Research On SOC Estimation Technology Of Lithium-ion Battery

Lithium-ion batteries are widely used in the automotive field because of their advantages in energy density,power density and number of cycles.As a parameter in the battery management system(BMS),the State of Charge(SOC)of the battery is of great significance to the BMS and the whole EV.On the one hand,the BMS can control the charge and discharge of the battery and other functions according to the SOC value;on the other hand,the driving range of the EV can be judged by the SOC value.Ensure safe travel.However,the SOC value of battery is a parameter that can not be measured directly,and can only be obtained by estimation.How to improve the accuracy of SOC estimation has always been the focus of researchers.This paper takes lithium-ion batteries as the research object and aims to accurately estimate the SOC of batteries.Firstly,the working principle and advantages and disadvantages of lithium-ion batteries were analyzed,and the three-element nickel-cobalt-manganese lithium-ion battery was selected as the object of this study.After that,a battery test platform was built to conduct charging and discharging experiments and related working conditions,and the relationship curve between SOC and open-circuit voltage of the battery was calibrated.Based on the experimental data,the characteristics of resistance,voltage and capacity of lithium-ion battery were analyzed,and the definition of SOC and its influencing factors were introduced.Secondly,based on the second-order RC equivalent circuit model,the model function relationship is derived,and the parameter values in the model are identified by off-line and online parameter identification methods.Offline parameter identification is based on constant-exile-current experiments and Forgetting Factor Recursive Least Square(FFRLS)is used to identify parameters at different SOC values by fitting method,while online parameter identification is realized by FFRLS.In addition,aiming at the problems of large fluctuation and partial divergence of initial identification results of FFRLS algorithm,data monitoring window is set up to ensure the stability of online parameter identification results.The simulation results show that the Mean Absolute Error(MAE)between the simulated voltage and the actual voltage at the off-line parameter identification result end is only 0.003 V under constant current exile condition.In the United States Federal Urban Driving Schedule(FUDS),MAE of the online parameter identification results after optimizing the data monitoring window is only 0.011 V between the simulated voltage and the actual terminal voltage.And by comparing the identification result graph and simulation error graph before and after the optimization of FFRLS algorithm,it can be seen that the parameter identification effect of the optimized FFRLS algorithm is better.Finally,the Extended Kalman Filter(EKF)algorithm combined with the equivalent circuit model was used to derive the process for estimating the SOC of Li-ion batteries.The influence of fixed Gaussian noise in EKF algorithm on the estimation accuracy is analyzed.In order to optimize the Gaussian noise in EKF algorithm,particle swarm optimization(PSO)algorithm is introduced to optimize the Gaussian noise of the system.Furthermore,PSO algorithm is improved according to quantum mechanics idea and particle swarm optimization(QPSO)algorithm with quantum behavior is proposed to further optimize the noise in EKF.At the same time,the corresponding model is built in MATLAB software to verify the effectiveness of different algorithms to estimate SOC under FUDS condition and dynamic stress test(DST)condition respectively.Under the two working conditions,both the EKF algorithm,PSO-EKF algorithm and QPSO-EKF algorithm can achieve accurate estimation of SOC for lithium-ion batteries,and the MAE estimated is within 1%.Moreover,the estimation accuracy of the EKF algorithm optimized by QPSO is better than that of the PSO-EKF algorithm and the EKF algorithm.It is proved that with the continuous optimization of the system noise,the estimation accuracy is also improved.