Electric vehicles (EVs) have emerged as a promising trend for future development. Serving as the core energy source for EVs, lithium-ion batteries offer advantages. Accurate SoC estimation is vital when you have to take care of battery management. A recent breakthrough study presented by researchers from the Chang'an University and RWTH Aachen University introduces an improved model combining machine learning and Kalman filtering architecture for state of charge estimation of lithium-ion batteries. This advanced method not only achieves accurate online SOC estimation and improves the safety of Lithium-ion batteries.
SOC estimation method is also a key technology component in battery management systems. Generally, it can be categorized into four main classes namely the ampere-hour integration (AHI) method, parameter-based approaches, model-based methods, and data-driven methods. Now, numerous scholars have employed hybrid approaches for battery SOC estimation. As one of the hybrid approaches, the improved SOC estimation model can address strongly nonlinear problems without requiring prior knowledge of the battery's physical characteristics and parameters. And, it can deal with the issue of poor model generalization and limited robustness.
This research proposes a closed-loop SOC estimation method based on simulated annealing-optimized support vector regression (SA-SVR) combined with minimum error entropy based extended Kalman filter (MEE-EKF) algorithm. Firstly, a probability-based SA algorithm is employed to optimize the internal parameters of the SVR, thereby enhancing the precision of original SOC estimation. Secondly, utilizing the framework of the Kalman filter, the optimized SVR results are incorporated as the measurement equation and further processed through the MEE-EKF, while the ampere-hour integral physical model serves as the state equation, effectively attenuating the measurement noise, enhancing the estimation accuracy, and improving generalization ability.
The results demonstrate that the improved method achieves a mean absolute error below 0.60% and a root mean square error below 0.73% across all operating conditions. Compared to traditional methods such as Back propagation (BP) Neural Network, Long Short-Term Memory (LSTM) with strong sequence processing capabilities, the improved method exhibits the smallest errors and significantly better estimation accuracy;
In the future, researchers need to explore the high-precision estimation of SOC with strong robustness in the presence of complex noise. To ensure that the method performs reliably and accurately in practical environments, the parameters of the SOC estimation module can be periodically updated based on the battery's operating and aging conditions, addressing robustness concerns during real-world operation. Besides, technologies such as digital twins and the Internet of Things can further empower this approach for advanced applications.
Reference
[1] Hu X, Feng F, Liu K, Zhang L, Xie J, Liu B. State estimation for advanced battery management: key challenges and future trends. Renewable Sustainable Energy Rev 2019;114:109334. https://doi.org/10.1016/j.rser.2019.109334 .
[2] Tian J, Xiong R, Shen W, Lu J. State-of-charge estimation of LiFePO4 batteries in electric vehicles: a deep-learning enabled approach. Appl Energy 2021;291:116812. https://doi.org/10.1016/j.apenergy.2021.116812.
[3] Wang Y, Kang X, Chen Z. A survey of digital twin techniques in smart manufacturing and management of energy applications. Green Energy Intell Transp 2022;1(2):100014. https://doi.org/10.1016/j.geits.2022.100014.
Author: Yan Li, Min Ye, Qiao Wang, Gaoqi Lian, Baozhou Xia
Title of original paper: An improved model combining machine learning and Kalman filtering architecture for state of charge estimation of lithium-ion batteries
Article link: https://doi.org/10.1016/j.geits.2024.100163
Journal: Green Energy and Intelligent Transportation
https://www.sciencedirect.com/science/article/pii/S277315372400015X
