Optimization Of Double Kalman Filter Based On Genetic Algorithm For SOC Estimation Of Lithium Battery

As one of the keys to solve the energy and environment problems,electric vehicle is attracting wide attention.As a key component of electric vehicles,the state of charge(SOC)of the battery needs to be accurately predicted in order to give full play to its performance.This paper focuses on the SOC estimation of ternary lithium-ion batteries.Through the designed battery test experiment,the accuracy of the battery model and the accuracy of the proposed algorithm for SOC estimation are analyzed and verified.The main work contents of this paper include:(1)Considering the multi-time scale effect of the battery and the complexity of the model,the second-order RC equivalent circuit model is adopted as the research model in this paper.According to the battery performance parameters and research content,design the battery test experiment,and through the battery discharge characteristics test experiment.The basic structure and working principle of the battery are introduced and analyzed,and the relationship between SOC and opencircuit voltage,internal resistance characteristics,capacity,charge-discharge rate and capacity is emphatically analyzed based on the test data of the battery.(2)In view of the time-varying characteristics of battery model parameters in the actual use environment of electric vehicles,a genetic algorithm optimized Kalman filter(GA-KF)was proposed for dynamic identification of model parameters.The model parameters at different discharge rates and SOC values were identified dynamically.In DST condition,the model was compared with the model based on the least square method to verify the accuracy of the proposed model.(3)Aiming at the problem of nonlinear state variables caused by electrochemical reactions in the battery of electric vehicles,a strong tracking Kalman filter(STF)was proposed to estimate SOC accurately.A GA-DKF based SOC estimation method is proposed in combination with the above model dynamic parameter identification method.KF was used to identify model parameters,STF was used to estimate SOC,and genetic algorithm was used to optimize the covariance matrix of system noise and observation noise of two Kalman filters.In FUDS condition,the SOC estimation is carried out and compared with EKF and UKF to verify the accuracy and robustness of the proposed algorithm.(4)In order to verify the feasibility of the proposed algorithm for the power battery pack,an EV model was built in AVL Cruise software,and the vehicle simulation was carried out under the designed two dynamic cycle conditions,and the battery pack current and voltage parameters were obtained.The proposed GA-DKF algorithm was used to estimate the SOC of the battery pack,and the SOC values obtained after vehicle simulation with Cruise software were taken as reference to verify the effectiveness of the proposed algorithm.