Long-Range Dependent Stochastic Prediction Model for Lithium-Ion Battery RUL Prediction Using CC-CV Dataset

Lithium-ion battery degradation does not follow the Markov property. Its current degradation state is influenced by a wealth of historical observations, which manifests long-range dependence (LRD). In this paper, capacity is taken as the characteristic parameter to quantify the degradation behaviour of lithium-ion batteries. A prediction model based on the fractional Poisson process (fPp) with long-range correlation is proposed. The constant-current constant-voltage dataset from the CALCE repository is used as the test set, and the Monte Carlo simulation algorithm is implemented to realize lifetime prediction. The effectiveness of the proposed method is verified by comparative experiments against the fractional Brownian motion (fBm), fractional levy stable motion (fLsm), Wiener, and LSTM models, where the root-mean-square error (RMSE), mean absolute error (MAE), coefficient of determination (R2), and mean relative error (MRE) are employed as evaluation metrics.