Interpretation of DGA for Transformer Fault Diagnosis with Step-by-step feature selection and SCA-RVM

Abstract

Oil-filled transformer is one of the important devices in power grid. To enhance the accuracy of transformer fault diagnosis and to ensure the stable performance of power system, an initial feature set, composed of the volume fraction of the seven dissolved gas and the constituted twenty-eight-set of dissolved gases selected by the step-by-step feature and SCA-RVM, is raised by analyzing the dissolved gas in oil. Then the ReliefF algorithm is used to select the sensitive features to be fused later. After that, the redundancy of the fused features is eliminated by the kernel LDA (KLDA), and lastly the step-by-step features are fed into the SCA-RVM diagnosis model. The result shows that, the accuracy of the diagnosis model can reach as high as 97.01%. Therefore, with the superior accuracy, this model can provide some references in transformer fault diagnosis.