Low-Speed Bearing Fault Diagnosis Based on ArSSAE Model Using Acoustic Emission and Vibration Signals

• Published in: IEEE access 2019, Vol.7, p.46885-46897
• Main Authors: Saufi, Syahril Ramadhan, Ahmad, Zair Asrar Bin, Leong, Mohd Salman, Lim, Meng Hee
• Format: Article
• Language: English
• Subjects: acoustic emission analysis; Acoustic emission testing; Adaptation models; adaptive resilient stacked sparse autoencoder (ArSSAE); Analytical models; bearing fault diagnosis; Data models; Deep learning; Emission analysis; Fault diagnosis; Feature extraction; Low speed; Roller bearings; Vibration analysis; Vibration measurement; Vibrations
• ISSN: 2169-3536; 2169-3536
• DOI: 10.1109/ACCESS.2019.2909756

The development of rolling element bearing fault diagnosis systems has attracted a great deal of attention due to bearing components having a high tendency toward unexpected failures. However, under low-speed operating conditions, the diagnosis of bearing components remains a problem. In this paper, the adaptive resilient stacked sparse autoencoder (ArSSAE) is proposed to compensate for the shortcomings of conventional fault diagnosis systems at low speed. The efficiency of the proposed ArSSAE model is initially assessed using the CWRU database. Then, the proposed model is evaluated on actual vibration analysis (VA) and acoustic emission (AE) signals measured on a bearing test rig at low operating speeds (48-480 rpm). Overall, the analysis demonstrates that the ArSSAE model is able to perform an accurate diagnosis of bearing components under low-speed conditions.