State estimation algorithms for electric vehicle 48 V battery management system

Abstract

The estimation of the State of Charge (SOC) is the most important function of the battery management system (BMS) of a vehicle. The algorithms for this estimation are conventionally based on adaptive methods but now improved with adaptive and conventional methods. The paper presents SOC estimation based on second order equivalent circuit and open circuit voltage-based capacity estimation models with extended Kalman filter approach for accurate estimation at pack level simulated in MATLAB. The BMS measures the voltage, current and temperature of the battery pack. The estimation of the equivalent circuit parameters is done using a nonlinear least squares solver from the measured quantities of the pack. The equivalent circuit parameter estimation is done using real time hybrid pulse power characterization (HPPC) data and the drive cycle data. The test data for the look up table is for charging and discharging at C/50 of a Lithium Iron Phosphate (LFP) chemistry at pack level. The SOC estimation using modified Leunberger observer with dynamic observer gain and extended Kalman filter algorithm has been tested for 0.1C and 0.5C rates. With input data available at 25 °C the accuracy for SOC estimation is 95%. With data from different temperatures the accuracy improved to 98%.