Research On State Of Charge(SOC) Estimation Of Lithium-ion Battery Based On Improved Central Difference Kalman Filter Algorithms

Under the environment of energy crisis and global warming,in order to promote energy conservation and emission reduction,the electric vehicle industry has developed rapidly.Battery Management System(BMS)is usually used to monitor and manage lithium-ion batteries,which are the main power source of electric vehicles.As one of the key indicators in BMS,State of Charge(SOC)represents the ratio of current capacity to full capacity.Accurate estimation of SOC can prolong the cycle life of battery,reduce the attenuation rate of battery capacity and allow battery to operate in a safe state.This paper takes Ternary lithium-ion battery as the research object,carries out the modeling and parameter identification of the battery,and focuses on the SOC estimation methods.The specific work is as follows:(1)The second-order RC equivalent circuit model is used to simulate the operating characteristics of the battery,and the corresponding state space equation is established and discretized.The pulse discharge method and the recursive least square method with forgetting factor are used to meet the needs of both offline and online parameter identification.The accuracy of the parameters identified offline is verified by the intermittent constant current pulse experiment.(2)Considering that the inherent innovation scalar in the Central Differential Kalman Filter(CDKF)algorithm can not reflect the recent change trend of state,the multi-innovation Central Differential Kalman Filter(MI-CDKF)algorithm is designed by combining CDKF with multi-innovation theory,which expands the innovation scalar to innovation vector,obtains better estimation performance with less calculation,and the robustness of MI-CDKF is verified when Gaussian white noise and bias value are added to voltage and current.(3)In order to suppress the negative influence of filter divergence caused by ill-conditioned matrix in CDKF algorithm and adjust the covariance matrix of process noise and measurement noise in real time,an adaptive square root Central Difference Kalman Filter(ASRCDKF)algorithm is designed,which updates the state covariance matrix in the form of square root.(4)CDKF assumes the measurement noise as Gaussian white noise,which is not suitable for the actual situation.In order to overcome the limitations brought by this assumption,the central difference sampling idea of CDKF is combined with H∞ Filter to form Central Difference H∞Filter(CDHF).Sage-Husa adaptive filtering is introduced to further improve CDHF into adaptive CDHF(ACDHF).Finally,CDKF,CDHF and ACDHF are tested under different noise interference conditions.The simulation results verify that ACDHF has the best robustness and higher estimation accuracy for different noise interference conditions among the three.