State Evaluation and Fault Prediction of Bearing Packets Based on Industrial Big Data
Aiming at issues such as long disintegration cycle of roughing mill bearings,the"black box"operating status,difficulties in providing accurate and effective disintegration strategies and information purification difficulties,an innovative industrial test system is presented in this paper.The dynamic information of the upper/lower work roll bearings in the rolling of different slabs is collected accurately,completely and in real-time.Acceleration signals from different passes are analyzed in the feature domain,and bearing mechanics models and failure mechanisms are fused to provide accurate and effective bearing package disintegration strategies for the production site.The results show that the acceleration signals of the three sets of bearing packs tested by tracking can clearly identify the four stages of roll idling,steel biting,stable rolling and steel throwing,in which there is an obvious periodic impact in the stable rolling stage,and the values of indices such as the waveform factor and the impulse factor can be used to clearly distinguish the different bearing packs'operating states.The frequency domain integration method is used to simulate the horizontal response characteristics of the system using the acceleration signal.In addition to biting and throwing steel instantly,external forces also cause the system to produce a large horizontal response,in which the lower roll response amplitude is significantly larger than that of the upper roll.The study provides a precise and effective strategy for the maintenance of the roughing mill's work roll bearings in the lower machine and the predictive repair of failures.
industrial big datarolling bearingdisintegration strategiesreliability assessmentfault forecast
NETL
NSTL
万方数据
