Secondary signal-induced large-parameter stochastic resonance for feature extraction of mechanical faults

Abstract

Aiming to solve the problem that it is difficult to extract large parameter signals from a strong noise background, a novel method of large parameter stochastic resonance (SR) induced by a secondary signal is proposed. The SR mechanism of high-frequency signals is expounded by analyzing the density distribution curve. High-frequency signals are converted to low-frequency signals using the scale transformation method, and then large-parameter SR is induced by the secondary signal. Ultimately, the method is applied to the feature extraction of mechanical faults. Simulation and experimental results indicate that (i) the effect of SR induced by the secondary signal is significantly enhanced when the frequency of the secondary signal is twice that of the signals to be detected after the scale transformation; (ii) when the frequency of secondary signal is twice the maximum frequency of the signals to be detected after the scale transformation, choosing an appropriate amplitude of secondary signal can alleviate the problem that the noise energy is excessively concentrated in the low-frequency channel with regard to the extraction of two-frequency or three-frequency high-frequency signals; and (iii) by adding the secondary signal to the engineering example, the fault power spectrum value of system output is 101% higher than that without the secondary signal.