We use a system model expressed in a subset of the Unified Modeling Language to perform fault isolation in large object oriented control systems. Due to the severity of the failures considered and the safety critical nature of the system we cannot perform fault isolation online. Thus, we perform post mortem fault isolation which has implications in terms of the information available; the temporal order in the error log can not be trusted. In our previous work we have used a structural model for fault isolation. In this thesis we provide a formal framework and a prototype implementation of an approach taking benefit of a behavioral model. This gives opportunities to perform more sophisticated reasoning at the cost of a more detailed system model. We use a model-checker to reason about causal dependencies among the events of the modeled system. The model-checker performs reasoning about temporal dependencies among the events in the system model and the scenario at hand, allowing for conclusions about the causal relation between the events of the scenario. This knowledge can then be transferred to the corresponding fault in the system, allowing us to pinpoint the cause of a system failure among a set of potential causes. / <p>Report code: LiU-Tek-Lic-2000:59.</p>
| Identifer | oai:union.ndltd.org:UPSALLA1/oai:DiVA.org:liu-5734 |
| Date | January 2001 |
| Creators | Lawesson, Dan |
| Publisher | Linköpings universitet, TCSLAB - Laboratoriet för teoretisk datalogi, Linköpings universitet, Tekniska högskolan, Institutionen för datavetenskap |
| Source Sets | DiVA Archive at Upsalla University |
| Language | English |
| Detected Language | English |
| Type | Licentiate thesis, monograph, info:eu-repo/semantics/masterThesis, text |
| Format | application/pdf |
| Rights | info:eu-repo/semantics/openAccess |
| Relation | Linköping Studies in Science and Technology. Thesis, 0280-7971 ; 863 |
Page generated in 0.0021 seconds