No abstract

Fault identification represents the preliminary step in solving or preventing a failure in reliable systems. The identification passes through the recording and analysis of events used as probes of underlying fault. Sometimes those events are incomplete, they can have in any order and novel in part. In this paper we propose a model based on Bayesian reasoning as means to isolate compatible faults.

No abstract

No abstract

System fault detection and recovery deals with a decision problem under uncertainty in which we first attempt to isolate a fault according to information we collect regarding the system behavior, and after to recovery from the failure by the application of some recovery actions. In this paper we propose a method which makes use of Bayesian networks to reason under uncertainty and decision analysis to search the best recovery action.