For the challenge of intricate the aging mechanism and the difficulty in measuring battery life, in-situ non-destructive diagnostic tests combined with Distribution of Relaxation Times (DRT) analysis capture the electrochemical process characteristics associated with time constants during battery cycle life. Subsequently, life prediction algorithm based on the Extreme Gradient Boosting (XGBoost) is developed. The prediction accuracy of dQ-dV, ECM (Equivalent Circuit Model), and DRT features is compared under dynamic conditions for two different batteries. The results indicate that the DRT analysis offers more precise mechanism characteristics than ECM and dQ-dV. The XGBoost-based estimation algorithm enhances the robustness of life prediction under dynamic conditions and improves the generalization across battery types.
Remaining Useful Life Prediction Based on Electrochemical Impedance Characteristic Distribution
2024-10-10
947630 byte
Conference paper
Electronic Resource
English
Prediction of Mining Railcar Remaining Useful Life
Springer Verlag | 2019
|Prediction of Remaining Useful Life for Aero-Engines
IEEE | 2021
|Remaining Useful Life Prediction Based on LSTM with Peephole for PEMFC
SAE Technical Papers | 2022
|Remaining Useful Life Prediction Based on LSTM with Peephole for PEMFC
British Library Conference Proceedings | 2022
|Remaining useful life prediction based on known usage data [5046-03]
British Library Conference Proceedings | 2003
|