Multi-source transfer network for cross-domain state of charge estimation of lithium-ion batteries

Abstract

The state of charge (SOC) in lithium-ion batteries plays a crucial role in battery management systems (BMS). Accurate SOC estimation is essential for extending battery life and ensuring safe operation. In recent years, many deep transfer learning methods have been applied to the field of SOC estimation. However, most existing approaches only consider single-source domains. In practical scenarios, multiple datasets may come from batteries with different materials or operate under varying conditions. These datasets not only differ from the target domain but also exhibit distributional differences among themselves. To leverage information from multiple source domains and enhance model performance in the target domain, this paper proposes a Multi-Domain Transfer Neural Network (MDTNN) framework. This method aligns the feature distributions between each source domain and the target domain in multiple feature spaces, enabling effective learning from multiple source domains and improving target domain performance. Extensive experimental results demonstrate that MDTNN outperforms other methods in various transfer scenarios, achieving accurate and robust SOC estimation.