Experience Based Approach for Li-ion Batteries RUL Prediction

Prognostics and Health Management, is an engineering discipline gaining an increasing interest as it aims at, improving the safety and the reliability of reducing maintenance costs by providing an accurate estimation about the remaining useful life RUL and current health status of the equipment. In practice, this RUL estimation is a challenging task. The challenges come from the nature of data and the context in which the equipment was run. The behavior of a system run under different operating profiles is difficult to predict. In this paper, an experience based approach for RUL prediction is presented and tested an real Li-ion batteries data set, collected at different operational profiles. The degradation of the components is represented by health indicators obtained by learning a regression model. The regression model is trained using run-to-failure data and knowledge extracted from the latter. RUL is predicted by retrieving the most, similar instance. obtained results are interesting. (C) 2015, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved.