Improved adaptive feedback particle swarm optimization-multi-innovation singular decomposition unscented Kalman filtering for high accurate state of charge estimation of lithium-ion batteries in energy storage systems

• Published in: Ionics Vol. 30; no. 9; pp. 5411 - 5427
• Main Authors: Li, Yang, Wang, Shunli, Liu, Donglei, Liu, Chunmei, Fernandez, Carlos, Wang, Xiaotian
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.09.2024; Springer Nature B.V
• Subjects: Accuracy; Algorithms; Artificial intelligence; Batteries; Chemistry; Chemistry and Materials Science; Condensed Matter Physics; Decomposition; Electrochemistry; Energy Storage; Feedback; Innovations; Kalman filters; Lithium; Lithium-ion batteries; Machine learning; Methods; Optical and Electronic Materials; Optimization; Optimization algorithms; Parameter identification; Particle swarm optimization; Real time; Renewable and Green Energy; Simulation; State of charge; Storage systems; Lithium-ion batteries; Multi-innovation singular decomposition UKF; Adaptive feedback particle swarm optimization; Second-order RC equivalent circuit model; SOC
• ISSN: 0947-7047; 1862-0760
• DOI: 10.1007/s11581-024-05663-6

Accurate estimation of the state of charge (SOC) of lithium-ion batteries is very important for the development of energy storage systems. However, batteries are subject to characteristic changes in complex environments, making it difficult to accurately estimate SOC online. In this paper, an adaptive feedback particle swarm with multi-innovation singular decomposition unscented Kalman filtering method is proposed. The idea of the real-time change of inertia weight and learning factor is used to balance the particle searchability, and the information feedback mechanism is established to make the local optimal position constantly updated, which solves the problem that the standard particle swarm optimization algorithm is easy to fall into the local optimal solution. Singular decomposition (SVD) is used to replace Cholesky decomposition in traditional UKF to avoid algorithm divergence. At the same time, a strategy of noise variance Q varying with multi-time errors is introduced to further improve the estimation accuracy. The results show that under different working conditions, the SOC estimation accuracy based on adaptive feedback particle swarm optimization and multi-information singular decomposition unscented Kalman filter is improved by 76.6% and 67.6% respectively, and the algorithm convergence speed is improved by 88.9% and 77.5%, respectively.