Oil Paper Insulation State Evaluation Method Integrating Fuzzy K-Nearest Neighbor and D-S Proof Theory

Abstract

In order to realize the accurate evaluation of transformer oil paper insulation state, this paper proposes the evaluation method of integrating Fuzzy K-Nearest Neighbor (FKNN) and D-S evidence theory. Firstly, the multi-characteristic parameter database of reply voltage method is constructed based on the measured data of reply voltage method transformer. Then, for the database, a basic probability allocation method based on FKNN is proposed to reduce the influence of subjective factors. Finally, each evidence is integrated through the D-S evidence theory to obtain confidence results for the insulating state proposition, which avoids the limitations of individual feature parameter evaluation. Using the proposed method of the transformer measured data of the database, the results show that the method of the confidence results can not only accurately reflect the transformer oil paper insulation state, provide guidance for the maintenance strategy, and can reflect the deterioration trend of transformer oil paper insulation to a certain extent.