System Identification and State Estimation of a Reduced-Order Electrochemical Model for Lithium-Ion Batteries
Yujie Wang,Xingchen Zhang,Kailong Liu,Zhongbao Wei,Xiaosong Hu,Xiaolin Tang,Zonghai Chen
DOI: https://doi.org/10.1016/j.etran.2023.100295
IF: 11.9
2023-01-01
eTransportation
Abstract:Lithium-ion batteries commonly used in electric vehicles are an indispensable part of the development process of decarbonization, electrification, and intelligence in transportation. From intelligent designing, manufacturing to controlling, an intelligent battery management system plays a crucial role in the long life, high efficiency, and safe operation of lithium-ion batteries. As a first-principle model, the electrochemical parameters of the electrochemical model have physical meanings and reflect the internal state of the lithium-ion batteries. The application of electrochemical models in an advanced intelligent battery management system is a future trend that promises to mitigate battery life degradation and prevent safety incidents. The reduced-order electrochemical model is expected to alleviate the requirements of advanced battery management systems for high accuracy and fast computing of lithium-ion battery models. However, the existing model order reduction methods have the drawbacks of high computational complexity and small application scope, so that inconvenient to apply onboard. In order to solve the existing obstacles, this paper applies the pseudo-spectral method to solve the solid-phase diffusion equation, while the liquid-phase concentration equation is simplified by the Galerkin method. Subsequently, a particle swarm optimization algorithm is used to identify 11 parameters of the electrochemical model. To further improve the accuracy of the electrochemical model, the above system identification method is applied to segment identification, especially for high or low state-of-charge (SoC) conditions in this study. Finally, based upon the derived model, estimation of SoC is performed using a particle filter. The results show that the proposed reduced-order electrochemical model achieves a low Mean Absolute Error (MAE) of 8.4 mV and a MAE of 0.54 % on estimation of SoC based on the envisaged particle filter. This work is expected to provide the basis for the subsequent development of lithium-ion battery electrochemical models with smaller identification parameters and faster identification processes.