Abstract
Aiming at the nonlinear characteristics of hydraulic pump vibration signals and low diagnosis accuracy in fault diagnosis, a feature extraction method based on ICEEMDAN sample entropy and LLTSA popular learning was proposed.Firstly, multiple intrinsic mode functions are obtained by ICEEMDAN decomposition method for signals of different modes, and the kurtosis index is used for preliminary screening of components. Then, the 6 dimensional feature vector is formed by calculating sample entropy of each component. Secondly, the low dimensional fault features with good clustering are obtained by using LLTSA for further dimensionality reduction of features. The low dimensional features are classified by SSA- SVM method.The experimental results show that the fault diagnosis model can accurately identify all kinds of faults of hydraulic pump, and the advantages of this diagnosis method are verified by comparing with other pretreatment methods.
Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
