Bearing Fault Diagnosis Based on Stochastic Resonance and BSS / ICA

A fault feature extraction method of rolling bearing based on cascaded mono-stable scale- transformation stochastic resonance and BBS/ICA was proposed. Noise energy was first transformed into noise signal energy through cascaded mono-stable scale-transformation stochastic resonance under the control of high frequency signal, then residual noise was separated from BSS/ICA. Theoretical analysis and numerical simulation results show that the approach can make use of noise to strengthen signal frequency characteristics, make great parameters signal get more energy from the system, and eliminate noise to achieve effective fault diagnosis. Simulation analysis indicates that this mono-stable system can effectively detect impact signal similar to rolling bearing fault, which is useful for engineering application. Simulation test performs in inner and outer fault of rolling bearing, and results sufficiently show the effectiveness of this approach.