Machine fault diagnosis using industrial wireless sensor networks and on-sensor wavelet transform

This paper proposes a novel approach for machine fault diagnosis using industrial wireless sensor networks (IWSNs) and on-sensor calculation. In this paper, the induction motor and vibration signal are taken as an example of the monitored industrial equipment and signal due to their wide use. The discrete wavelet transform and wavelet energy-moment are used for on-sensor machine fault feature extraction, while a minimum distance classifier is adopted for on-sensor fault classification. The data from three motor operating conditions - motor normal operating condition, bearing fault from the motor drive, and bearing fault from the motor fan - are employed to evaluate the proposed system. Experimental results show that compared with raw data transmission, the proposed method can reduce the payload data by more than 99%, and deliver 100% fault diagnosis accuracy on two test sets.