Fault diagnosis of centrifugal pump anchor bolt loosening based on RBF neural network
SONG Liwei1,ZHANG Yixun1,CHEN Zeyu2,ZHANG Yuhang2,FAN Chuanhan2,XIAO Xingxin2,DONG Liang2*
1. State Key Laboratory of Nuclear Power Safety Monitoring Technology and Equipment of CGN Engineering Co., Ltd., Shenzhen, Guangdong 518124, China; 2. National Research Center of Pumps, Jiangsu University, Zhenjiang, Jiangsu 212013, China
Abstract:In order to accurately identify the loosening fault of the anchor bolt of horizontal centrifugal pump, a diagnostic platform of horizontal centrifugal pump unit was built, and eddy current sensor was used to monitor the rotor displacement of centrifugal pump. The acquired rotor displacement signals were decomposed into multiple intrinsic mode functions(IMF)by empirical mode decomposition(EMD), and the fault sensitive component was obtained by analyzing the IMF spectrum characteris-tics, correlation coefficient and energy ratio of each layer. Finally, the radial basis function(RBF)neural network was used to identify and predict the loosening fault of the centrifugal pump. The results show that the EMD method can effectively extract the centrifugal pump loosening fault features, and the IMF5—IMF8 layer can be used as the fault feature components. An accuracy of 95% can be reached by inputting the correlation coefficient and energy ratio of IMF5-IMF8 layers into the RBF neural network as fault features for recognition.