Document Type : Original research papers
Authors
1
Department of Electrical Engineering, Faculty of Engineering at Shoubra, Benha University, Cairo, Egypt Faculty of Engineering, Delta University for Science and Technology, Gamasa, Egypt
2
Department of Electrical Engineering, Faculty of Engineering at Shoubra, Benha University, Cairo, Egypt
3
Extra High Voltage Research Centre, Egyptian Electricity Holding Company, Egypt
4
4Electrical Power Engineering Department, School of Electronics, Communications and Computer Engineering, Egypt-Japan University of Science and Technology, Alexandria, Egypt Electrical Power and Machines Engineering Department, Faculty of Engineering, Tanta University, Tanta, Egypt
Abstract
Premature diagnosis of transformer oil faults enables the operator for diagnosing the transformer condition and hence operating the transformer continuously without outage. DGA is one of the most popular chemical tests used for fault diagnosis and there are many techniques which have been developed in this regard. This paper presents a proposed DGA technique for transformer oil fault identification based on the results of the recently published techniques. The proposed technique has been constructed based on the integration between the outputs of two recently established DGA techniques; Conditional probability and artificial neural network. A total of 532 datasets, obtained from the Egyptian Electricity Transmission Company (EETC) with known faults, have been used for designing and testing the proposed technique. The proposed fault diagnostic technique's accuracy attained 86.6 %, which is higher than the results of the combined techniques; 81.7% for ANN and 82.8% for conditional probability technique. The proposed developed technique came to the conclusion that integrating several DGA techniques, with higher accuracy, enhances fault detection's overall accuracy.
Keywords
Main Subjects
)
View on SCiNiTO