Concentration Prediction of Dissolved Gases in Transformer Oil by Relevance Vector Regression Algorithm

Abstract

Concentration prediction of dissolved gases in power transformer oil is very significant to detect incipient failures of transformer early. Concentration prediction of dissolved gases in power transformer oil is complicated problem due to its nonlinearity and little training data. Relevance vector regression algorithm (RVR) is applied to concentration prediction of dissolved gases in transformer oil in this paper. Compared with traditional support vector machine, relevance vector regression algorithm has the higher prediction accuracy because it has less support vectors than support vector regression algorithm. The concentration prediction model of dissolved gases in power transformer oil is established based on regression algorithm of RVR. The experimental results indicate that the RVR method can achieve greater prediction accuracy than support vector regression algorithm. Consequently, the RVR model is a proper alternative for predicting the concentration of dissolved gases in power transformer oil. The article presents some promising patents on concentration prediction of dissolved gases in transformer oil by relevance vector regression algorithm. Keywords: Concentration prediction, chromatographic analysis transformer, relevance vector regression, time series prediction