A rolling bearing status monitoring method based on subband spectral fuzzy description

Abstract Vibration signals provided by rotating machinery are informative about their operating states. By nature, the vibration signal behavior is non-stationary. To this end, extraction of discriminating and fault-sensitive parameters a major challenge in field monitoring machines. Conventional fault diagnosis methods based on processing use statistical feature time domain, frequency domain time-frequency representation. In article, new method proposed for detection classification bearing defects spectral subband using membership functions. Statistical including energy, Center frequency, root variance Shannon entropy considered. Compared to common features, extracted can provide information. These finally fed into generalized RBF neural network system trained with Resilient Backpropagation (Rprop) algorithm classify seven pre-established types ball bearings under multiple shaft speeds load conditions. The results suggest that significantly improve diagnostic performance terms accuracy estimation level.