Research On Key Technologies Of Big Data-driven Bearing Useful Life Prediction

Bearing as an important component,plays an important role in industrial manufacturing equipment.With the rapid development of sensor technology,the number and type of sensors are increasing quickly.And the sensors,such as vibration sensors,current sensors,and temperature sensors,are widely applied in bearing condition monitoring.Different kinds of sensor signals may reflect the bearing operation state in different aspects.And the volume of sensor signals is increasing exponentially as the monitoring time.The 4V characteristics of the big data are appeared in the sensor signals finally,those are volume,variety,velocity and veracity.These massive sensor signals gradually beyond the range of the traditional signal technology can handle.Therefore,it is of great theoretical and practical meanings to research the bearing useful life prediction method in big data environment.This paper focuses on the key technology of big data driven bearing useful life prediction.The main research contents are as follows:(1)The bearing useful life prediction framework in big data environment is proposed.First of all,the characteristic parameters of sensor signals,which may reflecting the performance degradation of the bearing,are extracted in the time domain,frequency domain and time-frequency domain by using the big data technology.And then the health index of the bearing may be obtained by the features selection and fusion,based on the Spearman rank correlation coefficient and principal component analysis.Next the bearing status may be identified by the degradation indictors of the health index.Afterwards the change of the bearing degradation performance may be reflected by exponential fitted method.Finally,the method of empirical Bayes estimation is applied to predict the useful life of the bearing.(2)The slewing bearing experiment is analyzed to verify the proposed method.The vibration sensor,current sensor and temperature sensor are used to monitor the degradation state of the slewing bearing.And the vibration sensor signals are analyzed by using large data technology to demonstrate the feasibility of the proposed method.(3)The prototype software system of bearing useful life prediction based on Hadoop is designed and developed.This system is designed based on the theory of bearing useful life prediction and the basic structure of Hadoop distributed system.On this basis,this system and its relevant functional modules based on C / S is developed on Linux system by using Java programming language.