Mahalanobis-Taguchi method based anomaly detection for lithium-ion battery

• Published in: Electrochimica acta 2024-03, Vol.479, p.143890, Article 143890
• Main Authors: Shimura, Jusuke, Takata, Daisuke, Watanabe, Hikari, Shitanda, Isao, Itagaki, Masayuki
• Format: Article
• Language: English
• Subjects: Anomaly detection; Electrochemical impedance spectroscopy; Lithium-ion battery; Mahalanobis-Taguchi method
• ISSN: 0013-4686; 1873-3859
• DOI: 10.1016/j.electacta.2024.143890

•Can find a weird single cell that is not aligned with other cells in the module.•Mahalanobis-Taguchi method was used for anomaly detection.•Seven impedance parameters were considered to describe the cell.•Considered new and thermally degraded cells as normal and abnormal ones.•Cells with a capacity loss ratio of 5 % or less could be detected. Lithium-ion batteries (LIBs) are power sources that contain many single cells, and are commonly used in electric vehicles and energy storage systems. In order to improve the safety of LIBs for large-scale applications, it is important to establish a high-performance anomaly detection method that can detect abnormal cells among a large number of normal cells. In this study, the Mahalanobis-Taguchi (MT) method was used for anomaly detection by considering new and thermally degraded cells as normal and abnormal ones, respectively. New and thermally degraded cells were prepared, their electrochemical impedance spectra were obtained using a frequency response analyzer, and their degree of anomalies were calculated using the corresponding impedance parameters, which were obtained by fitting, using an equivalent circuit. Thermally degraded cells with only 4.7 % capacity loss could be detected with high sensitivity using seven impedance parameters. [Display omitted]