Study on vibration characteristics and fault diagnosis method of oil-immersed flat wave reactor in Arctic area converter station

Wenqing Lai,Yuandong Wang,Wenpeng Li,Guang Sun,Guomin Qu,Shigang Cui,Mengke Li,Yongqiang Wang

DOI: https://doi.org/10.1063/1.5005264

2017-01-01

AIP Conference Proceedings

Abstract:Based on long term vibration monitoring of the No.2 oil-immersed fat wave reactor in the ±500kV converter station in East Mongolia, the vibration signals in normal state and in core loose fault state were saved. Through the time-frequency analysis of the signals, the vibration characteristics of the core loose fault were obtained, and a fault diagnosis method based on the dual tree complex wavelet (DT-CWT) and support vector machine (SVM) was proposed. The vibration signals were analyzed by DT-CWT, and the energy entropy of the vibration signals were taken as the feature vector; the support vector machine was used to train and test the feature vector, and the accurate identification of the core loose fault of the flat wave reactor was realized. Through the identification of many groups of normal and core loose fault state vibration signals, the diagnostic accuracy of the result reached 97.36%. The effectiveness and accuracy of the method in the fault diagnosis of the flat wave reactor core is verified.