Synergistic internal temperature estimation and fault diagnosis of lithium-ion batteries via online sequential extreme learning machine

Temperature is a critical indicator for safety monitoring of lithium-ion batteries. However, due to uncertain thermal diffusion from the cell interior to the surface, surface measurements cannot accurately or promptly reflect internal states. To address this limitation, this study proposes a novel internal temperature estimation method applicable under both normal operation and short-circuit fault conditions. A synergistic framework is developed based on the online sequential extreme learning machine, enabling simultaneous internal temperature estimation and fault diagnosis. An innovative experimental setup was established using controlled external short circuits (ESC) and shallow, slow nail penetration to trigger internal short circuits (ISC). Experimental results demonstrate that the proposed method achieves high accuracy, with maximum errors of 1.0963 °C, 3.0876 °C, and 2.2119 °C under normal, ESC, and ISC conditions, respectively. Moreover, the method successfully distinguishes between ESC and ISC based on distinct internal temperature dynamics, confirming its capability for reliable fault classification. These results highlight the method's promise for real-time fault detection and safety monitoring in lithium-ion battery systems.