Long Short-Term Memory With Attention Mechanism for State of Charge Estimation of Lithium-Ion Batteries

Abstract

Evaluating the state-of-charge of the battery's current cycle is one of the major tasks in the charge management of rechargeable batteries. We propose a long short-term memory model with an attention mechanism to estimate the charging status of two lithium-ion batteries. Data from three dynamic tests such as dynamic stress test, supplemental federal test procedure-driving schedule, and federal urban driving schedule are used to evaluate our model at different temperatures. One dataset or two datasets are used as the training data, and the other datasets are used as the test data. The model achieves the predictive root mean square errors of 0.9593, 0.8714, and 0.9216 at three different temperatures for the FUDS dataset. Moreover, the predictive RMSE of the proposed model is lower than 1.41 for all our experiments. We use the Monte Carlo dropout technique to verify the robust of the proposed model.