A robust approach to state of charge assessment based on moving horizon optimal estimation considering battery system uncertainty and aging condition. (10th October 2020)

• Record Type: Journal Article
• Authors: Ren, Hongbin; Zhang, Hongwei; Gao, Zepeng; zhao, Yuzhuang
• Is Part Of: Journal of cleaner production. Volume 270(2020)
• Publication Date: 2020-10-10
• Subjects: State of charge estimation -- Moving horizon estimation -- Nonlinear optimization -- Robust -- Uncertainties; Factory and trade waste -- Management -- Periodicals; Manufactures -- Environmental aspects -- Periodicals; Déchets industriels -- Gestion -- Périodiques; Usines -- Aspect de l'environnement -- Périodiques; 628.5
• DOI: 10.1016/j.jclepro.2020.122508
• Languages: English
• ISSNs: 0959-6526

Abstract

Abstract: Accurate battery state information is essential for battery management system application and safety monitoring. However, it is a challenge task to obtain satisfied estimation results due to the uncertainties and inconsistencies of battery packs caused by aging. To solve this challenge, an optimization-based moving horizon estimation approach is presented in this paper for battery state and parameter online estimation. The dynamic battery parameters including open circuit voltage and internal resistance in equivalent circuit model are described by polynomial function of state of charge and input current for estimation algorithm design. And the intrinsic connection and difference between extended Kalman filter and moving horizon estimation algorithm are explicitly explained. Both of them are least square based estimation approach, and Kalman filter is a special form of moving horizon estimation, while moving horizon estimation relax Markov assumption compared with extended Kalman method. And then the optimization-based moving horizon estimation is designed for parameters and state of charge online assessment for battery dynamic system. To reduce computing time, the software framework CasADi is used for differential-algebraic calculation and nonlinear optimization. Three mismatch working conditions are studied for estimation performance validation, including mismatched initial guess values and battery dynamic characteristics difference caused by aging condition. The experimental results demonstrate that optimization-based moving horizon estimation performs better than Kalman filter-based approaches in terms of estimation precision, convergence time and robust. The proposed optimization-based moving horizon estimation is a promising approach for state of charge estimation in commercial battery management system applications. Highlights: The intrinsic connection between Kalman filter- and moving horizon estimation approaches are explained. A nonlinear optimal based moving horizon estimation approach is proposed to estimate SOC in real-time. The maximum SOC estimation error of MHE is less than 2.95% and the average error is less than 0.2%. Optimal based MHE guarantees the robust to the initial value mismatch and battery aging conditions.