Adaptive stochastic resonance quantified by a novel evaluation index for rotating machinery fault diagnosis

Abstract

Stochastic resonance (SR) provides the enhancement of fault characteristic signals with the assistance of noise. The performance of adaptive SR must be evaluated using an appropriate index to automatically enhance various characteristic signals. This paper proposes a weighted impulse (WI) index to evaluate the performance of the adaptive general scale transformation SR (AGSTSR) in rotating machinery fault diagnosis and uses an intelligent optimization algorithm to obtain the optimal WI. The comparison results in the simulation experiment revealed that the proposed WI-based AGSTSR method presented the fault characteristics more clearly than the adaptive SR method based on the impulse index or weighted kurtosis index. Moreover, the proposed method offered the best anti-noise performance in the simulation experiment. Finally, two experimental case studies validated that the proposed method can adaptively realize early fault diagnosis of rotating machinery through the analysis of weak fault characteristics.