Abstract:

To address the problems of massive and dispersedly distributed rollers in belt conveyors, low accuracy, susceptibility to environmental impact, and high monitoring costs of existing fault diagnosis methods, a diagnosis method based on sound signals was proposed. Firstly, the sound signals of 30 sets of rollers of the belt conveyor under working conditions were collected through a BOURNS omnidirectional microphone. Then, statistical methods were used to analyze TD features of the collected sound signals, achieving preliminary fault feature mining. Finally, FD and TFD features of the sound signal were analyzed to improve the ability to express fault features. The results show that TD features of the roller sound can detect rust faults. FD features can detect serious faults, and TFD features can detect various typical bearing damages, having application prospects in the monitoring of roller faults in industrial belt conveyors.

Key words: sound signal, belt conveyor, roller, fault characteristic, time-domain (TD) feature, frequency-domain (FD) feature, time-frequency-domain (TFD) feature