Online diagnostic method for fault behavior of lithium batteries with secondary utilization

With the continuous increase in battery deployment, the repurposing scale of retired batteries in energy storage and other secondary utilization scenarios has expanded significantly. However, performance degradation during cycling may elevate risks such as thermal runaway, posing critical safety concerns. This study addresses the safety concerns in retired lithium iron phosphate (LFP) battery echelon utilization by establishing a online failure diagnosis system based on mechanism model parameters. First, a battery model is established and an online collaborative identification method is constructed using the dual adaptive extended kalman filter (AEKF) algorithm. Subsequently, the evolution patterns of the model parameters are quantitatively analyzed, and a fault behavior diagnostic framework is proposed through a fault boundary construction method. Finally, the reliability of the failure boundary-based diagnostic method is validated through destructive analysis techniques encompassing scanning electron microscopy (SEM) and X-ray photoelectron spectroscopy (XPS). The results show that the voltage prediction errors of MAE less than 20 mV and RMSE less than 26 mV remain within the acceptable thresholds when voltage prediction is performed using the online identification parameters. The diagnostic accuracy of the failure behavior diagnostic method for the detected anomalies is 94.1%, and the leakage rate for all sample points is 8.4%, indicating that the failure behavior diagnosis method based on failure boundary has strong reliability and accuracy.