Research On Adaptive Co-Prediction Of The State Of Energy And Peak Power For Power Lithium-ion Batteries

With the concepts of “Carbon Neutral” and “Carbon Peak” are promoted,the new energy industry is developing rapidly,and power lithium-ion battery shipments are rising fast.Power lithium-ion batteries are widely applied in the fields of vehicles,drones,and so on.State of Energy(SOE)and peak power are key state parameters of the battery management system to realize the safety and efficient application of batteries.In this thesis,the research on the adaptive collaborative prediction of the SOE and peak power of the power lithium-ion battery is conducted.(1)This thesis designed some power lithium-ion battery working characteristic test experiments to explore the changes of the internal resistance,energy,and open-circuit voltage under different working conditions.The working characteristic supplies the basis for the improved Thevenin equivalent circuit model which is constructed in this thesis,and the forgetting factor recursive extended least square algorithm is adopted to realize accurate online parameter identification of the model.The experiment verification maximum voltage error is38.40 m V of the proposed battery model.(2)Aiming at the extended Kalman filtering algorithm noise condition cannot be satisfied in actual application,which will cause the estimation error of SOE.This thesis constructs an innovation sequence through a sliding window to estimate the observation noise and system noise of the power lithium-ion battery system.The sliding window multi-innovation-adaptive extended Kalman filtering(SWMI-AEKF)algorithm is designed,which can realize the adaptive estimation of the observation noise covariance matrix and the system noise covariance matrix for the accurate SOE estimation by the innovation sequence.(3)According to the system state estimated by the SWMI-AEKF algorithm,the peak power estimation based on the equivalent circuit model is carried out.At the same time,a peak power estimation model is constructed considering the influence of SOE on the peak power estimation to realize the SOE and peak power adaptive co-estimation of the batteries.(4)To verify the estimation accuracy of the SOE and peak power adaptive co-estimation method proposed of the power lithium-ion batteries.The co-estimation method of SOE and peak power are verified by the Beijing buses dynamic stress test conditions and dynamic stress test conditions.Based on the SWMI-AEKF algorithm under different working conditions,the SOE estimation maximum root mean square error and maximum average absolute errors are1.06% and 1.03% respectively.The maximum root mean square error and maximum mean absolute error of the peak power estimation method proposed in this thesis are 26.97 W and21.11 W respectively.This research constructs an improved equivalent circuit model according to the battery working characteristics and obtains accurate model parameters by online parameter identification.Besides,this thesis adopts the SWMI-AEKF algorithm for accurate SOE estimation and combines the equivalent circuit model to achieve the peak power adaptive coestimation of the power lithium-ion battery.The experimental verification results show that the proposed method can effectively realize the SOE and peak power co-estimation,which can provide the theoretical basis for the management and application of power lithium-ion batteries.