Multiscale noise tuning of stochastic resonance for enhanced fault diagnosis in rotating machines

Abstract

The interference from background noise makes it difficult to identify incipient faults of a rotating machine via vibration analysis. By the aid of stochastic resonance (SR), the unavoidable noise can, however, be applied to enhance the signal-to-noise ratio (SNR) of a system output. The classical SR phenomenon requires small parameters, which is not suited for rotating machine fault diagnosis as the defect-induced fault characteristic frequency is usually much higher than 1Hz. This paper investigates an improved SR approach with parameter tuning for identifying the defect-induced rotating machine faults. A new method of multiscale noise tuning is developed to realize the SR at a fixed noise level by transforming the noise at multiple scales to be distributed in an approximate 1/f form. The proposed SR approach overcomes the limitation of small parameter requirement of the classical SR, and takes advantage of the multiscale noise for an improved SR performance. Thus the method is well-suited for enhancement of rotating machine fault identification when the noise is present at different scales. A new scheme of rotating machine fault diagnosis is hence proposed based on the SR with multiscale noise tuning and has been verified by means of practical vibration signals carrying fault information from bearings and a gearbox. An enhanced performance of the proposed fault diagnosis method is confirmed as compared to several traditional methods.