Thermal Runaway Prediction of Lithium-ion Batteries based on Fusion Model

Implementing a proactive alert system to detect thermal runaway in lithium-ion batteries is essential for ensuring the safety and dependability of energy infrastructures. In this research, a new approach is introduced that integrates an electrothermal coupling model with deep learning techniques to detect thermal runaway in its early stages. The proposed frameworks incorporate Convolutional Neural Networks (CNN), Bidirectional Long Short-Term Memory (BiLSTM) systems, and sequence-to-sequence neural models augmented with attention mechanisms. First, an electrothermal coupling model is constructed to simulate internal battery temperature changes. Subsequently, deep learning algorithms are employed to estimate battery voltage and surface temperature metrics. Finally, an internal temperature prediction model is established based on these predicted values. Experimental validation shows that this method controls the internal temperature prediction error within $3.4^{\circ} \mathrm{C}$. It effectively achieves thermal runaway early warning and significantly improves the operational safety of lithium-ion batteries.