Stochastic Capacity Loss and Remaining Useful Life Models for Lithium-ion Batteries in Electric Vehicles

Abstract:

This paper proposes and validates a stochastic prognostic model for capacity loss and remaining useful life (RUL) in lithium-ion pouch cells with graphite anode and NMC-LMO cathode. The model was developed using data from an experimental campaign which studied the effect of C-rate, minimum SOC, temperature, and charge-depleting usage on aging in plug-in hybrid electric vehicle (PHEV) batteries. The proposed algorithm estimates capacity loss and RUL as a function of resistance and operating conditions including charge sustaining/depleting use and temperature, and its stochastic nature is able to capture the variability of the data. The battery resistance is estimated using a Particle Filter developed for an experimentally validated Equivalent Circuit Battery Model. The Particle Filter is designed to perform combined estimation of State of Charge and resistance, which is used as an input to the stochastic model. Finally, the stochastic model predicts capacity loss with an RMS error of less than 1% and RUL with an RMS error of 1.6 kAh, and can be integrated into on-board battery management systems in Electric Vehicles (EVs) with similar aging conditions to monitor the health of lithium-ion batteries.