Experimental research on sensitivity of frequency domain feature parameters to hydraulic pump faults

It is critical in fault diagnosis to extract the feature vector of fault signal effectively and to find the sensitive factors. Therefore, a series of vibration signals are collected at pump end cover under some faults, such as loose sliding shoe, sliding shoe wear, sloping cam plate wear plate and the center spring failure, etc. Then, low frequency signals are obtained by means of suitable signal processing method and feature parameters are extracted in frequency; next, the sensitivity analysis of every fault has been processed on the basis of frequency-domain feature parameters so as to find a dimensionless sensitivity factor. Using this method it can provide reliable sensitive feature information, increase the completeness of fault information and improve the fault diagnosis rate of the system.