Base On Multi-stable Stochastic Resonance Machinery Of The Weak Fault Signal Feature Extraction Method Research

With the continuous development of modern production technology, improving the real-time monitoring technology and fault diagnosis technology of mechanical equipment has become an inevitable trend. Mechanical fault signal is a weak characteristic signal in the strong noise background. It has become an important research direction and has been widely used in communication, biology, physics and other important aspects. In this paper, aiming at the shortcomings and limitations of stochastic resonance, a new machinery fault feature extraction method based on multi-stable stochastic resonance is proposed. A further study is conducted on the shortcomings and limitations of stochastic resonance and the application of multi-stable stochastic resonance in multi-frequency weak signal feature extraction. A method base on wavelet transform and parameter compensation band-pass multi-stable stochastic resonance is proposed with certain reference significance and application value for mechanical fault diagnosis.Aimed at the difficult problem of detecting weak signal buried under heavy background noise, a multi-stable SR model is proposed. First, study the langevin equation and Fokker-Planck equation; then derived multi-stable stochastic resonance of the output signal-to-noise ratio(SNR) formula, through the graphical system to illustrate the influence of system parameters and noise to the signal to noise ratio, adjusting parameters and noise intensity can improve the SNR and find the highest SNR optimal parameters values.Aiming at the shortcomings and limitations of stochastic resonance(SR), a novel method of weak signal detection based on variable scale multi-stable SR is proposed. First, the signal which is under large-scale frequency conditions while can not occur SR is processed by variable scale sub-sampling compression to make it meets the conditions of stochastic resonance, then make the signal through a multi-stable system and adjust the parameters to get the spectral characteristics of the signal, finally, compare it with the characteristic frequency gained from the bistable stochastic resonance method., the simulation and experiment results show that: under the same conditions, it is more accurate using the multi-stable SR method to obtain frequency than the bistable SR method, the proposed method also can enhance the signal amplitude, effectively detect weak signal submerged by strong noise, it is valuable and available at bearing fault signal analysis.Aimed at the difficult problem of detecting multi-frequency signals buried under heavy background noise, a method based on wavelet transform and parameter compensation band-pass multi-stable stochastic resonance is proposed. First, the noisy signal is processed by parameter compensation to counteract the effect of the damping term. Then the processed signal is decomposed into multiple signals of different scale frequencies by wavelet transform. Following this, we adjust the size of the scaled signals’ amplitudes and reconstruct the signals; the weak signal frequency components are then enhanced by multi-stable SR. The enhanced components of the signal are processed through a band-pass filter, leaving the enhanced sections of the signal. The processed signal is analyzed by FFT to achieve detection of the multi-frequency weak signals. The simulation experiment and the bearing case and gear case analysis showed that the method is simple and easy to operate, and can effectively detect the multi-frequency weak signal and greatly improve the SNR and it has obvious advantages compared with SR.The detection method of the weak signal of stochastic resonance under colored noise background is proposed. First,derive probability density function formula of steady state under multiplicative and additive colored noise excitation and mean first passage time formula. Analysis the influence on probability density function and the mean through time of various parameters. Finally use the medium speed shaft outer fault data of a company to analysis the detection of weak signal in colored noise, the result shows that this method can effectively extract weak signal feature information in colored noise.