Application of Adaptive Hierarchical Multi-class SVM to Transformer Fault Diagnosis

On the basis of the relationship between dissolved gases in transformer oil and transformer fault,a tran sformer fault diagnosis method adopting adaptive hierarchical multi-class SVM is proposed.Based on the concept of feature extraction in pattern recognition,a hierarchical structure is employed with different input vectors for transformer fault judgment and fault type identification.By optimizing the multi-class SVM with adaptive optimization method and comparing the diagnosis effects and recognition rates of different fault types,it indicates that the composition ratio of dissolved gases in transformer oil is more sensitive to transformer fault types.Comparison of test results and analysis of support vector machine parameters demonstrate that the proposed method has high accuracy and excellent generalization.