result, the application of DSM to controller design and adaptation is discussed in particular for model predictive control (MPC) and PID controller.
Moreover, an FDI scheme based on the analysis of the DSM is proposed. Since it is difficult to isolate different types of faults using a single model, a multi-model approach is employed in this FDI scheme. The proposed FDI scheme is not restricted to a special type of fault.
In some faulty situations, recovering the system performance to the nominal one cannot be fulfilled. As a result, reducing the output performance is necessary in order to increase the system availability. A framework of FTC system is proposed that combines the proposed FDI and the controllers design based on DSM, in particular MPC, with accepted degraded performance in order to generate a reliable FTC system.
The DSM concept and its applications are illustrated using simulation examples. Finally, these applications are implemented in real-time for an experimental two-tank system. The results demonstrate the fruitfulness of the introduced approaches.