MKurt-LIA: mechanical fault vibration signal measurement scheme with frequency tracking capability for bearing condition monitoring

Abstract

Monitoring the operational status of rolling bearings is crucial to ensuring the reliable operation of mechanical equipment, and it is considered as an important task of prognostics and health management in mechanical industry. Finding fault pattern directly from vibration signals measured on bearings can be challenging due to strong noise. Lock-in amplifiers (LIAs) can effectively measure the strength of weak fault signals distorted by noise, thus providing insights for fault diagnosis. However, LIAs are sensitive to the reference frequency, and an incorrect reference frequency can render the measurement even meaningless. In this study, a measurement scheme of LIA with the ability to track time-varying fault frequencies is proposed for measuring fault vibration signal components of in-service bearings, termed MKurt-LIA. The MKurt spectrum is employed for searching the local fault frequencies in the time domain, overcoming the negative impact introduced by using theoretical fault characteristic frequencies calculated based on bearing geometry in previous works. Guided by the statements in Guide to the Expression of Uncertainty in Measurement and its supplements, this work demonstrates through the analysis of propagation of distributions that the negative impact of fault frequency deviation on measurement results is significant and cannot be ignored. Experimental results confirm that the MKurt-LIA scheme effectively measures the amplitude of fault frequency components during bearing service, enabling the detection of fault evolution in the early stages where fault characteristics are completely submerged in the time domain waveform. MKurt-LIA scheme fully respects the fact that fault frequency always deviates from theoretical value, and achieves significantly robust performance in assessing in-service bearing’s operation condition.

Lock-in spectrum: a tool for representing long-term evolution of bearing fault in the time–frequency domain using vibration signal

“…This assumption is based on the belief that the characteristic frequencies of bearing faults are strictly linearly correlated with the rotational speed, with linear relationship determined by the geometric properties of the bearing itself. The latest applications have realized that this assumption may be inadequate (Zhang, 2024a).…”