A Dynamic Spatial-Temporal Attention-Based GRU Model With Healthy Features for State-of-Health Estimation of Lithium-Ion Batteries

A proper battery management system (BMS) plays a vital role in ensuring the safety and reliability of electric vehicles (EVs) other electronic products. Accurate State-of-Health (SOH) estimation Lithium-ion (Li-ion) batteries is key factor BMS. It difficult to determine SOH because complexity electrochemical reactions within battery. To improve accuracy estimation, dynamic spatial-temporal attention-based gated recurrent unit (DSTA-GRU) model proposed this paper. First, we extract six features from battery's charging discharging processes that can reflect aging degree some extent. Second, paper proposes combine spatial attention temporal not only consider effects states at different time step on results, but also space domain. Third, trained tested NASA datasets compared with conventional models. Experiments carried these data sets demonstrate our achieves higher than