Application of stochastic resonance in bearing fault diagnosis

Due to their low frictional resistance, fast startup, low power consumption, and high mechanical efficiency, rolling bearings are widely used in medical apparatus, auto industry, mineral industry, aerospace industry, etc. It is necessary to monitor the running condition of them to prevent breakdown during their operation. This paper presents an approach for detecting the initial failure of rolling bearings based on Stochastic Resonance (SR) combined with wavelet analysis. SR is a kind of nonlinear phenomena which can enhance weak characteristic signal by utilizing noise. Compared with linear method it can detect signal with low signal-to-noise ratio (SNR). One of the challenging issues in SR is the determination of noise intensity of input signal. This paper uses wavelet transform to address this issue. The procedure of SR method is as follows: (1) Estimate the frequency range of input signal. (2) Figure out the optimal parameters of the system. (3) Calculate the noise intensity of the input signal through wavelet transform. (4) Take the parameters into SR model and solve it, then the output signal would be obtained. (5) Analyze the output signal, and fine-tune the parameters to get the optimal results.