A Novel Method for Estimating the State of Charge of Lithium Batteries Adopts a Bidirectional Gated Recurrent Unit With Attention Mechanism and Piecewise Estimation

Accurate estimation of the state of charge (SOC) in lithium-ion batteries is fundamental to ensuring their safe and reliable operation. however, the influence of factors such as battery state, discharge conditions, and environmental temperature poses significant challenges to achieving precise SOC estimation. To tackle these challenges, this paper is proposed a hybrid deep learning framework for SOC estimation that integrates the bidirectional information extraction capabilities of Bidirectional Gated Recurrent Units (BiGRU) with the adaptive feature weighting mechanism of the attention mechanism, forming the BiGRU-Attention model. To further enhance the model’s performance under varying temperatures and operational conditions, the Whale Optimization Algorithm (WOA) is employed to optimize its hyperparameters. Additionally, a redefined SOC calculation method to mitigate the impact of data disturbances caused by incomplete discharge cycles and battery aging. Furthermore, piecewise estimation addresses the problem of increasing root mean square error (RMSE) in low-temperature environments. Experimental results demonstrate that the proposed method achieves highly accurate and robust SOC estimation, outperforming existing approaches. The RMSE is maintained below 0.8%, while the maximum absolute error is constrained to within 4.3%, highlighting the superiority of this approach and offering a novel perspective for SOC estimation of lithium-ion batteries.