Application Research Of Rolling Mill Bearing Fault Diagnosis Based On Reinforce Ensemble Local Mean Decomposition

Rolling bearing is one of the important basic parts of rolling mill in metallurgical enterprises.Its good working condition is directly related to the safe and efficient operation of the equipment,and determines the quality and efficiency of the enterprise.Usually,the mill work in high temperature and high pressure,high humidity,dust environment,and in response to rapidly changing market demand,with many steel mills opened many varieties,specifications,small batch and batch production mode,make the rolling equipment operation condition is extremely complex,appear easily mill transmission system parts,such as rolling bearing fault,for rolling mill operations management brought huge challenge to health.Based on the collection of Local empirical Mode Decomposition(Ensemble Local Mode Decomposition,ELMD),rapid spectral kurtosis(Fast spectral kurtosis,FSK)signal of time-frequency analysis method based on the theory of rolling bearing as the research object,based on the theory of algorithm combined with experimental verification of technical route,in view of the rolling mill operation and service conditions,the vibration signals obtained in strong background noise and non-stationary and frequency and amplitude modulation,the characteristic frequency is to cover up,The difficulty of fault identification is greatly increased,and the method of extracting the characteristic indexes that highlight the running state of rolling bearing vibration signal is emphatically studied.The specific research contents include: proposed the feature extraction method of rolling bearing with enhanced ELMD,developed the data acquisition module of rolling bearing vibration based on LabVIEW software,laboratory bench verification test,and field data analysis and nuclear test calibration.(1)feature extraction method of RELMD rolling bearing vibration signal.For several PF components depend on human experience gained from ELMD algorithm directly when screening characterization of fault feature component empirical,blindness,and practical fast spectral kurtosis graph algorithm is found to kurtosis biggest band center frequency and bandwidth as a highlight of the fault component selection criterion,easy to cause fault components missing,lead to the problem of inaccurate fault diagnosis results.In view of this,a feature extraction method of enhanced total local mean decomposition(Reinforce,RELMD)was proposed according to the idea of "first enhance the impact components of each component,and then obtain the spectrum of each component",so as to increase the accuracy of fault diagnosis.(2)development of rolling bearing vibration data acquisition module.According to the requirements of bench test,engineering application and validity verification of algorithm,the data acquisition module of rolling bearing vibration was developed based on LabVIEW18.0 software.Firstly,the architecture and function of the acquisition module are designed based on the above requirements.Secondly,the interface layout of the front panel is based on the principle of friendly interaction.Finally,a matching block diagram is designed.(3)bench experiment and field application.On the basis of the detailed test process,the test system was set up and the vibration signal was collected by means of the comprehensive simulation bench of mechanical failure,vibration data acquisition module of rolling bearing and fault parts of rolling bearing of Spectra Quest corporation of the United States,which is located in the institute's mechanical fault diagnosis laboratory.By comparing and analyzing the measured vibration data of rolling bearing with the acquired vibration data in the field,the results show that the ELMD algorithm and RELMD algorithm are effective and reliable.