Research On Fault Diagnosis Method Of Rolling Bearing Based On Improved Bistable Stochastic Resonance

As the load element of mechanical equipment,rolling bearings are very prone to failure.Once a failure occurs,it will cause certain economic losses and even endanger the lives of staff.Therefore,it is important to research rolling bearing failure signals,especially early failure signals.The early failure signal of rolling bearings is very weak and often submerged in noise.Stochastic resonance can enhance the failure characteristic signal from the signal containing noise interference.Different from other fault diagnosis methods that improve the signal-to-noise ratio by weakening the noise,stochastic resonance improves the signal-to-noise ratio by transferring part of the noise energy into the weak failure characteristic signal through a nonlinear system.In this paper,based on the traditional bistable stochastic resonance system,three different stochastic resonance systems are proposed and used for rolling bearing fault diagnosis.The main research questions of this paper are as follows:The basic structure and common failure types of rolling bearings are analyzed.The fault vibration signals of different parts of bearings and the characteristic frequencies of failure in different parts are studied.The theory of stochastic resonance is described,and common stochastic resonance evaluation indexes are analyzed.The variable scale stochastic resonance and particle swarm optimization algorithms are studied,and the variable step size and optimal noise stochastic resonance systems are proposed.The classical bistable stochastic resonance system is optimized by changing the step size and adding noise.Finally,the system before and after optimization are compared and analyzed by simulation and by bearing example.To solve the output saturation problem of traditional stochastic resonance,a superposed segmented bistable stochastic resonance system based on particle swarm optimization algorithm is proposed.A superimposed,segmented potential function model is developed to alleviate the output saturation problem to some extent.Derive Kramers rates and signal-to-noise ratio formulas for the system and analyze the effect of specific parameters on the signal-to-noise ratio.An evaluation function is established to evaluate the system’s performance and assist the system in screening parameters.Finally,the effectiveness of the proposed system is verified by comparing the simulated signals and rolling bearing failure signals with other systems.The effect of potential function shape on the signal-to-noise ratio is studied,and a concave-convex bistable stochastic resonance system based on particle swarm optimization algorithm is proposed.The effects of potential well width,potential barrier height,potential wall tilt,and potential function concavity on the signal-to-noise ratio are analyzed.A concave-convex potential function model is constructed,and the effect of each parameter of this model on the signal-to-noise ratio is analyzed.An evaluation function is established to evaluate the system’s performance and assist the system in selecting parameters.Finally,the simulation experiment and rolling bearing failure experiment are compared and analyzed with other stochastic resonance systems.In summary,the method proposed in this paper can enhance a weak signal with a low signal-to-noise ratio and realize the enhancement of a weak fault characteristic frequency under strong background noise.It is beneficial to the early fault diagnosis of rolling bearings and has a broad application prospect.