Transformer Fault Diagnosis Based on Hierarchical Fuzzy Support Vector Machines

Abstract

Large power transformer, as the key equipment of power system, plays an influential role to ensure the safe operation of power system. In this paper, transformer fault diagnosis model is built based on Fuzzy Support Vector Machines（FSVM） which combines Support Vector Machines (SVM) with fuzzy degree of membership. Hierarchical classification algorithm for multi-class classification is applied to diagnose the transformer fault. The membership value of the FSVM is obtained by Fuzzy C Means. Parameters of the FSVM model are optimized with Genetic Algorithm (GA). The transformer states are divided into trouble-free(normal), low temperature overheating T1, medium temperature overheating T2, high temperature overheating T3, low-energy discharge D1, high-energy discharge D2, and partial discharge PD. A mass of fault samples are analyzed and the results are compared with those obtained by the methods of Back-Propagation Neural Network (BPNN) and SVM, which shows that the proposed model is more effective and accurate. So the given method of transformer fault diagnosis based on Fuzzy Support Vector Machines is feasible. DOI: http://dx.doi.org/10.11591/telkomnika.v11i10.3414