Investigation on the effects of measurement and temporal uncertainties on rolling element bearings prognostics

摘要: Estimation of remaining useful life (RUL) rolling element bearings (REBs) has a major effect on improving the reliability in industrial plants. However, due to complex nature fault propagation these components, their prognosis is affected by various uncertainties. This intensified when recorded data offline, which very common for many machines lower cost rather than online monitoring strategy. In present paper, order overcome shortcoming feed-forward neural network (FFNN) REBs prognostics, new method considering two main uncertainties (caused measurement and process noises) proposed, presence offline acquisition. Inthe proposed method, primary RUL probability distribution corresponded each measured predicted, utilizing outputs trained FFNNs. Then, predicted will become more robust confronting temporal changes, taking into account approval pervious stage predictions prediction. As result, overall also its confidence levels (CLs) areobtained. Finally, evaluation performed byutilizing bearing experimental datasets. The results show that capability express estimated CLs acquisition effectively. By providing probabilistic perspective, can improve asset decision-making about future