Thermal Coupling Simplified First Principle Model And SOC Estimation For Lithium-Ion Battery

Accurate state of charge(SOC)estimation for lithium-ion battery is one of important functions for battery management system(BMS)with high performance.Compared with equivalent circuit model,first principle(FP)model can describe battery electrochemical and dynamic processes more accurately in micro scale,and it can be utilized in BMS to further enhance the accuracy of SOC estimation.However,the complexity of the model and parameter identifiability is a great limit,and FP model fails to simulate battery external behaviors accurately at different ambient temperatures.This work focuses on establishing a thermal coupling simplified first principle(TC-SFP)model and developing a model-based SOC estimator with high accuracy.To address the problem of the reduction of SOC estimation accuracy along with battery aging,model parameter update methods are also developed to ensure the accuracy of SOC estimation during battery life cycle.Firstly,based on an improved single particle model,the formulas of battery internal average and surface temperatures are obtained through a thermal resistance model.Considering the effects of the variation of temperature on electrochemical paramters and internal processes,a TC-SFP model is developed.According to the basic theory of excitation response analysis and the requirement of parameter decoupling identification,multi-step identification conditions including a 0.02 C-rate discharge condition and shorttime constant charge and discharge condition at three ambient temperatures and so on,are specially designed to obtain 17 model parameters accurately.Secondly,a framework of SOC estimation based on TC-SFP model is developed,and SOC is separately estimated through extended kalman filter algorithm,particle filter(PF)algorithm and dichotomy algorithm.The accuracy,computational efficiency,astringency and anti-interference performance of the three algorithms are assessed and SOC estimation results based on TC-SFP model and PNGV model are also compared.The comparison results of performance of different SOC estimation methods indicate that TC-SFP model-based SOC estimation method using dichotomy algorithm has more advantages.To address the problems of multiple model parameters and complex identification conditions,model parameter sensitive analysis is conducted.9 sensitive parameters which are needed to estimate accurately during battery life cycle are selected according to their sensitivities and identification difficulty.An improved identification condition with alternant charge and discharge processes under various C-rates is designed to identify the sensitive parameters.The accuracy of SOC estimation using the identified sensitive paramters during battery life cycle is assessed to validate the effectiveness of the sensitive parameter identification method.To update the sensitive parameters during battery life cycle,a multi-time update method which combines offline prediction and capacity parameter online estimation is developed.9 sensitive parameters are obtained at different aging stages and their predictive values are calculated via different fitting functions.An online estimation method based on PF algorithm combined with external feature extraction is developed to update 3 capacity parameters with the predicted results of other sensitive parameters.Validation results at two C-rates aging indicate that the developed multi-time update method can ensure the SOC estimation accuracy during battery life cycle.