Skip to Main Content
Data models play an important role in making an application easy to adopt in an industrial environment. In this paper, we discuss a data model built in XML Schema for supporting a framework for Inference and Diagnostics. The framework uses a probabilistic inference engine based on Bayesian networks. The primary function of the framework is to infer the possible cause for malfunctioning of a system from the symptoms or deviations from the normal behavior exhibited by the system. The data model not only supports building of Bayesian networks, it also enables building the repair strategies to put the system back into action. Another important feature of this data model is that it also supports the hierarchical component model, a model based diagnosis method which enables analyzing the system at different levels of abstraction. The paper discusses the data model in the context of the application being used, the way it is employed to represent Bayesian networks and also in the context of folding of Bayesian network according to the level of abstraction chosen. We conclude with a comparison of our effort with the existing efforts in the literature.