A comparison of compiled reasoning systems and model-based reasoning systems and their applicability to the diagnosis of avionics systems

The two approaches to diagnostic knowledge-based systems, model-based reasoning and compiled reasoning systems, are compared. The focus of this comparison is the basic problem-solving strategies and knowledge representation issues involved in these two broad categories of diagnostic reasoners. The investigation highlights the differences in the approaches. Two representative shells are used to build prototype diagnostic systems for the same device, using the same experts. The compiled knowledge shell used in this study is CSRL, a generic task tool developed at the Ohio State University. The model-based shell used was the fault isolation system (FIS), a tool developed by the US Navy specifically for the diagnosis of electronic and analog devices. The device for this study was a black-and-white TV. The compiled-knowledge and model-based approaches are compared based on run-time behavior, knowledge requirements, knowledge acquisition, training utility, and potential integration with avionics systems