Research On Electrochemical Model And SOC Estimation Of Lithium-ion Battery With Temperature And Current Dependence

Electric vehicles,including hybrid electric vehicles(HEVs),plug-in hybrid electric vehicles(PHEVs)and pure electric vehicles(PEVs),have been widely developed due to the explosion of the energy crisis and global warming.Recently,compared with other batteries,lithium-ion batteries(LIBs)have been widely utilized not only for their high density,high power density,long cycle life and environmental friendliness,but also for their low self-discharge rate and lack of a memory effect.Reliability and safety issues coupled with the above mentioned advantages are challenging and crucial for LIBs,especially under extreme conditions.To address this problem,a battery management system(BMS)was proposed to monitor and protect the battery.In the design of a BMS,the accurate estimation of state of charge(SOC)can not only prevent battery from over-charge/discharge but also optimize battery performance,extend cycle life and increase driving range.SOC cannot be directly evaluated.Accordingly,various battery models are employed to estimate SOC,which are closely dependent on the related battery parameters.Compared with the equivalent circuit models,the electrochemical models are more accurate and can capture important dynamics,including solid-phase diffusion,at the expense of computational resources,but they are more sophisticated and unsuitable for online applications.Thus,a simplified electrode-average electrochemical model combined with extended Kalman filtering algorithm(EKF)is proposed to estimate SOC of LIBs.Firstly,two assumptions based on the pseudo-two-dimensional(P2D)model are proposed:the concentration of Li~+in the electrolyte is constant,and the solid concentration distribution along the electrode is negligible.A simplified electrode-average model with a solid-phase diffusion equation reduced by polynomial approximation and a three-variable method is proposed,which not only captures the dynamic behaviour of the battery but also simplifies the physics-based equations expressing concentration transport and conservation of charge for the solid and electrolyte phases to reduce the model's complexity.Secondly,a novel parameter identification method considering temperature and current is also proposed to reduce the parameter deviation caused by different working conditions.The model parameters are identified by the genetic algorithm(GA)offline at different temperatures and currents to create lookup tables for online estimation.To remove the redundant parameters and adjust the model,local sensitivity analysis of model parameters is employed.Thirdly,3.5 Ah 18650-type cells are chosen to validate the simplified model and the proposed estimation method through several operating conditions.The experiment platform and process are introduced in detail.The results suggest that the simple method possesses low complexity,sufficient accuracy and excellent adaptability as temperature and current rate changing.The simplified model combined with the proposed parameter identification method updates the parameters with a voltage error of 0~0.056 V under current rate<1C and an average relative error of 0.2%under DST test.Finally,the simplified electrochemical model combined with EKF is proposed to estimate the battery SOC and compared with coulomb counting(CC)method.1C pulse discharge test and DST test are utilized to verify the efficiency of the proposed method respectively.The results show that the EKF algorithm has higher precision compared with the CC method with the maximum SOC error of 2.72%and average error of 0.61%.The initial value dependence and cumulative error in CC method are not generated.The proposed method possesses the self-correct ability and can quickly converge to the true value.