| Popis: |
In view of the frequent occurrence of power system disasters in extreme environments, extreme weather will lead to transformer faults and increase the risk. Therefore, it is very important to identify transformer faults in extreme weather. In this paper, a transformer fault diagnosis method based on MDT-AdaBoost algorithm and kernel principal component analysis is proposed. Firstly, the transformer state characteristics and weather characteristics are taken as input features, and the kernel principal component analysis is used to reduce the dimensionality Then the MDT-AdaBoost algorithm is used to diagnose and identify transformer faults. Finally, simulation results show that the accuracy rate of transformer fault identification using the proposed method reaches 94.81%, and the recognition accuracy converges rapidly with the increase of decision tree depth and iteration times. The accuracy and effectiveness of the proposed method are verified by simulation, which provides a reliable reference for transformer fault diagnosis under extreme weather conditions. |
