Research On Mechanical Fault Feature Extraction Method Based On Time-delayed Stochastic Resonance And Variational Mode Decomposition

Rotating machinery is one of the main types of mechanical equipment failure,and its fault signal detection is of great significance for monitoring the state of mechanical equipment.Aiming at the problem that the fault signal of rolling bearing is difficult to detect under strong noise,this paper adopts the method of time-delay stochastic resonance and variational mode decomposition(VMD)to de-noise and extract the features of the fault signal,and uses the least squares support vector machine(LSSVM)optimized by particle swarm optimization(PSO)to recognize the type of fault signal.The details are as follows:In this paper,the Langevin equation of the general tristable stochastic resonance(TSR)model is studied,and the Fokker-Planck equation with time delay is derived.The fourth-order Runge-Kutta equation is used to solve the stochastic resonance model numerically.For the high frequency signal limitation problem in the adiabatic approximation theory,the variable-scale stochastic resonance method is adopted.The validity of stochastic resonance system for feature extraction of high frequency weak signals is verified by simulation signals,which provides a basis for processing actual fault signals.A tristable stochastic resonance model based on Gaussian white noise induced delay feedback is proposed,and the influence of each parameter on the model is analyzed.By comparing with the general tristable stochastic resonance model,the time-delayed feedback tristable stochastic resonance(TFTSR)model is more effective in reducing the noise of the actual fault signal.Aiming at the shortcomings of the time-delayed feedback tristable stochastic resonance model,a linear time-delayed feedback tristable stochastic resonance(LTFTSR)is proposed,and the time-delayed Fokker-Planck equation and the mean first passage time(MFPT)are derived,and the effects of various parameters on the two models are compared.The feasibility and superiority of the model are verified by comparing the results of the feature extraction of the simulation signal and the actual fault signal indifferent models.To solve the problem of extracting weak fault signals from strong noise background,a feature extraction method based on time-delay stochastic resonance model and VMD is proposed.By comparing the results of fault signal feature extraction with time-delay stochastic resonance model and empirical mode decomposition(EMD),it is found that VMD has the advantage of extracting fault features.Finally,the energy entropy after signal decomposition is selected as the feature input of pattern recognition,and the PSO-LSSVM method is used to identify the fault signal type intelligently.