Reserach On Data-Driven-Based Remaining Useful Life Prediction Of Equipment

Remaining useful life prediction of equipment is a hot issue in the field of reliability.It has great theoretical significance and practical value to carry out degradation process modeling and remaining useful life prediction based on degraded data.However,the influence of imperfect maintenance on degradation process is seldom considered in the existing research.it is known that imperfect maintenance activities can bring objective economic benefits in engineering applications based on practical engineering experience,which can slow down the degradation process of equipment to a certain extent,and prolonging the life of equipment.Therefore,this paper mainly focuses on the three aspects of equipment degradation process modeling and remaining useful life prediction based on degradation data as follows:1.The research status of data-driven-based remaining useful life prediction is reviewed from the perspective of direct condition monitoring(CM)data and indirect CM data,and the progress and shortcomings of each method are discussed.2.Considering the time-varying uncertainty of degradation state,the influence of imperfect maintenance activities on equipment degradation process is studied.Based on the direct CM data,the degradation model is established,and the probability density function of remaining useful life is deduced under the concept of first hitting time.Then,the parameters of the model are estimated by using maximum likelihood estimation method through the degradation data and maintenance data of equipment.Finally,the effectiveness of the proposed method is verified by simulated data.3.Considering the difficulty of obtaining direct CM data in engineering practice,the remaining useful life of equipment based on indirect CM data is discussed.Firstly,the degradation model is established.Under the concept of first hitting time,Kalman filter algorithm is used to deduce the remaining useful life of hidden degradation state.Then,the parameters of the model are estimated based on the maximum likelihood estimation method using equipment's degradation data and maintenance data.Finally,the effectiveness of the proposed method is verified by the drift data of gyroscope.