Probability Forecasting of Lithium-ion Batteries Remaining Useful Life by Using the Additive Quantile Regression Model with Splines

Batteries play a key role in the reliability of electronic and electrical devices. The remaining useful life prediction is a crucial issue for providing maintenance and replacement for the battery system in time. Different from other existing research that they pay attention to the conditional mean of the remaining useful life, a probability forecasting method based on quantile regression is proposed. This provides an opportunity for a more complete view of the degradation of the lithium-ion battery. The results showed that the additive quantile regression model with splines significantly outperformed the linear quantile regression model.