HPPC参数辨识时间域对电池模型精度影响的研究

Study on the influence of time domain of HPPC parameter identification on accuracy of the battery model

为保证电动汽车电池系统的安全性,需配备电池管理系统(battery management system,BMS)对电池进行状态估计,而准确的电池状态估计依赖于合适的电池模型以及精确的模型参数。为了研究混合脉冲功率性能测试(hybrid pulse power characteristic,HPPC)实验的不同阶段辨识参数对电池模型精度的影响,分别在3个温度下,对索尼US18650VTC6锂离子电池进行HPPC实验,并通过对HPPC实验的2个阶段(10 s零状态响应阶段、40 s零输入响应阶段),提取Thevenin等效电路模型的参数,然后通过0.5 C、1 C恒流放电工况进行验证。结果表明:不同温度下模型精度不同,模型精度与温度存在相关性。通过偏差分析,排除了放电过程中电池温升以及连接阻抗的影响。为了进一步解决5、25℃下仿真与实验结果之间存在偏差的问题,利用HPPC实验的6 min阶段,通过改变时间域进行参数辨识。5℃下,0.5 C恒流放电工况平均相对误差从1.49%降低到40 s及60 s时间域下的0.55%,结果表明通过改变HPPC参数辨识时间域的方法可以有效提高不同温度下的模型精度。

In order to ensure safety of battery packs in electric vehicles,a battery management system(BMS)is required to estimate the state of battery packs.The state estimation accuracy depends on a suitable battery model and accurate model parameters.The offline parameter identification is based on the measured data in a laboratory.It can be used to quickly analyze the correlation between the change law of the battery parameters and the battery performance.It can also provide a reference for online parameter identification.Offline parameter identification is mostly carried out through Hybrid Pulse Power Characteristic(HPPC)experiment.Based on the previous discussion of offline parameter identification,most of the researchers conduct parameter identification through the zero-input response on“rebound characteristic”stage or zero-state response stage of the HPPC experiment,focusing on the establishment of different equivalent circuit models.Some researchers also find that the pulse time and standing time of the HPPC experiment affect the accuracy of the battery model,and the battery model parameters need to be modified at a low temperature.In order to study the effects in different stages of HPPC test on the accuracy of battery model,the HPPC tests are performed on Sony US18650VTC6 lithium-ion batteries at 5℃,25℃and 55℃respectively.Through the two stages(10 s zero-state response stage and 40 s zero input response stage)of the HPPC test to get the model parameters of the Thevenin equivalent circuit model,the parameters are imported into the equivalent circuit model based on Simscape in Matlab/Simulink,and then verified under 0.5 C and 1 C discharge conditions.In the process of verification,the dynamic stress test(DST)condition is also verified.Through comparison,it is found that the discharge condition is more conducive to describing the deviation phenomenon than the DST condition.At 5℃and 25℃,there is a relatively stable voltage deviation between the simulation value and the experimental value under discharge conditions.At 55℃,the simulation value is basically close to the experimental value.During the experiment,temperature rise of the battery is not very high,and slight temperature rise mainly occurs in the later stage of discharge.Through an analysis of the overall deviation of the voltage platform period under discharge conditions,there is no correlation between the deviation and the discharge rate.The influence of battery temperature and connection impedance during discharge can be excluded.Results show that the accuracy of the model at different temperatures is different.There is a correlation between accuracy of the model and temperature.According to the characteristics of the lithium-ion battery,the interior of the battery is an organic electrolyte system.With an increase of battery temperature,electrochemical reaction inside the battery and lithium-ion diffusion both accelerate,and polarization internal resistance reduces.At a low temperature,polarization time of the battery is longer than that at a high temperature.In order to further solve the problem of the deviation at 5℃and 25℃,the parameter identification is carried out by changing the time domain during the 6-min stage of HPPC test.The parameters are identified in three time domains of 20 s,40 s and 60 s respectively.At 5℃,the average relative error at 0.5 C discharge current reduces from 1.49%to 0.55%in the 40 s and 60 s time domains,and the average relative error under 1 C discharge condition is only 0.42%.Similarly,the accuracy of the model apparently improves at 25℃.The results show that different time domains should be selected for parameter identification at different temperatures.The model accuracy at different temperatures can be effectively improved by changing the HPPC parameter identification time domain method.