Fault diagnosis and condition monitoring of bearing using multisensory approach based fuzzy-logic clustering

This paper investigates the application of multisensor fault feature extraction and fuzzy-logic based clustering for the condition monitoring of bearing. Multiple independent sensors on an electric motor drive system provide valuable early indication of a fault, and can be effectively utilized to perform high reliable and optimal fault detection. Through utilizing common sensors including current sensor and vibration sensors in motor, motor current signature analysis (MCSA) and vibration analysis have been used to extract the bearing fault energy. The discrete wavelet transform (DWT) has been applied to monitor energy of the bearing fault signals. Then, the fuzzy c-mean (FCM) has been developed to utilize the data from single sensor and multisensor to identify the severity of bearing fault. Extensive theoretical analysis and experimental test has been performed to demonstrate the advantages of proposed approach. The validity of this study has been confirmed through analysis of the 1/6 HP single phase induction motor and drive system.