Research on the Method of Remaining Useful Life Prediction of Lithium-Ion Battery Based on LSTM

Lithium-ion batteries are prone to capacity degradation during use, leading to a decline in their reliability and potentially causing catastrophic failures. This necessitates the study of advanced methods for estimating the lifespan of lithium-ion batteries. To address this need, this paper conducts research in three areas: the extraction of degradation features, the construction of health factors, and the development of predictive models, proposing a comprehensive framework for estimating the remaining useful life of lithium-ion batteries. Firstly, the general trends of external feature parameters as a function of cycling are observed to identify the features that indicate battery degradation. Next, the relationship between these features and battery capacity is determined using Grey Relation Analysis, allowing for the construction of an indirect health factor. Finally, a data-driven prediction method is employed to construct a predictive network model using Long Short-Term Memory networks, with the health factor used for training to accurately predict the health state of the lithium-ion battery. The effectiveness of the proposed remaining life prediction method is validated using the NASA Battery Set dataset from NASA Ames Research Center.