Research On Rolling Bearing Fault Diagnosis Based On Time-frequency Post-processing Algorithm

As a key component of rotating machinery,the health status of rolling bearings directly affects the normal operation and overall performance of the equipment.Therefore,it is important to carry out the research of rolling bearing fault diagnosis.The rolling bearing vibration signal contains rich fault information,but it is difficult to obtain clear and accurate features from the signal by traditional signal processing methods due to the influence of time-varying working conditions and background noise.In recent years,many excellent post-processing algorithms have been developed on the basis of the original time-frequency analysis methods,which provide new ideas for rolling bearing fault diagnosis.In this paper,the research of rolling bearing fault diagnosis method is carried out based on time-frequency post-processing algorithm.The main research contents are as follows:The rolling bearing failure characteristics and failure forms are studied,and the vibration mechanism of rolling bearing failure is described.The advantages and limitations of existing time-frequency analysis methods and their post-processing algorithms in nonlinear signal processing are analyzed.For the problem of nonreassignment points of multisynchrosqueezing transform method,the local maximum squeezing operator is defined to further squeeze the energy diverged in the time-frequency results.The effectiveness of the improved algorithm is verified by analyzing numerical signals and bat echo signals.For the non-stationary nonlinear characteristics of rolling bearing fault signal,the local maximum iterative matching Chirplet transform is proposed.The matching Chirplet transform is introduced,and the chirp rate parameter is defined as a function of timefrequency,which adaptively fits the time-varying multi-component signal.The improved multisynchrosqueezing transform is applied to the time-frequency results provided by the matching Chirplet transform.An iterative compression operator based on the matching Chirplet transform is derived so that the fuzzy energy in the time-frequency plane is gradually concentrated near the time-frequency ridge.Finally,the local maximum squeezing operator is used to concentrate all the energy to the ridge line.The proposed algorithm is examined in terms of energy concentration,noise robustness and practicality using numerical signals and rolling bearing fault signals.Detection of transient shock components is a powerful tool for rolling bearing fault diagnosis.To accurately extract transient features from complex background disturbances,the horizontal synchrosqueezing extracting transform is proposed based on the idea of time-frequency post-processing in the time direction.For the fast time-varying and wideband characteristics of pulse-like signal,the redistribution of signal energy along the time direction is beneficial to enhance the fault characteristics of the signal.A second-order time group delay estimate is first computed to squeeze the divergent energy to the timefrequency ridges.Then a new extraction operator is defined to extract the transient shock features of the signal from the noise background and further sharpen the time-frequency spectrum.The transient feature extraction capability of the proposed algorithm is verified by simulation signals and rolling bearing fault signals under variable speed conditions.