Fault feature enhancement of gearbox in combined machining center by using adaptive cascade stochastic resonance

The difficulty to select the best system parameters restricts the engineering application of stochastic resonance (SR). An adaptive cascade stochastic resonance (ACSR) is proposed in the present study. The proposed method introduces correlation theory into SR, and uses correlation coefficient of the input signals and noise as a weight to construct the weighted signal-to-noise ratio (WSNR) index. The influence of high frequency noise is alleviated and the signal-to-noise ratio index used in traditional SR is improved accordingly. The ACSR with WSNR can obtain optimal parameters adaptively. And it is not necessary to predict the exact frequency of the target signal. In addition, through the secondary utilization of noise, ACSR makes the signal output waveform smoother and the fluctuation period more obvious. Simulation example and engineering application of gearbox fault diagnosis demonstrate the effectiveness and feasibility of the proposed method.