Rotating machine fault diagnosis through enhanced stochastic resonance by full-wave signal construction

This study proposes a full-wave signal construction (FSC) strategy for enhancing rotating machine fault diagnosis by exploiting stochastic resonance (SR). The FSC strategy is utilized to transform a half-wave signal (e.g., an envelope signal) into a full-wave one by conducting a Mirror-Cycle-Add (MCA) operation. The constructed full-wave signal evenly modulates the bistable potential and makes the potential tilt back and forth smoothly. This effect provides the equivalent transition probabilities of particle bounce between the two potential wells. A stable SR output signal with better periodicity, which is beneficial to periodic signal detection, can be obtained. In addition, the MCA operation can improve the input signal-to-noise ratio by enhancing the periodic component while attenuating the noise components. These two advantages make the proposed FSCSR method surpass the traditional SR method in fault signal processing. Performance evaluation is conducted by numerical analysis and experimental verification. The proposed MCA-based FSC strategy has the. potential to be a universal signal pre-processing technique. Moreover, the proposed FSCSR method can be used in rotating machine fault diagnosis and other areas related to weak signal detection. (C) 2016 Elsevier Ltd. All rights reserved.