Prediction of dissolved gases content in power transformer oil by support vector regression machine

Support vector regression machine (SVRM) has been successfully employed to solve regression problem of nonlinearity and small sample. However, SVRM has rarely been applied to predict dissolved gases content in power transformer oil, though it has a great potential in this area. In this study, SVRM was proposed to predict dissolved gases content in power transformer oil. The prediction model of dissolved gases content in power transformer oil was established based on regression arithmetic of SVRM, among which the leave-one-out cross validation method was used to determine free parameters of SVRM. The experimental data from several electric power companies were used to illustrate the performance of proposed SVRM model. The experimental results indicate that the SVRM method can achieve greater accuracy than grey model method under the circumstances of small training data. Consequently, the SVRM model is a proper alternative for predicting dissolved gases content in power transformer oil.