Research of Bayesian networks application to transformer fault diagnosis

Abstract

The power transformer as the key equipment in electrical power systems, its operation reliability directly influences security of electrical power systems. Three-ratio method based on the Dissolved Gases Analysis is most widely used for transformer fault diagnosis currently. Considering the incomplete encoding and the over absolute faults classification zone of three-ratio method, this paper proposes no-code ratio method and Bayesian Network to diagnose the faults of transformer. The Bayesian Network diagnostic model is built by Bayesian Network Tool in MATLAB, and the simulation result shows the validity of this method. © 2011 Springer-Verlag.