Research On Fault Diagnosis Of Printing Machine Bearing Based On Periodical Stochastic Resonance

Fault diagnosis of the printing equipment bearing is of great significance to ensure the good operation of the printing machine.As the transmission support component of the printing equipment,the bearing works for a long time under the harsh environment of uneven force,friction,heat generation,etc.,and is prone to wear,internal an.d external ring rupture,ball rupture and other failures.The working state of the printing machine is often accompanied by large background noise,the early fault signal of the bearing is a weak signal,which is difficult to find.This paper uses the periodical stochastic resonance(PSR)method to identify the weak fault of the bearing from the vibration signal by using the noise to enhance the bearing fault characteristics.The specific research contents are as follows:Aiming at the output saturation defect of the classic bistable stochastic resonance(BSR),This paper studies and establishes a PSR system model.The parameters of the system model can be independently adjusted and do not affect each other,so the output saturation phenomenon can be effectively avoided.In order to overcome the classic stochastic resonance and only satisfy the small parameter condition,the variable-scale PSR is studied for fault detection under large frequency,and the simulation signal is used to verify the effectiveness of the proposed method.For the traditional stochastic resonance,it is difficult to achieve stochastic resonance when the fault frequency is unknown.Aiming at this problem,an adaptive periodical stochastic resonance(APSR)bearing fault detection method based on grey wolf optimizer(GWO)algorithm is proposed in this paper.This method uses the output signal-to-noise ratio(SNRout)as the optimization index,and can realize the bearing fault detection adaptively.The simulation signal and experimental signal verify the effectiveness of the proposed method.The diagnosis results show that the proposed method not only can effectively identify the fault frequency,but also obtain a large SNRout,showing a strong fault extraction capability,and has certain engineering practical value.The traditional intelligent diagnosis method is a shallow model.The characteristics learned through training are limited,and the ability is insufficient when dealing with complex faults.Aiming at this problem,This paper presents a bearing fault identification method based on periodical stochastic resonance and deep belief network(PSR-DBN).This method uses PSR to enhance the characteristics of the fault data,and then inputs the data into the DBN deep network model for training.The diagnosis results show that the training accuracy rate based on PSR-DBN method reaches 98.65%,and the test accuracy rate reaches 98.32%,showing strong fault identification capability.