To ensure the reliability and safety of lithium-ion battery operation and to monitor its health status in a timely manner

an adaptive feature-aware battery health state fusion estimation model based on the transposed Transformer is proposed. This model is built on the Autoformer and iTransformer models and combined with a linear regression model. First

health factors are extracted from the charging curve. Then

the capacity degradation is decomposed into a degradation trend part and a capacity regeneration part. The degradation trend of battery capacity is predicted by using a linear regression model

and the capacity regeneration part is estimated by using a transpose Transformer model. The combination of these two parts provides an estimate of the battery capacity degradation. Finally

attention weights are utilized to provide interpretability to the model. The research results indicate that this method can be adopted to significantly reduce the prediction errors compared to other time series prediction models in simulation experiments on the NASA lithium battery aging dataset

thus verifying the prediction accuracy and reliability of the proposed method. This paper provides a reference for further in-depth research on accurate estimation of battery health state.