A fractional-order model-based battery external short circuit fault diagnosis approach for all-climate electric vehicles application

The impact of SOC and temperature on external short circuit (ESC) faults characteristics of lithium-ion batteries, including the current and voltage variation and temperature increase, are analyzed. A fractional-order model (FOM) and a first-order RC model are both employed to describe the electrical behavior of the battery cells with the ESC fault. While the model parameters are identified by the genetic algorithm (GA). A comparison study is made on the prediction accuracy for the two models. An effective classification method based on a random forests (RF) model is proposed to recognize the electrolyte leakage behavior that occurs during the ESC fault experiments. Based on the above efforts, the three steps model-based diagnosis algorithm for identifying the ESC fault and even electrolyte leakage of the battery in real-time is proposed. Two indicators of the root mean square error (RMSE) of battery predicting voltage are applied to diagnose for the ESC fault only and ESC-leakage merged fault. The result of the leakage condition is obtained by a pre-trained RF classifier to confirm the leakage detection result based on the RMSE indicator. Several cases are verified that all the ESC cells can be diagnosed efficiently.