| As electric vehicle ownership increases,the problem of extended driving range of electric vehicle attracts more attention.Lithium-ion battery is an important energy storage unit of electric vehicle.More efficient management of lithium-ion battery can improve the energy utilization rate of the battery,which requires an effective evaluation of the energy state of the battery.In Battery Management System(BMS),battery state estimation is commonly achieved based on equivalent circuit model(ECM).However,ECM simulating characteristics of battery with electronic components lacks of description of electrochemical mechanism inside the lithium-ion battery,which contributes to limit of strong polarization simulation at low state of charge(SOC)range.To solve this problem,an optimized equivalent model over full-range SOC is established based on the simplification of the internal electrochemical processes of lithium-ion battery.In the proposed model,not only is solid-phase diffusion considered into the update of open circuit voltage(OCV)improving the simulation accuracy at low SOC range,but also the Butler-Volmer equation is used to describe the charge transfer reaction at the solid-liquid interface with the double layer capacitive effect considered.Besides,the proposed approximate method for describing internal electrochemical micro-variables with external electric macro-variables allows the model to avoid a series of partial differential equations.Thus,the proposed model is adaptive to dynamic working conditions with resonnable calculation.In this paper,a method of offline identification combined with online identification is proposed to identify the model parameters both in the current time domain and the predictive time domain.In the current time domain,for the parameters related to the solid-phase diffusion process,of which the response cannot be directly obtained,offline identification is carried out by fitting the response of the step current under different SOC,and other parameters are online identified based on the forgetting factor recursive least square(FFRLS)method.In the predictive time domain,for lack of the feedback of battery terminal voltage,the model parameters are calibrated over full-range SOC by fitting the response of the battery terminal voltage under hybrid pulse power condition(HPPC).The remaining discharge energy(RDE)and the remaining discharge time(RDT)are chosen as the evaluation indexes after analyzing the definitions and the influencing factors of SOE.And a prediction method is put forward combining battery current state estimation based on the Unscented Kalman Filter(UKF)algorithm with open-loop terminal voltage sequence prediction based on the proposed model.In the current time domain,the closed-loop estimation of the battery real-time SOC based on the UKF algorithm is used to achieve a high-accuracy estimation of the current battery terminal voltage.The following average SOC is predicted according to the current state and future working conditions,which is used to determine the model parameters in the predictive time domain.The terminal voltage sequence is predicted based on the proposed equivalent model,which determines the discharge cut-off point of the battery.Then RDE and RDT are calculated based on the predicted terminal voltage sequence.Litium-ion battery experiments are carried out on the experiment bench.Based on the experiment data of HPPC and dynamic stress test(DST),the terminal voltage output accuracy is verified.The results show that the proposed model is proven to provide better performance over full-range SOC especially at the low-range area below 20%,compared with the runtime-based model(RTM).The energy state prediction method proposed in this paper is varified based on the constant current discharge experiment.The RDE error obtained from the terminal voltage sequence based on the improved model was less than 0.05 Wh,which was about 90% lower than that based on the RTM.And the average error of RDT is 10.05 s,which is much lower than that based on RTM. |
PDF Full Text Request