Features fusion based diagnosis method for the internal short-circuit fault in a lithium-ion battery pack

The internal short-circuit (ISC) fault is one of the main faults, which results in thermal runaway of lithium-ion batteries. However, it is still challenging to detect the ISC fault when it comes to a battery energy storage system composed of battery packs. In response to this challenge, this paper proposes a novel ISC fault‑diagnosis method based on fused features that could reflect the transient response and steady-state characteristics of the ISC fault. The features are obtained by measuring the battery voltage and estimating the state of charge (SOC) via an unscented Kalman filter. Then, the features are normalized and fused using Euclidean distance for fault determination, in order to collectively improve reliability and prevent misdiagnosis. The results of internal short-circuit experimental tests demonstrate that, compared to an existing method based only on SOC, adding voltage can significantly improve fault‑diagnosis sensitivity and achieve at least 1.3 hrs earlier fault detection, with only a slightly increased computational cost. The proposed fault‑diagnosis method for battery packs shows great potential for practical application.