Data-physics-driven estimation of battery state of charge and capacity

Abstract

High-power density lithium-ion batteries have been utilized in both energy storage and high rate charging and discharging applications. Accurate state estimation is fundamental to enhancing battery life and safety. Therefore, a data-physics-driven estimation of the state of charge and capacity for lithium-titanate batteries was conducted using Gaussian distribution fusion. Firstly, a fractional order model was selected as the physical analytical model for lithium-titanate batteries. Secondly, a data-driven model that combines convolutional neural networks and long short-term memory networks was employed to predict the battery state of charge. Thirdly, the physical analytical model and data-driven model were fused to estimate the state of charge by employing the principle of Gaussian distribution fusion. Finally, both the state of charge and actual capacity of the battery were jointly estimated. The results showcase the capability of the proposed method to accurately estimate the state of charge and capacity, ensuring an accuracy level within 1%. Furthermore, this approach exhibits a 20% improvement in accuracy compared to traditional methods.