Fault Tree Reuse Across Multiple Reasoning Paradigms

The development of a diagnostic model can be a very time-consuming and manually intensive process. One must first analyze the Test Program Set (TPS) to determine the fault tree and then integrate with that any additional knowledge that can be obtained from external data sources (such as test results, maintenance actions from the various maintenance levels, run-time failure information, etc.). The diagnostic models defined in the IEEE Std. 1232-2002 (AI-ESTATE) each define a different method that can be used for a diagnostic reasoner. It has been determined that each of these models utilize the information found in the basic TPS fault tree. As the fault tree represents hard won engineering knowledge that is expensive to reproduce, it is desirable to share the fault tree representations across multiple reasoner models. This paper will layout how each model type in the AI-ESTATE standard utilizes the fault tree to perform diagnostics and how, through the use of the XML representation of the AI-ESTATE fault tree model, that basic fault tree can be shared between reasoner models. It would also be desirable to find a way to gain that fault tree knowledge without having to manually reproduce it. As such, this paper will also describe how that information can at least be semi-automatically extracted from TPS design artifacts.