An improved adaptive weights correction-particle swarm optimization-unscented particle filter method for high-precision online state of charge estimation of lithium-ion batteries.

Li, Zehao; Wang, Shunli; Yu, Chunmei; Qi, Chuangshi; Shen, Xianfeng; Fernandez, Carlos

Abstract

In the battery management system (BMS), the state of charge (SOC) of lithium-ion batteries is an indispensable part, and the accuracy of SOC estimation has attracted wide attention. Accurate SOC estimation can improve the efficiency of battery use while ensuring battery safety and improving battery life. Taking ternary lithium battery as the research object, this paper proposes a parameter identification method using adaptive forgetting factor recursive least squares and an improved joint unscented particle filter algorithm to estimate SOC. Firstly, an adaptive method is used to select the appropriate forgetting factor value to improve the accuracy of the forgetting factor recursive least squares (FFRLS) method. Meanwhile, the improved particle swarm (IPSO) optimization algorithm that incorporates variable weights and shrinkage factors is utilized to make the best choice of the noise for the unscented Kalman filter (UKF) algorithm in order to improve the estimation accuracy of the UKF algorithm. At the same time, the UKF algorithm is used as the suggestion density function of the particle filter (PF) algorithm to form the unscented particle filter (UPF) algorithm. In this paper, the AFFRLS algorithm and IPSO-SDUPF algorithm are combined to estimate the SOC of Li-ion batteries in real time. Experimental results under different working conditions show that the proposed algorithm has good convergence and high stability for SOC estimation of lithium-ion batteries. The maximum estimation errors of this algorithm are 1.137% and 0.797% for BBDST and DST conditions at 25°C, and 1.015% and 1.029% for BBDST and DST conditions at 35°C, which are lower than those of the commonly used algorithms of EKF, SDUKF, IPSO-SDUKF, and SDUPF, and provide a reference for future. The maximum estimation errors are lower than those of the commonly used EKF, SDUKF, IPSO-SDUKF, and SDUPF algorithms, which provide a reference for the future high-precision SOC estimation of Li-ion batteries.

Citation

LI, Z., WANG, S., YU, C., QI, C., SHEN, X. and FERNANDEZ, C. 2023. An improved adaptive weights correction-particle swarm optimization-unscented particle filter method for high-precision online state of charge estimation of lithium-ion batteries. Ionics [online], Online First. Available from: https://doi.org/10.1007/s11581-023-05272-9

• Journal Article Type: Article
• Acceptance Date: Oct 17, 2023
• Online Publication Date: Nov 9, 2023
• Deposit Date: Dec 15, 2023
• Publicly Available Date: Nov 10, 2024
• Journal: Ionics
• Print ISSN: 0947-7047
• Electronic ISSN: 1862-0760
• Publisher: Springer
• Peer Reviewed: Peer Reviewed
• DOI: https://doi.org/10.1007/s11581-023-05272-9
• Keywords: Lithium-ion battery; Charge state; Adaptive unscented Kalman filter algorithm; Adaptive forgetting factor recursive least squares method; Particle filter algorithm
• Public URL: https://rgu-repository.worktribe.com/output/2153261