Study On The Estimation Of Charge State Of Series Lithium-ion Battery Packs Based On Multi-time Scale DEKF

With the rapid growth of the global economy,the energy crisis and air pollution problems have become increasingly acute.Studies have shown that vehicle exhaust emissions are the main source of air pollution.New energy electric vehicles instead of traditional fuel vehicles will be a major effective way to solve this problem,lithium batteries as the main energy source of new energy vehicles,it is necessary to study its state of charge,SOC accurate monitoring can directly reflect the range of electric vehicles,in addition to high-precision SOC prediction can also improve energy utilization and ensure battery safety.In this paper,taking the monolithic lithium battery and the battery pack composed of four monomer batteries in series as the research object,the experimental test platform is first built and a series of test schemes are designed for testing,and the basic characteristics of lithium batteries and the inconsistent characteristics exist after the series are grouped.Then the existing single cell model is analyzed,the DP model with the balance of computation and accuracy is selected to model the single cell,the adaptive genetic algorithm is used to identify the parameters of the DP model,the state space equation is established for the DP model and the SOC estimation of the single cell is carried out by using the traditional extended Kalman filtering algorithm,but the parameters of the battery model for complex dynamic conditions are not set in stone.This leads to the inaccuracy and poor applicability of traditional SOC estimates based on offline parameter identification.Therefore,in view of the above problems and the problem that the computational burden existing in the subsequent calculation of the battery pack SOC is too large,the multi-time scale double extended Kalman filtering algorithm that updates the model parameters online with time is proposed,and a macroscopic time scale extended Kalman filter is used to identify the battery parameters online,and a microscopic time scale extended Kalman filter is used to estimate the SOC.The multicondition applicability of the algorithm is verified under the FUDS,HWFET and UDDS working conditions,and the optimal macro time scale for online parameter identification is verified and selected under the BJDST working condition,and the convergence of the algorithm is verified.Finally,based on the previous SOC estimation analysis of a single battery and the inconsistencies between monomer batteries after being grouped in series,the SOC definition of the tandem battery pack is introduced,and the SOC estimation problem based on the definition is too large,and the improved Rint model is proposed to select the SOC estimation problem representing the battery so as to convert the SOC estimation problem on behalf of the battery group into the SOC estimation problem on behalf of the battery.The multi-scale DEKF algorithm introduced in Chapter 3 is used to estimate the SOC of the representative battery to obtain the battery pack SOC.The effectiveness and accuracy of the battery pack SOC estimation method under NEDC condition are verified by comparing the battery pack SOC in both SOC consistent and SOC inconsistent conditions with the traditional definition-based method,SOC maximum and minimum method,signal cell equivalent method and battery pack SOC estimation method based on terminal voltage selection.Figure [44] Table [15] Reference [65]...