Power transformer

as one of the most important equipment in the modern power system

undertakes the conversion and distribution of electrical energy

which provides a crucial guarantee for the reliable operation of the power grids.Due to the intricate mapping relationship between the gas dissolved in oil data and transformer fault types

the accuracy of the existing fault diagnosis methods is not high enough so far.In this paper

a transformer fault reasoning method based on relay protection and gas dissolved in oil is proposed.Firstly

the Bayesian networks are established based on the operating condition of the transformer and the characteristics of the relay protection device.Secondly

the probabilistic support vector machines are trained by the dissolved gas data in oil to acquire probability of operation state and fault degree.And then

the Monte Carlo simulation and maximum likelihood estimation are implemented to estimate the conditional probability of nodes in Bayesian networks based on historical statistics.At last

three case studies show that the proposed method can not only solve the drawback of only using dissolved gas data

but also provide more accurate maintenance decision for maintenance personnel.