Detection and quantitative assessment of internal short circuits in lithium-ion battery packs based on differential voltage analysis and mahalanobis distance

Accurate and prompt diagnosis of internal short circuits at an early stage is critical for preventing severe safety incidents and ensuring the reliability and safety of lithium-ion batteries. However, existing early-stage internal short-circuit diagnosis methods often rely heavily on high-precision battery models and large volumes of high-quality labeled training data, limiting their practicality and robustness in real-world applications. To address these limitations, this paper proposes a novel method for early detection and quantitative assessment of internal short circuits in lithium-ion battery packs, based on differential voltage (DV) analysis and Mahalanobis distance. The proposed approach extracts a median DV curve from the sorted terminal voltages of individual cells within a battery pack, which serves as a reference to characterize the normal cell behavior. The Mahalanobis distance between each cell's DV curve and the reference curve is then calculated and compared against a threshold to distinguish short-circuited cells from healthy ones. For the identified faulty cells, the short-circuit current and resistance are estimated by analyzing the differences between charging voltage curves across adjacent cycles, enabling precise quantification of fault severity. Experimental validation is conducted using simulated internal short circuits with varying severities. Results show that the proposed method can accurately detect short-circuited cells when the short-circuit resistance is less than or equal to 300 Ω. The maximum and minimum relative errors of short-circuit resistance estimation are 5.21 % and 1.20 %, respectively, demonstrating the effectiveness and accuracy of the proposed method. • A lightweight diagnostic method for early ISCs in battery packs is proposed. • The method does not require the development of battery models. • The method also eliminates the need for complex training procedures. • The method is characterized by low computational complexity and ease of implementation. • Moreover, the method enables accurate quantification of short-circuit severity.