Parameter Identification of Lithiumion Battery Electrochemical Model Using Sensitivity Analysis and Neural Networks

Le résumé fourni par la source

The electrochemical model can accurately reflect the internal state of lithium-ion battery, showing great potential in the electric vehicles and energy storage industry. However, strong nonlinearity of the model makes it challenging to obtain the accurate parameter values. This paper presents a comprehensive parameter identification framework based on electrochemical pseudo-two-dimensional model. Firstly, parameter sensitivity analysis is performed under constant current discharge profile and hybrid pulse power characteristic condition and parameters are divided into 3 groups. Secondly, deep learning method is used to obtain the initial values of these parameters which can improve the convergence of model. Finally, using optimization algorithm to estimate the values of parameters to improve the accuracy of the model. The verification results show that there are 14 parameters with high sensitivity in the electrochemical model and these parameters are estimated accurately. The maximum root mean square error of terminal voltage between simulation and experiment under all conditions is less than 53mV.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.