Abstract: In view of the problem that the vibration signal of check valve contains background noise,fault feature extraction is difficult and the diagnostic accuracy is not high,a fault diagnosis method combining complementary ensemble empirical modal decomposition (CEEMD),singular value decomposition (SVD) and least squares support vector machine (LSSVM) is proposed.Firstly,the vibration signals of check valve are decomposed by CEEMD method,and the useful intrinsic mode functions (IMF) are obtained by energy analysis method and cross correlation analysis method.Then,the corresponding fault features are extracted according to the SVD method and input into the LSSVM for fault diagnosis.Compared with the ensemble empirical modal decomposition (EEMD) and support vector machines (SVM),this method can not only eliminate the modal aliasing and reduce the background noise,but also can effectively obtain the fault characteristics of the check valve,and finally get a higher diagnostic accuracy rate.