Adaptive sliding mode observers for lithium-ion battery state estimation based on parameters identified online. (15th June 2018)

• Record Type: Journal Article
• Authors: Ning, Bo; Cao, Binggang; Wang, Bin; Zou, Zhongyue
• Is Part Of: Energy. Volume 153(2018)
• Page Start: 732
• Page End: 742
• Publication Date: 2018-06-15
• Subjects: Lithium-ion battery -- Parameters estimation online -- Adaptive sliding mode observer -- State of charge estimation -- State of health estimation; Power resources -- Periodicals; Power (Mechanics) -- Periodicals; Energy consumption -- Periodicals; 333.7905
• DOI: 10.1016/j.energy.2018.04.026
• Languages: English
• ISSNs: 0360-5442

Abstract

Abstract: Simplicity and accuracy are both important factors in real-time battery states estimation applications. However, a battery model initialized with static parameters which are identified in ideal laboratory conditions will not be able to get an accurate estimation in various actual applications. Besides, it is time-consuming and complex in implement. To solve the above problem, a new battery states estimation method is proposed. Firstly, an adaptive battery model is proposed according to a new online parameter estimation algorithm. Based on it, the parameter adaptive sliding mode observer for state of charge is proposed. Thus, the state of charge systematic error led from various work environments could be effectively reduced. The parameter adaptive sliding mode observer for state of health is proposed by tracing the derivative of open circuit voltage estimated online. As the reference open circuit voltage is estimated based on measurable inputs and outputs, rather than conventional observer with an assumed constant capacity. The estimated battery capacity could converge to the actual value while the error of battery open circuit voltage converges to zero. The proposed method is verified through the urban dynamometer driving schedule driving cycle. The results indicate that:1) parameters estimated online are accurate, 2) the absolute error of state of charge is less than 2%, 3) the estimated lithium-ion battery capacity could converge to the actual value with small capacity error. Highlights: An adaptive battery model with parameters estimated online is proposed. Parameter adaptive sliding mode observers are proposed for battery states estimation. Lyapunov stability criterion has been employed to verify the robustness of the proposed method. The proposed approach is verified by lithium-ion battery experiment. The proposed method could improve the battery estimation results of state of charge and state of health.