Research On LiFePO4 Battery Modelling And State Of Charge Estimation

LiFeP04 batteries have been widely used in mobile phones, electromobiles because of the excellent charging and discharging performance and being free from battery memory effects. This thesis focuses on the research of LiFePO4, providing accurate and reliable estimation of the State of Charge (SOC) for battery systems implemented on deep water autonomous underwater vehicles (AUVs). The SOC estimation is one of the important indications of the battery management systems on such AUVs. LiFePO4 batteries have prominent features such as weak self-discharging, high power density and stable chemical structures, etc. While charging and discharging cycles, the voltage profile of LiFePO4 batteries maintains stably around 3.3V during process of extraction and insertion of Li+, nonlinear characteristics of voltage occur when battery is fully charged or discharged.SOC estimation method is normally considered to be Current Counting. However, due to the initial error and process error of SOC, the estimation results of Current Counting are quite poor in accuracy. In contrast, Kalman Filter provides an effective and powerful way for SOC estimation, especially for reducing the impact of initial error and process error of SOC on estimation results. For the reason that estimation accuracy of Kalman algorithm strongly relies on the precision of battery model, a Labview-based battery performance test platform is established for acquiring experimental data of battery while charging and discharging. From the experimental results, one is able to extract the nonlinear functional relationship between SOC and equivalent power potential (Voltage of Open Circuit, Uoc) by using Curvefit Toolbox. Furthermore, MATLAB/SIMSCAPE is employed to build LiFePO4 on the basis of 2nd-order RC circuit. Parameter Estimation Toolbox in SIMSCAPE is used for RC parameters estimation in order to render the simulated model matching experimental data.According to the state space equations based on the RC equivalent circuit of LiFePO4, Extended Kalman Filter (EKF) is used to estimate SOC for battery package, the estimation results demonstrate that estimated SOC enables to approximate the reference SOC. Considering the scenario that deep water AUVs require battery instantaneously to output high power, causing SOC value to immediately drop down, Strong Tracking Extended Kalman Filter (STEKF) is used for SOC esitimation to better tracking reference SOC values. In order to enhance the stability of STEKF, eigenvalue decomposition is applied to modify the priori state error covariance matrix for realizing the filter of better performance.Given that different signals in continuous physical system change by different rate, the multirate control strategy is introduced to improve the performance of EKF and STEKF. With theoretical deduction, NEDC (New European Driving Cycle) battery discharging profile is exploited for evaluating the effectiveness and efficiency of SOC estimation results given by Input Multirate Extended Kalman Filter (IMEKF), Input Multirate Strong Tracking Extended Kalman filter (IMSTEKF), Output Multirate Extended Kalman Filter (OMEKF) and Onput Multirate Strong Tracking Extended Kalman filter (OMSTEKF) under the scenarios that multiplicities are 2 and 4 respectively.