Diagnostic model maturation

Performing fault diagnosis is an extremely difficult problem [1]. Given a complex system and a set of tests or alarm outcomes, the user wants to determine the most likely faulty component. Some of the most common diagnostic algorithms used are rule based, fault dictionaries, and fault trees [2], [3], [4]. Model-based methods provide another common set of approaches to performing system level diagnosis, which compare observations from the system with those expected from a model [5], [6].