Prediction of lithium-ion battery degradation trajectory in electric vehicles under real-world scenarios

Accurate prediction of lithium-ion degradation trajectory is essential to ensure the safe and reliable operation of electric vehicles (EVs). Owing to the complicated operating environment and different loads of batteries, it is challenging to guarantee the applicability of existing prediction techniques in the real world. In this work, we proposed a lithium-ion battery degeneration trajectory prediction method under real-world scenarios. First, a correction function is developed to alleviate the coupling influences of different temperatures and charging currents on the available capacity calculation of EV batteries. Second, a trend extraction algorithm combined with the rain flow counting method is proposed to reconstruct the equivalent full cycle-capacity relationship under random charging depths. Further, a Gamma process with Gibbs sampling and turning point detection method is used for degradation prediction and provides a probability distribution of the first passage time for individual vehicles and fleets. The effectiveness of the proposed method is verified by four years fleet operation data, with the mean absolute percentage error (MAPE) and root mean square error (RMSE) of capacity prediction kept below 0.49 % and 2.67 Ah, respectively. In addition, three maintenance strategies with increasing risk levels are suggested based on the reliability curve of the fleet.