Online Fault Detection and Diagnosis of Complex Systems Based on Hybrid Component Models

Up to now model-based online fault detection and diagnosis is rarely applied in process automation and chemical industries. The main reason is the big effort, which is necessary to develop a comprehensive model for a technical system under various circumstances. However the growing complexity of plants and facilities requires increasingly the use of formal methods to analyse and monitor the system behaviour. In this paper a fault detection and diagnosis method based on qualitative models and combined with dynamic models is proposed. The method is component-oriented. A basic feature of the concept is its ability to build automatically clusters of qualitative and dynamic components, which can be reused as single components. An application example of a three-tank-system shows that such kind of models, the so called hybrid models, are capable of solving fault detections and diagnosis problems.