Development of a novel asset management system for power transformers based on ontology

This paper aims to develop a formalised framework, which can perform reasoning with uncertainty in semantic web, for adopting rules with interlinked relationships to form an interoperable knowledge base for power transformers and developing a probabilistic diagnosis system to provide quantified confidence support if uncertainties occur. The framework provides a set of structural translation rules to map an OWL taxonomy into a Bayesian Network (BN) directed acyclic graph. Firstly, the advantages and shortages of crisp logic based ontology are introduced. Secondly, the essential concepts of BNs are introduced, which are graphical representations of uncertain knowledge. The algorithm of knowledge integration is used to refine an existing BN with more reliable sources. Finally, the framework, which augments and supplements OWL with additional functions for representing and reasoning with uncertainty based on BN, will be demonstrated by an small-scale transformer diagnosis example.