Estimation of State of Charge for Li-ion Batteries using Data Driven Technique

Accurate estimation of State of Charge (SoC) in Li-ion batteries using data-driven techniques is challenging due to the tedious and resource-consuming nature of real-time data generation. To address this issue, this paper proposes a data-light model for SoC prediction. The model is based on a hybrid architecture combining Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) networks. To optimize data usage, the model is trained with progressively decreasing amounts of data. Additionally, the number of nodes in the hybrid model is optimized to enhance the accuracy of SoC prediction. Experimental results demonstrate that the proposed hybrid model achieves accurate SoC estimation with significantly reduced data requirements, making it a promising solution for real-time Li-ion battery monitoring and management.