Design, implementation and validation of live state of charge estimator based on the extended Kalman filter

Abstract

The increasing awareness of environmental issues related to fossil fuels, combined with their increasing price and decreasing availability has lead to the rise of Electric Vehicle’s (EV) and Hybrid Electric Vehicles (HEV). One of the most important tasks of the Battery Management System (BMS) in automotive applications is determining the state of charge (SoC) of the battery, which represents the amount of remaining capacity in the battery. An overview of the different SoC estimation methods is given in this thesis from which two methods are chosen to design SoC estimator models : the Coulomb Counting method and the Extended Kalman Filter. Hereafter, both models were implemented on an Arduino Mega 2560 to perform live SoC estimations on a Li-NMC battery. The knowledge of some battery parameters such as capacity, the OCV-SoC relationship and equivalent model parameters is paramount to model battery behaviour and calculate SoC accurately. As such, these parameters were determined for the tested Li-NMC battery with a capacity test, OCV test and HPPC test respectively. Hereafter, both SoC estimator models were extensively tested using three tests with increasingly dynamic profiles : Constant Current Constant Voltage test (CCCV), Dynamic Discharge Pulse test (DDP) and Worldwide harmonized Light vehicles Test Cycle (WLTC). The measurements of these tests were used as inputs in an established second order battery model created by the Battery Innovation Centre of the VUB. The similarity between the simulated voltage of the battery model and the measured voltage during the test was used as an indicator for the accuracy of the SoC estimation.
While the Coulomb Counting SoC estimator showed an acceptable error during charging and discharging with a constant current, its performance proved inadequate during the more dynamic DDP and WLTC tests. However, the presented Extended Kalman Filter SoC estimator model offers an accurate and robust live SoC estimation that can handle real-life dynamic load profiles and as such, could be of interest for real-life applications.

Date of Award	24 Jun 2016
Original language	English
Awarding Institution	Vrije Universiteit Brussel
Supervisor	Alexandros Nikolian (Promotor)