Stochastic Resonance Of A Class Of Mechanical Systems And Its Applications

Stochastic resonance is a phenomenon caused by the interaction of noise,weak periodic signal and nonlinear environment.It not only changes the stereotype that noise is harmful,but also achieves a series of application results based on stochastic resonance principle.Stochastic resonance is generated by the amplification of weak signals induced by noise,which converts part of the noise energy into signal energy to enhance the output of the system,thus improving the signal-to-noise ratio(SNR).Through stochastic resonance processing,the amplitude and energy of weak signals are improved,and the accuracy of fault detection is also enhanced.In addition,stochastic resonance method does not need to filter noise,and useful signals will not be weakened,so it has high theoretical and practical research value.In this paper,stochastic resonance of nonlinear system is studied and applied to bearing fault diagnosis,and the diagnostic performance of the system is evaluated.The main research contents are as follows:1.The asymmetric bistable system under Gaussian color noise excitation is described.The expressions of mean first passage time(MFPT)and SNR are calculated,and the effects of various parameters on SNR are analyzed,such as noise intensity,asymmetry coefficient,signal amplitude and frequency.And the relationship between MFPT in two different directions and parameters is investigated.The SNR parameters are optimized by using the adaptive particle swarm optimization(APSO)algorithm,and the simulation signals are analyzed.2.Aiming at the problem of low output performance of traditional stochastic resonance,an underdamped piecewise time-delayed feedback system(UPTFSR)is proposed,and its feasibility in bearing fault diagnosis is discussed.Based on the underdamped system method,a piecewise potential function which can adjust the steepness independently is constructed to change the shape of the bistable state trap,and time-delayed feedback is introduced.In this way,the historical information is used to enhance the periodicity of the signal in the feedback process.By selecting appropriate parameters,the optimal matching effect of noise,signal and potential is achieved,and clearer output waveform and higher SNR are obtained.Through simulation and experiment,the reliability of this strategy is verified.3.In order to improve the ability of signal detection,the influence of the model under the condition of multi-time-delayed feedback and tri-stable state is further discussed.Firstly,an improved tri-stable model and multi-time-delayed structure are introduced,and the expressions of equivalent potential function and SNR are derived.Secondly,the influence of each parameter on the system performance is studied,and the parameters are optimized to make the system performance reach the optimal level.In simulation and experiment,compared with envelope analysis method and classical stochastic resonance method,the proposed method obtains higher SNR and clearer output waveform.The results show that this method has certain advantages in extracting bearing fault characteristics.