Fault Diagnosis of Power Transformer Based on Large Margin Learning Classifier

The fault diagnosis of power transformer is important for safety of the device and reliability of the power system. This paper proposes the large margin learning classifier, which is well designed for multi-class problem based on the large margin learning of SVM hyper-planes theory. Each time it attempts to find the separating hyper-plane with maximum margin to split the clusters. As a novel tool, the large margin learning classifier is applied into the fault diagnosis of power transformer. Due to its extraordinary generalization capability, it has excellent performance on reliability and training speed. The experimental results show the feasibility and effectiveness of this method