Online available capacity prediction and state of charge estimation based on advanced data-driven algorithms for lithium iron phosphate battery. (1st October 2016)

• Record Type: Journal Article
• Authors: Deng, Zhongwei; Yang, Lin; Cai, Yishan; Deng, Hao; Sun, Liu
• Is Part Of: Energy. Volume 112(2016)
• Page Start: 469
• Page End: 480
• Publication Date: 2016-10-01
• Subjects: Battery management system -- Online parameter identification -- State of charge estimation -- Available capacity prediction -- Dual adaptive extended Kalman filter -- Least squares support vector machine; Power resources -- Periodicals; Power (Mechanics) -- Periodicals; Energy consumption -- Periodicals; 333.7905
• DOI: 10.1016/j.energy.2016.06.130
• Languages: English
• ISSNs: 0360-5442

Abstract

Abstract: The key technology of a battery management system is to online estimate the battery states accurately and robustly. For lithium iron phosphate battery, the relationship between state of charge and open circuit voltage has a plateau region which limits the estimation accuracy of voltage-based algorithms. The open circuit voltage hysteresis requires advanced online identification algorithms to cope with the strong nonlinear battery model. The available capacity, as a crucial parameter, contributes to the state of charge and state of health estimation of battery, but it is difficult to predict due to comprehensive influence by temperature, aging and current rates. Aim at above problems, the ampere-hour counting with current correction and the dual adaptive extended Kalman filter algorithms are combined to estimate model parameters and state of charge. This combination presents the advantages of less computation burden and more robustness. Considering the influence of temperature and degradation, the data-driven algorithm namely least squares support vector machine is implemented to predict the available capacity. The state estimation and capacity prediction methods are coupled to improve the estimation accuracy at different temperatures among the lifetime of battery. The experiment results verify the proposed methods have excellent state and available capacity estimation accuracy. Highlights: A dual adaptive extended Kalman filter is used to estimate parameters and states. A correction term is introduced to consider the effect of current rates. The least square support vector machine is used to predict the available capacity. The experiment results verify the proposed state and capacity prediction methods.