Research On The Online Estimation Of State-of-charge Of Lithium-ion Battery

With the development of battery technology, lithium-ion battery as the representative of high quality battery has been widely used in all walks of life. In order to ensure its working performance and security, Battery Management System(BMS) has always been a hotspot. State of Charge(SOC)estimation, the most fundamental and pivotal function of BMS, has been recognized as a challenging job for the coupling influence by many factors. This thesis mainly focused on the research of the online SOC estimation of lithium-ion battery which considered factors such as current and ambient temperature. The research content contained following aspects:(1) Battery equivalent circuit model and off-line identification of its parameters were accomplished. On the basis of a large number of experiments, the hysteresis effect of open circuit voltage(OCV) and the rule of practical capacity were studied. With the help of 1stOpt, the nonlinear function relations about the OCV-SOC curve and the coulomb efficiency coefficient were fitted.(2) Considering the drawbacks of Unscented Kalman Filter(UKF) which needs an accurate model and a priori noise statistics, an improved Unscented Kalman Filter based on Recursive Least Squares(RLS) and the optimal unbiased Maximum a Posterior(MAP) was proposed. Firstly, the state space equation was deduced from Thevenin equivalent circuit model. Secondly, parameter adaptive adjustment was realized by RLS. Then, the covariance matrix of noise was adjusted adaptively by the optimal unbiased MAP method. Finally, physical experiments which contained simple and complex working conditions were designed. The experimental results revealed that this approach's estimate accuracy was excellent with acceptable robustness and computational complexity.This method was suitable for these applications which have strict requirements on computational complexity.(3) Considering the same drawbacks, another algorithm was proposed to further improve the estimation accuracy. The main point of this algorithm was that the parameter adaptive algorithm and noise adaptive algorithm were augmented Kalman filter and Covariance Matching(CM)algorithm, respectively. In order to provide the appropriate parameter for the model adaptive algorithm, sensitivity analysis experiment was designed. The experimental results revealed that the robustness and real-time performance were slightly decreased. However, this method had a better estimation accuracy and stability. This method was suitable for these applications which have strict requirements on estimation accuracy.(4) PC software about battery test and online SOC estimation system based on LabVIEW was programmed. With the help of this software, the test experiment efficiency and data precision was improved, as well as the online monitoring ability of battery.