Research On HHT And Fuzzy C-means Clustering Based Hydraulic Pump Fault Diagnosis

Hydraulic pump is the power source part and playing an important role in the wholehydraulic system.The quality of the system is affected directly by the hydraulic pumpworking condition normal or not,so the condition monitoring and fault diagnosis of thepump is particularly important.The axial piston pump fault signals are typical non-linearand non-stationary signals,so a method processing non-linear and non-stationary properlyis required in order to extract the fault feature accurately .An axial piston pump in Electro-hydraulic Servo System of a materials testingmachine in this paper, the vibration acceleration and pressure signals are collected. Thetime-frequency of collected signals is conducted by use of the method of Hilbert-HuangTransform(HHT). The intrinsic mode function(IMF) sensitive to faults is chosenaccording to the short-term maximum entropy spectrum and power analysis from theIMFs of the decomposed vibration signals. The fault frequency of kinds of fault states arefound.A method of constructing characteristic vector based on the characteristic energy ofthe local Hilbert marginal energy spectrum is proposed. In order to choose the right IMFsto construct characteristic vector,a method according to the cross correlation and theenergy percentage between reconstructed signal and original signal is proposed.The vibration and pressure signals are treated by means of the empirical modedecomposition (EMD) and the ensemble empirical mode decomposition (EEMD)respectively when constructing the characteristic vector, the faults are recognized by useof the fuzzy C-means clustering method. The recognition results show that thecharacteristic vector proposed in this paper can reflect the fault characteristic of axialpiston pump accurately;the fault recognition accuracy rate based on EEMD is improvedsubstantially as comparing with based on EMD;the characteristic vector of pressuresignals is better than the vibration signals.