A Lithium Battery Management System Using SOC For Active Balancing

With the proliferation of new energy vehicles,the research of power battery has received extensive attention.Due to the limited capacity of the power battery,maximizing battery energy utilization,extending battery life and ensuring battery safety are the core issues of battery management system(BMS).The state of charge(SOC)estimation has always been the focus of BMS.SOC is an important indicator to characterize the present charge capacity of the battery.BMS can use it to calculate cruising range of electric vehicles,optimize energy recovery strategy,balance battery charges,perform fault diagnosis,etc.There are many issues concerning SOC estimation,firstly,large computational complexity and complicated implementation are inevitable when high-order equivalent circuit models are utilized;secondly,long convergence time is needed when the initial model parameter values are not selected properly;thirdly,SOC estimation will not reflect the true value if the influence of ambient temperature is not considered;fourthly,large SOC estimation error will result if conventional coulomb counting method is used;finally,the tracking of the change of battery model parameters during operation is not possible if no corrective measure is implemented.This thesis presents a temperature-incorporated RLS SOC Kalman estimation technique to solve the above mentioned issues.Firstly,a simplified first-order resistor-capacitor(RC)equivalent circuit model is used to reduce the model complicity but yet is able to obtain accurate results.Secondly,a proprietary lookup table of model parameters is employed to determine a proper initial value of the model parameters.Thirdly,a multi-temperature battery model is established to take care of the ambient temperature influence.Fourthly,Unscented Kalman Filter(UKF)is implemented to effectively minimize the large SOC estimation error.Finally,Recursive Least Square(RLS)algorithm is adopted to track the change of the battery model parameters during operation in real time.The proposed SOC estimation technique is simulated in MATLAB/Simulink.The simulation results show that less than 0.87% of SOC estimation error is attained.