2025, Vol. 7, Issue 2, Part C

The fast developments of renewable energy has increased the need for reliable and efficient energy storage, and lithium-ion batteries characteristic to a large extent in current applications. Accurate and fine prediction of state of charge (SoC), state of health (SoH), with optimized charging are still challenging. This work presents a machine learning based support that is the combination of prediction and smart charging. Comparative analysis shows that LSTM outperforms traditional models with R² = 0.97 and RMSE = 0.067. A reinforcement learning based charging scheme is benchmark with traditional CCCV charging, demonstrating a reduction of about 18% in charging time, an efficiency increase of 2.7%, and a longer cycle life of about 15%. The proposed approach highlights how advanced ML and RL can improve battery reliability, reduce the cost of storage (LCOS), and support large-scale renewable integration.