Characterization Research Of LiFeO₄Batteries For Application On Pure Electric Vehicles

LiFePO₄battery has been widely applied to electric vehicles due to its high safetyand long cycle life. The research of the battery performance provides the main referencefor the construction of battery management system (BMS), and has great significance toimprove the security and energy economy of the electric vehicles.The research has been conducted on the basic characterization of the battery, thecycle durability under multi-stress and the approaches for the state-of-charge (SOC)estimation.The basic parameters such as battery capacity, internal resistance andopen-circuit-voltage (OCV) demonstrate a high nonlinearity with the changes inenvironment and working conditions. First, the relationship between capacity andcurrent ratio is studied by steady-state discharge experiments. Second, the charge/discharge ohmic internal resistance and polarization resistance under different SOChave been analyzed through a step-input method. Then, the OCV curve is studied andthe discussion on reasonable experimental methods for the stable OCV curve has beencarried out. Finally, the impact of the different ambient temperature on battery capacity,internal resistance and OCV is considered.In the research of battery durability, a battery life prediction model has beenestablished. The model takes several influence factors into account, such as ambienttemperature, discharge rate, discharge cut-off voltage, charge rate and charge cut-offvoltage. The existence of coupling between these factors has been verified, and therelationship between stress levels and the coupling intensity is obtained by experiments.Then, the influence of coupling on capacity fade is analyzed under different stress levels,and the life prediction model is developed based on the coupling intensity determinationand multi-factor input. Additionally, a model-based sensitivity analysis is conducted toreveal the weight of these factors on battery life. After the battery has entered the stabledecay phase, the life prediction model, which has taken into account the couplingbetween factors, has reached an error of15%or less.In the study of battery SOC estimation, the model-based algorithm and theimproved Ampere-hour integration method have been applied. The estimation result of the direct model-based algorithm shows excessive fluctuation and after combined withKalman filtering recursive algorithm, the estimation quality has been greatly enhanced.To calibrate the initial SOC, the traditional single OCV curve method has beensubstituted by a multi-factor SOC-OCV curve-cluster. This new calibration map cansignificantly improve the calibration accuracy, expand the SOC range that the methodcan cover, increase the correction frequency of SOC initial values and reduce the erroraccumulation. This curve-cluster method has achieved an accuracy of less than5%inthe range of SOC <40%or SOC>70%, and15%in the range of40%-70%SOC.In addition, some of the related experiment methods and main conclusions in thispaper are applicable to hybrid electric vehicles and other types of lithium-ion batteries.