Research On Fault Diagnosis Method Of Low-speed Rolling Bearing

Low-speed rolling bearings are widely used in subway escalators,extrusion granulator,wind generator and other equipment.As the key supporting components,it is of great practical significance to study the fault diagnosis methods of low speed rolling bearings.However,the low speed leads to weak bearing fault signal,and the complex structure of large equipment is accompanied by serious environmental noise,which increases the difficulty of signal denoising and fault feature extraction.Therefore,based on the failure vibration response characteristics of low speed rolling bearings,The fault diagnosis methods of low speed bearing are studied,which mainly include:(1)The fault vibration characteristics of low-speed rolling bearings are manifested as low signal-to-noise ratio,weak transient impact.Based on the frequency domain response characteristics of bearing faults,research is carried out from two aspects:background noise separation and fault feature enhancement.Modulated bispectral analysis is combined with particle swarm optimization algorithm.The optimal filtering center frequency is determined by bispectral analysis in advance to reduce the complexity of the optimization problem.The range of slice analysis is obtained based on the optimal filter bandwidth solved by particle swarm calculation,and the background noise of the signal is separated.The bispectral coherence function is used to reconstruct the modulation spectrum,and finally the fault features are enhanced and the bearing faults are diagnosed.(2)The theory of the traditional shock pulse method is studied.Based on the mechanism of bearing fault shock generation,the shock quantification method and the feature value alarm mechanism,a bearing fault quantitative characterization method based on the high frequency resonance signal of the conventional acceleration sensor is proposed.Aiming at the weak fault impact of low-speed rolling bearings,the mechanism of sensor resonance excited by impact is used to capture and enhance the fault impact.By calculating the pulse characteristic value of the signal resonance frequency band response,the damage degree of the bearing is effectively quantified and the working state of the bearing is judged.(3)The limitation of the conventional feature index in the spectral kurtosis algorithm is analyzed.Aiming at the characteristics of different types of noise and the sensitivity of the conventional feature index,the simulation signal is used to verify that a single feature index is easily interfered by related types of noise.An improved Protrugram algorithm based on weighted kurtosis value is proposed.The product of the time domain kurtosis value and the impulse modulation factor is used as the characteristic index to comprehensively evaluate the transient impact and cyclic smoothness of the signal.The accuracy of the proposed method is verified by simulation signals.(4)Aiming at the problem that the high-frequency resonance frequency signal of the sensor cannot be collected under the limited sampling rate,a quantitative diagnosis method for bearings based on Protrugram algorithm and digital SPM is proposed.The frequency response function of the digital simulation sensor is constructed by selecting the bearing fault characteristic components in the low sampling rate signal.The frequency response function of the simulated sensor is convolved with the bearing fault impact signal,and the sensor resonance phenomenon in the SPM method is simulated to enhance the fault impact.Then,the sparse filter of the redundant dictionary is used to denoise the impact-enhanced signal to obtain the final reconstructed signal.Finally,the envelope spectrum characteristic value of the reconstructed signal is calculated to standardize the fault impact.The method proposed in this paper is verified by bearing vibration signal of subway escalator test bench,and the result shows that the quantitative diagnosis of low-speed rolling bearing can be realized.