Application of fuzzy support vector machine for determining the health index of the insulation system of in-service power transformers

• Published in: IEEE transactions on dielectrics and electrical insulation 2013-06, Vol.20 (3), p.965-973
• Main Authors: Ashkezari, A. D., Hui Ma, Saha, T. K., Ekanayake, C.
• Format: Article
• Language: English
• Subjects: Algorithms; Condition assessment of transformer; Electric power generation; Fuzzy; fuzzy support vector machine (FSVM); Health; health index; Indexes; Insulation; insulation system; Oil insulation; oil test; power transformer; Power transformer insulation; support vector machine (SVM); Support vector machines; Training; Transformers
• ISSN: 1070-9878; 1558-4135
• DOI: 10.1109/TDEI.2013.6518966

With the integration of data and information obtained from a variety of chemical and electrical tests on transformer insulating oil, it is possible to evaluate the health condition of the insulation system of an in-service power transformer. This paper develops an intelligent algorithm for automatically processing the data collected from oil tests and determining a health index for the transformer insulation system. This intelligent algorithm adopts a fuzzy support vector machine (FSVM) approach, which constructs a statistical model using a training database based on the historic data collected from 181 in-service power transformers. The procedure of constructing the training database, the formulation and implementation of FSVM and the data preprocessing methods for dealing with a class imbalanced training database is presented in this paper. Numerical experiments are also conducted to evaluate the performance of the algorithms developed in the paper.