Weak Signal Detection Method Based On Underdamped Stochastic Resonance Potential Model And Application

Due to the periodic movement of rotating machinery,there is a high probability of failure.If the state monitoring and maintenance of the rotating machinery are not carried out in time,the machine will stop if the danger occurs,causing production stagnation.Heavy weight will cause casualties to the operator and produce irreparable losses.Bearings and gears are the main components in rotating machinery,so the weak fault diagnosis of bearings and gears needs more and more attention.The working conditions of bearings and gears are generally accompanied by strong noise,which makes it difficult to identify weak fault signals.Traditional signal processing methods,such as wavelet transform and empirical modal demodulation,etc.,are based on signal filtering for fault identification.Stochastic resonance solves the damage of the original filtering method to the original signal.The method is to use the noise component to convert the noise energy into a weak noise signal ability.When the best match among noise,weak signal and nonlinear system is achieved,the effect of noise reduction will be achieved,and the effect of enhancing weak fault signals will be achieved while reducing noise.The main factors affecting the filtering effect of stochastic resonance are the signal type,the nonlinear system(potential model)and the model order.Based on the above analysis,this paper introduces three new potential models based on second-order underdamped stochastic resonance: continuous potential,single-parameter potential,and periodic potential.(1)First draw the graph of the potential model,analyze the influence of the potential model parameters on the potential structure,and conclude that the parameter changes can have different effects on the model's potential well width,time barrier height,and potential wall steepness.(2)Then the theoretical derivation of the theoretical SNR signal-to-noise ratio is carried out.When the noise intensity gradually increases,the signal-to-noise ratio will reach the highest value at a certain moment,and then when the noise intensity continues to increase,the signal-to-noise ratio will gradually decrease.It tends to be stable.Overdamped stochastic resonance is a first-order differential equation, and second-order underdamped stochastic resonance expression is a second-order differential equation,which adds a damping term to overdamped stochastic resonance.Second-order underdamping is equivalent to secondary filtering,and its output effect is better than traditional overdamped stochastic resonance.(3)Finally,the proposed new potential model is applied to the second-order underdamped stochastic resonance,and the ant colony algorithm is used to optimize the potential parameters to obtain the best output signal to noise ratio.The method is verified by simulation signals and experimental data.Compare the output effect of the proposed method with the traditional stochastic resonance method,and draw a conclusion.