Researchers have developed a two-step method for diagnosing micro short circuits in lithium-ion batteries, aiming to catch an early-stage fault that can eventually contribute to serious safety failures and thermal runaway. The new approach combines Hellinger distance with an Inverse Markov Method and is designed to detect subtle abnormal behavior even when the warning signals are partially hidden by complex charging and discharging conditions.
Micro short circuits, often abbreviated as MSCs, are especially difficult to diagnose because they do not necessarily produce the dramatic signals associated with a severe battery failure. Instead, they can begin as minor internal faults, creating small but dangerous anomalies that may evolve over time. In real battery systems, those anomalies are often masked by normal operating variation, dynamic driving cycles, and interactions among cells connected in series. That makes early diagnosis technically challenging, even though early detection could be crucial for preventing more serious safety events later on.
The researchers addressed this problem with a two-step diagnostic strategy intended to improve both sensitivity and reliability. In the first step, they estimated internal resistance using the forgetting factor recursive least squares, or FFRLS, method. They then used the Hellinger distance to compare each cell with a reference value and identify cells that were suspicious enough to warrant closer examination. This screening step is important because it narrows the search space and helps separate likely abnormal cells from the broader background of normal variation in a battery pack.
The second step then focuses on confirmation. For cells flagged as suspicious, the team applied an Inverse Markov Method to analyze anomalies in voltage transfer behavior. That means the method does not rely on a single instantaneous signal spike. Instead, it looks for abnormal transition patterns in the cell's voltage behavior, helping determine whether the observed irregularity is more consistent with a genuine micro short circuit than with ordinary operating fluctuation. The combination of the two methods is meant to provide a more robust diagnosis than either technique could deliver on its own.
To validate the approach, the researchers created a labeled dataset specifically designed for this problem. Micro short circuits were induced through slight extrusion, and the faulty cells were then operated in series with normal cells under the Urban Dynamometer Driving Schedule, or UDDS, cycle. This matters because the resulting data more closely reflect the dynamic conditions under which detection would be difficult in practice. By constructing a dataset containing both normal and MSC data segments, the team gave itself a way to test whether the diagnosis framework could distinguish subtle faults under realistic, non-static operating conditions.
According to the paper, the method performed strongly in those validation experiments. The authors report that the two-step strategy was able to identify MSC faults effectively, achieving a precision of more than 98.0%. That level of performance is notable because the target fault is both minor and safety relevant. A method that can reliably identify such early-stage abnormalities could help battery management systems intervene sooner, potentially reducing the chance that hidden internal damage progresses to a more dangerous state.
The broader value of the work lies in its contribution to battery safety management. Much attention in battery diagnostics has focused on faults that are already pronounced, but the earliest stages of failure can be the hardest and most important to detect. A diagnosis method that can operate under frequent charge-discharge cycling and identify mechanically induced MSC behavior could be useful in electric vehicles, energy storage systems, and other applications where lithium-ion batteries work under varying load profiles. It may also help improve maintenance strategies by identifying cells that appear healthy on the surface but are already beginning to deviate internally.
The study is also notable because it moves beyond purely theoretical diagnosis concepts and builds its validation around a fault dataset generated from mechanically stressed cells operating under a realistic driving cycle. That makes the work more relevant to practical battery monitoring, where algorithms must cope with dynamic, noisy, and highly variable signals rather than ideal laboratory traces.
Further work will still be needed before the approach can be considered ready for broad deployment in commercial battery packs. Additional validation across different chemistries, pack architectures, and real operating environments will matter, as will understanding how the method performs under noise, aging, and multiple fault types. Even so, the study offers a promising framework for early MSC detection. By combining statistical distance measurement with state-transition analysis, it suggests a practical route toward identifying one of the most elusive early warning signs in lithium-ion battery safety.
Reference
Author:
Hong Liang a , Renjing Gao a , Zeyu Chen b , Qingyi Tao b , Zilu Zhang c
Title of original paper:
Micro short circuit diagnosis for lithium-ion batteries based on Hellinger distance and Inverse Markov Method
Article link:
https://www.sciencedirect.com/science/article/pii/S2773153725001148
Journal:
Green Energy and Intelligent Transportation
DOI:
10.1016/j.geits.2025.100364
Affiliations:
a State Key Laboratory of Structural Analysis, Optimization and CAE Software for Industrial Equipment, Dalian University of Technology, Dalian 116024, China
b School of Mechanical Engineering and Automation, Northeastern University, Shenyang 110819, China
c Department of Mechanical Engineering, National University of Singapore, Singapore 117575, Singapore
Experimental study
Not applicable
27-Feb-2026