Effective band-selection algorithm for rolling element bearing diagnosis using AE sensor data under noisy conditions

Abstract

When diagnosing a rolling element bearing (REB), it is important to select the frequency band that has the most defect information. Many band-selection methods have been developed in recent years. Most existing methods target the vibration signal; hence, these methods are often unsuitable for use with acoustic emission (AE) sensors. With existing methods, the large sampling rate and high sensitivity of AE sensor causes huge computing costs and susceptibility to noise. To realize sensitive diagnosis with AE sensors, it is necessary to develop a proper band selection algorithm that operates under noisy conditions and with low computing cost. Thus, this paper proposes a segment-based fault information assisted band selection method for AE sensor data. The proposed method is validated by applying it to both simulated and experimental data. The test data contain random impulsive and non-Gaussian noises to represent the signal from other components and electrical noise from the motor system, respectively. With traditional methods, these noises either interrupt the proper band selection or increase the computing cost; however, the proposed method handles these noises and provides proper band selection with moderate computing cost.