Redundant Thermal Fault Diagnosis and Localization for Lithium-ion Battery Strings

Detecting and addressing thermal fault in battery systems is crucial for averting potential safety hazards. A novel redundant thermal fault diagnosis and localization method is proposed in this paper. This approach combines the indirect electrical and thermal characteristics to accurately identify the faulty cell in a battery string under sparse temperature sensing. An electro-thermal coupling model integrated with back propagation neural network (BPNN) is established to depict the electrical and thermal behavior. Based on this model, two proportional integral observers (PIOs) are utilized to estimate the residual internal resistance and thermal fault power in real-time. Several experiments are conducted regarding the thermal fault occurred in one of the cells to validate the effectiveness and robustness of the proposed method. The results show that the thermal fault can be rapidly detected upon triggering, and the faulty cell can be located accurately from both the electrical and thermal characteristics under various operating conditions.