A multi-fault diagnosis method based on modified Sample Entropy for lithium-ion battery strings

• Published in: Journal of power sources Vol. 446; p. 227275
• Main Authors: Shang, Yunlong, Lu, Gaopeng, Kang, Yongzhe, Zhou, Zhongkai, Duan, Bin, Zhang, Chenghui
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 15.01.2020
• Subjects: Electric vehicles; Fault diagnosis; Lithium-ion batteries; Open circuit; Sample entropy; Short circuit; Fault diagnosis; Lithium-ion batteries; Sample entropy; Open circuit; Electric vehicles; Short circuit
• ISSN: 0378-7753; 1873-2755
• DOI: 10.1016/j.jpowsour.2019.227275

The conventional fault-diagnosis methods are difficult to detect the battery faults in the early stages without obvious battery abnormality because lithium-ion batteries are complex nonlinear time-varying systems with absolute cell inconsistency. Therefore, this paper proposes a real-time multi-fault diagnosis method for the early battery failure based on modified Sample Entropy. By detecting the modified Sample Entropy of the cell-voltage sequences in a moving window, the proposed diagnosis method can diagnose and predict different early battery faults, including short-circuit and open-circuit faults, and can also predict the time of the faults occurring. The experimental results and the comparison with the conventional methods verify the validity of the proposed solution with strong robustness, high reliability and low computational cost, and without the need of a precise model. In summary, the proposed multi-fault diagnosis approach is feasible and promising in real electric vehicle applications. •A multi-fault diagnosis method is proposed to detect the early battery faults.•The proposed diagnosis method is based on the modified Sample Entropy.•A coefficient is introduced to detect the fault type and time of fault occurring.•A moving window is used to maintain the detection sensitivity and less computation.•By optimizing the tolerance, the proposed method can prevent false detections.