Data-driven SOC Estimation with Adaptive Residual Generator for Li-ion Battery

Xiaoyi Xu,Cong-Sheng Huang,Mo-Yuen Chow,Hao Luo,Shen Yin

DOI: https://doi.org/10.1109/iecon43393.2020.9255139

2020-01-01

Abstract:Lithium-ion batteries are widely used in many fields of modern life, e.g. wearable devices, electric vehicles and electric grids, etc. The safety and reliability of the lithium-ion battery are critical issues during the battery operation, where the battery management system (BMS) plays a key role. An accurate estimation of the state-of-charge (SOC) of the battery is essential for the BMS. However, due to the intrinsic nonlinearity of the lithium-ion battery, the accurate estimation of the SOC is technically challenging and has drawn lots of attention both from academic and industrial fields. In order to tackle this difficulty, many SOC estimation approaches have been proposed, in which an identification method for the parameters of the battery is normally implemented. However, the additional parameter identification approach greatly reduces the efficiency of SOC estimation and the bias from identification may significantly affect the accuracy of the SOC estimation. This paper proposes a novel data-driven SOC estimation approach based on the adaptive residual generator, which realizes integrating the parameter identification and the SOC estimation into a simultaneous procedure, where the convergences for both the parameter identification and SOC estimation are guaranteed. The proposed adaptive residual generator can estimate the SOC of the battery accurately due to real-time parameter identification that proactively minimizes the modeling error. The effectiveness and the performance of the proposed method are demonstrated through the case studies on a battery simulator. Also, owing to accurately identified parameters, the SOC of the battery is estimated accurately with almost 0% SOC estimation error.