پديدآورندگان :
Behzad Mehdi m_behzad@sharif.edu Professor, School of Mechanical Engineering, Sharif University of Technology, Tehran, Iran ; , Mirfarah Motahareh MS Student, School of Mechanical Engineering, Sharif University of Technology, Tehran, Iran , Addin Arghand Hesam PhD Cadidate, School of Mechanical Engineering, Sharif University of Technology, Tehran, Iran.
كليدواژه :
Vibration condition monitoring , Rolling element Bearing , Prognostics , Neural Network , Confidence Level
چكيده فارسي :
Neural network (NN) is used for prognostic purpose in many applications. Although this method is very powerful for making a prognostic model in complex systems, finding the uncertainty for its prediction result is an important challenge. ISO 13381-1 emphasizes on the confidence level (CL) determination for any remaining useful life (RUL) prediction. It provides a procedure for calculating CL. In this paper, a procedure based on that offer is proposed for calculating CL of RUL prediction for rolling element bearings (REBs) with NN method. Two sources of error have been considered and a model has been proposed to study them. Experimental bearing accelerated life test data of PRONOSTIA dataset have been employed to study the performance of proposed method. It is shown that the determined CL for prediction results shows reasonable behavior.
