Fault diagnosis for oil-filled transformers using voting based extreme learning machine
- 99 Downloads
Extreme learning machine (ELM) based fault diagnosis for oil-filled transformers overcomes some drawbacks faced by that using traditional learning algorithms. Since the randomized hidden nodes are used and they remain unchanged during the training phase, some samples may be misclassified near the classification boundary. To reduce the number of such misclassified samples, fault diagnosis using voting based ELM (V-ELM) was proposed in this paper. The V-ELM-based diagnosis method incorporates multiple independent ELMs to improve the classification performance. Firstly, the user-specified parameter of individual ELM was chosen for dissolved gas analysis samples through experiment. Then, the unstable performance of individual ELM was demonstrated on testing samples. Finally, the network complexities and performance of V-ELM-based diagnosis were compared with original ELM approaches. Experimental results show that the proposed method achieves a much higher correct classification rate and the performance is more reliable.
KeywordsPower transformers Fault diagnosis Dissolved gas analysis Extreme learning machine Majority voting method
The authors acknowledge the Doctoral Scientific Research Foundation of Northeast Electric Power University (no. BSJXM-201401), China.