Pulse Signal Detection And Recovery Via Stochastic Resonance System

Weak signal detection technology is getting increasing attention,in the field of mechanical fault diagnosis.In the early stages of failure,features of the failures are usually hidden in the background noise,which are difficult to detection,thus accidents may happen with prediction remain undiscovered.Traditional detection methods tent to reduce the noise to increase SNR.However stochastic resonance methods use noise to increase the features of the signal to realize the detection of signal.Since the discovery of stochastic resonance,stochastic resonance theory and research continues to develop.Stochastic resonance is used to detect both periodic signal and aperiodic signal.In this thesis,with the Gaussian pulse signal as the model,the detection and recovery of the pulse signal is studied.Firstly the construction feature of bistable stochastic resonance systems is analysis.Single well oscillation of bistable system is used to detect pulse signal practically.The particle oscillates in single well rather than double well.For the positive impulse,the detection effect in the negative well in the bistable system is better than the positive well.However the transition phenomenon of the particle has the adverse effect on the detection effect.Secondly response feature of pulse signal via mono-stable system is studied.Pulse wave distortion nearby the equilibrium position of pulse signal has negative influence in the detection of signal.A method of moving pulse series is put forward in the thesis to solve the problem.The mechanism of the proposed method reducing the distortion is discussed in detail.The moving pulse series method can adjust the distribution of the noise power spectrum,improve monostable stochastic resonance and detect pulse signal effectively by setting offset.The recovery system functions of monostable system and bistable system were given to realize the recovery of pulse signal.The efficiency of signal waveform recovery of the recovery systems is analysis.The bias-monostable system which is similar with the negative potential of bistable system is formed by setting offset in the thesis.The detection effect and the recovery effect of the two systems is analysis and compared.The conclusion that the bias-monostable system and bistable system have the same detection capability and the same recovery capability on pulse signal is draw.Finally that the method of moving pulse series is beneficial to detect pulse signal masked by noise is verified by the experiment of rolling bearing fault signal detection.The bistable SR and bias monostable SR technique are compared in analyzing the metal cutting course.