Research On Noise Induced Multi-stable Stochastic Resonance Weak Signal Detection

Weak signal detection technology is widely used in areas like fault diagnosis,biology,and so on which is an important branch of signal processing.In recent years,stochastic resonance has been applied into the weak signal detection method,which has transformed the traditional suppression of noise detection into a non-linear detection method that make use of noise.In the weak signal detection,the noise and the signal achieve synergy through the stochastic resonance steady-state system,and part of the noise energy is transferred to the useful signal to be measured so as to “highlight” the target signal submerged in the noise.Due to the superior characteristics of the stochastic resonance,it becomes more and more popular in the research of weak signal detection.This thesis briefly introduces the research progress of stochastic resonance in the field of weak signal detection,analyzes the dynamic behaviors of the classical bistable system,and focuses on the stochastic resonance phenomena and the applications in the weak signals detection of the second-order tri-stable and hybrid tri-stable systems.To build a joint detection system,a variety of weak signal detection technologies are used in this thesis.The main work of the dissertation is as follows:1.Aiming at the problem that the classical bistable system can't effectively use noise energy to make the detection in the detection of weak signals unsatisfactorily,the second-order tri-stable stochastic resonance system is studied.Firstly,the tri-stable system is constructed from the trajectory of the cantilever beam,and the signal-to-noise ratio of the tri-stable system and the classical bistable system is compared theoretically.Then,to explored the stochastic resonance phenomena of the second-order tri-stable system,the background noise of the typical real noise model is regarded as background and the signal-to-noise ratio gain is used as metrics.Finally,the simulations is conducted to implement the weak signal detection,and the actual bearing fault data of Case Western Reserve University is detected with this system.The fault signal detects and compares the detection effect of a classical bistable system.The research shows that the stochastic resonance can be generated by adjusting the parameters of the second-order tri-stable potentials and the damping ratio when the noise and weak signals are constant.Therefore,by adjusting the potential function and the damping ratio,the noise efficiency can be improved and the noise energy can be promoted.The noise energy is transfer to the useful signal,at the same time the energy of the useful signal in the output of the system can be increased.The noise efficiency of the second-order tri-stable in the detection of weak signals and actual bearing signals is better than that of the classical bistable and can achieve better detection results.2.It is difficult to extract the fault feature signal because it is often submerged in the noise during diagnosis.A hybrid tri-stable stochastic resonance model is proposed by combining the woods-saxon monostable and the hybrid bistable.Firstly,the hybrid tri-stable construction method is described.Then stochastic resonance phenomena with different noise pulse exponents,various system parameters in symmetrical parameter and damping ratio changes are explored.Finally,the simulated fault signals in the background of ? noise and the actual fault data are used.The data of the rolling bearing data center in Case Western Reserve University verify the effectiveness of the system and compare the classical bistable system.The results show that the system can generate stochastic resonances by changing the system parameters and the damping ratio under different ? noise distributions.The hybrid tri-stable system is superior to the classical bistable system in the simulation of fault signals and rolling element fault detection.It can detect the signal to be measured while the classic bistable cannot.3.To further enhance the ability of stochastic resonance to detect weak signals,a joint detection system is proposed.Firstly,the noisy signal is suppressed by wavelet threshold and the suppressed signal is subjected to empirical mode decomposition.Then the signal-to-noise ratio localization modal method is used to select the high model component of signal-to-noise,and the selected modality is synthesized so that the features to be measured can be highlighted.Finally,it is input into a tri-stable stochastic resonance system.After the simulation signal and the simulation results of fault diagnosis in the inner and outer rings of the rolling bearing of Case Western Reserve University,it is shown that the system can accurately detect the target signal and improve the effectiveness of the detection system.