Bearing fault diagnosis method with unknown variable speed based on multi-curve extraction and selection

Abstract

Aiming at the bearing fault diagnosis with unknown time-varying speed, a fault diagnosis method of variable speed bearing based on multi-curve extraction and selection, Vold-Kalman filter (VKF) and generalized demodulation (GD) is proposed. The designed filter decomposes the original vibration signal into signal components carrying speed information and bearing fault information respectively. Fast path optimization algorithm is used to extract time-frequency curves iteratively from the time-frequency representation (TFR) of components. The high-precision instantaneous shaft rotation frequency (ISRF) curve is selected and calculated based on the proposed selection index, and then the demodulation phase function is constructed. Using VKF and GD, the bearing fault pulse harmonics are extracted and demodulated. Finally, by establishing reconstructed demodulation spectrum, bearing fault state identification and classification can be achieved. The experimental and analysis results confirm that the proposed method can successfully extract and demodulate fault pulse harmonics without tachometer. Different fault states can be accurately identified, and the effects of interference components and extraction errors can be avoided. Further analysis demonstrates that it can also identify weak bearings fault. Thus, the proposed method is an effective fault diagnosis technology for unknown time-varying speed bearing.