Feature extraction of frictional vibration based on CWT time-frequency image

A method to extract features of frictional vibration signals was developed, using continuous wavelet transform (CWT) time-frequency image (TFI) and image processing to analyze the vibration signals, since the frictional vibration signals contain a wealth of information about the interface´s fault state and friction wear state. TFIs were generated by CWT theory from the vibration signals produced by cylinder and piston ring samples sampled from 6S50MC marine diesel engine under two different lubrication statuses. And characteristic parameters obtained by image segmentation of TFIs were analyzed in order to establish the intrinsic relationship between the characteristic parameters and friction wear status. The results show that regular variations occur to the characteristic parameters with the conduct of the friction process, and characteristic parameters of different lubrication condition show significant differences. The CWT TFI characteristic parameters of vibration signals can be used to quantitatively describe vibration signal characteristic.