Research On SOC Estimation Method Of Lithium Battery Based On On-line Identification Of Multi Innovation Least Square Method

The development of new energy vehicles is the only way for China to move from a big automobile country to a strong automobile country.It is an important strategic action to adapt to climate change and promote green environmental protection.At present,the most mature development of new energy vehicles is electric vehicles,electric vehicles need battery management system for energy and safety management,and the State of Charge(SOC)of the battery is one of the key parameters of the battery management system of electric vehicles,affecting the vehicle performance and safety.Therefore,accurate estimation of battery capacity is of great theoretical and engineering significance for the development of electric vehicles.This paper takes nickel-cobalt-Manganese ternary lithium battery as the research object,studies the online identification method of the battery equivalent circuit model parameters,and realizes the real-time accurate estimation of battery SOC with Extended Kalman Filter(EKF).The main research work of this paper includes:(1)A ternary lithium battery test platform was established,and capacity characteristics tests were carried out on ternary lithium batteries under the conditions of temperature 0℃,5℃,15℃,25℃,35℃ and charge and discharge rates of C/10,C/5,C/3,C/2,7C/10,and 1C,respectively.The test results show that the higher the temperature is,the lower the charge-discharge ratio is,and the higher the maximum discharge capacity is.Based on the experimental data,the charging and discharging capacity correction coefficient was introduced to establish the relationship between the charging and discharging capacity correction coefficient and the temperature and the rate,so as to realize the online correction of the ampere-hour integral calculation formula in the process of battery SOC estimation.Open-circuit voltage tests were carried out on ternary lithium batteries at ambient temperatures of 5℃,25℃ and 35℃,and the internal resistance characteristics of ternary lithium batteries were studied through experimental analysis.The test results show that the average open circuit voltage at 5℃ decreases about 0.4%compared with that at 25℃ and 35℃.The internal resistance of the battery also presents different characteristics at different temperatures.The ohm resistance at 5℃ is 1.95 times and 2.54 times of that at 25℃,35℃,respectively.(2)The first-order RC equivalent circuit model was selected as the basic model by comparing the characteristics of different equivalent circuit models,and the online identification model by recursive least squares(RLS)was derived based on the first-order RC equivalent circuit model.An online parameter identification model based on multi-innovation least squares(MILS)algorithm was proposed to solve the problem of low identification accuracy caused by only using the new information of the current system in each parameter change of RLS.MILS can make full use of the new information of the system at every moment,thus increasing the data utilization efficiency,accelerating the convergence of the algorithm and improving the accuracy of parameter identification.Through DST working condition test,the error between the estimated terminal voltage of RLS and MILS algorithm and the experimental value is compared.The research results show that:The end voltage estimation errors of the two algorithms are 2.017%and 1.568%,respectively.The corresponding calculated output terminal voltage of the two algorithms has good tracking effect on the test terminal voltage,indicating that the equivalent circuit model parameter identification has good accuracy.MILS algorithm has higher model parameter identification accuracy and smaller end voltage estimation error.The online identification algorithm based on MILS can further improve the accuracy of parameter identification.(3)Based on the selected equivalent circuit model and its internal electrical relationship,the state space equation of estimating battery SOC using EKF algorithm is derived.The SOC-OCV relation needs to be substituted into the EKF algorithm,and the SOC-OCV relation can be obtained through off-line SOC-OCV data fitting.First,the SOC-OCV curve fitted with order 3,6 and 9 of terpolymer lithium battery is compared to compare its fitting accuracy,and then the SOC-OCV curve fitted with order 5,6 and 7 of terpolymer lithium battery is further compared.Compared with its fitting accuracy,the 6-order polynomial fitting result of battery SOC-OCV offline data was selected as the SOC-OCV curve relation and substituted into EKF algorithm.(4)RLS and MILS online identification algorithms are combined with EKF respectively to establish a comparison between RLS-EKF and MILS-EKF SOC estimation joint algorithms,and compare the accuracy of the two joint estimation algorithms.The research results show that:The milS-EKF combined estimation algorithm can further improve the accuracy of battery SOC estimation,and the estimation accuracy is 19%higher than rLS-EKF.In addition,in the face of a large initial SOC error deliberately set,the algorithm can stably converge to the theoretical value in a short time,and the algorithm has good robustness.