Remaining energy estimation for lithium-ion batteries via Gaussian mixture and Markov models for future load prediction

Other than upgrading the energy storage technology employed within electric vehicles (EVs), improving the driving range estimation methods will help to reduce the phenomena, known as range anxiety. The remaining discharge energy (RDE) of the battery affects the remaining driving range of the vehicle directly and its accurate calculation is crucial. In this paper a novel approach for the RDE calculation of the battery is proposed. First a stochastic load prediction algorithm is prepared via a Markov model and Gaussian mixture data clustering. Then, the load prediction algorithm is connected to the battery second order equivalent circuit model (ECM) coupled with a bulk parameter thermal model. Based on the extrapolated load and the battery dynamics, the battery future temperature conditions, future parameter variations and its internal states are predicted. Finally, the battery end of discharge time is prognosed and its RDE is calculated iteratively. In order to prove the proposed concept, lithium-ion battery cells are selected and the performance of the method is validated experimentally under real-world dynamic current charge/discharge profiles.