TY - JOUR
T1 - Co-estimation of capacity and state-of-charge for lithium-ion batteries in electric vehicles
AU - Li, Xiaoyu
AU - Wang, Zhenpo
AU - Zhang, Lei
N1 - Publisher Copyright:
© 2019 Elsevier Ltd
PY - 2019/5/1
Y1 - 2019/5/1
N2 - Precise capacity and state-of-charge (SOC) estimations are vital in order to guarantee safe and efficient operation of battery systems in electric vehicles. In this paper, a co-estimation scheme for battery capacity and SOC estimations is proposed, in which an equivalent circuit model (ECM) is used to represent battery dynamics. The recursive least squares (RLS) method and adaptive extended Kalman filter (AEKF) are leveraged simultaneously to achieve online model parameters identification and SOC estimation. Accelerated aging tests are conducted to investigate the relationship between partial voltage curves and aging levels of batteries. The Elman neural network is then employed to realize battery capacity prediction in real-time, which is used to refurbish the actual capacity of battery SOC estimator. The effectiveness of the proposed co-estimation scheme is experimentally verified under different driving cycles at varied temperatures. The results show that the SOC estimation error at room temperature can reach as high as 2% against discrepant aging levels, while the maximum estimation error is within 6% at varied temperatures.
AB - Precise capacity and state-of-charge (SOC) estimations are vital in order to guarantee safe and efficient operation of battery systems in electric vehicles. In this paper, a co-estimation scheme for battery capacity and SOC estimations is proposed, in which an equivalent circuit model (ECM) is used to represent battery dynamics. The recursive least squares (RLS) method and adaptive extended Kalman filter (AEKF) are leveraged simultaneously to achieve online model parameters identification and SOC estimation. Accelerated aging tests are conducted to investigate the relationship between partial voltage curves and aging levels of batteries. The Elman neural network is then employed to realize battery capacity prediction in real-time, which is used to refurbish the actual capacity of battery SOC estimator. The effectiveness of the proposed co-estimation scheme is experimentally verified under different driving cycles at varied temperatures. The results show that the SOC estimation error at room temperature can reach as high as 2% against discrepant aging levels, while the maximum estimation error is within 6% at varied temperatures.
KW - Adaptive extended kalman filter
KW - Battery capacity prediction
KW - Electric vehicles
KW - Elman neural network
KW - Lithium-ion batteries
KW - State-of-charge estimation
UR - http://www.scopus.com/inward/record.url?scp=85062231249&partnerID=8YFLogxK
U2 - 10.1016/j.energy.2019.02.147
DO - 10.1016/j.energy.2019.02.147
M3 - Article
AN - SCOPUS:85062231249
SN - 0360-5442
VL - 174
SP - 33
EP - 44
JO - Energy
JF - Energy
ER -