Fault-Tolerant Fusion Framework for Robust State-of-Charge Estimation in Lithium-Ion Battery Packs
点击次数：
影响因子：7.0
DOI码：10.1109/TIM.2026.3676196
教研室：R. Yang, Z. Sun, M. Lin, X. Liu, J. Wu
发表刊物：IEEE Transactions on Instrumentation and Measurement
关键字：Closed-loop fusion; fault-tolerant; lithium-ion battery pack; representative cells; state of charge (SOC)
摘要：Lithium-ion batteries are widely adopted in electric vehicles (EVs) due to their excellent safety and long cycle life. However, inconsistencies in capacity, internal resistance, and aging among individual cells are common throughout the battery lifecycle, which poses challenges for battery management systems (BMSs) in achieving accurate and reliable state-of-charge (SOC) estimation. In addition, to ensure robustness under sensor noise and abnormal conditions, fault-tolerant performance is increasingly critical for practical deployment. Here, we propose a data-driven and filter fusion-based algorithm with strong fault-tolerant capability for battery pack-level SOC estimation. First, representative cells are selected using a differential voltage (DV) method to reliably identify the cells most representative of the pack’s capacity and SOC. Subsequently, the voltage and current data from the selected cells serve as inputs to dual parallel gated recurrent unit (PAGRU) networks, establishing a mapping between pack SOC and voltage–current characteristics. Finally, by integrating the model with a Kalman filter (KF), a closed-loop estimation framework is constructed to suppress noise-induced deviations and enhance reliability under abnormal operating conditions. Experimental results demonstrate that the proposed method achieves accurate SOC estimation and maintains high performance under unknown operating conditions, sensor faults, and noise disturbances, offering strong fault tolerance and reliability for real-world applications.
论文类型：期刊论文
学科门类：工学
文献类型：J
卷号：75
页面范围：9002416
是否译文：否
发表时间：2026-03-20