Prashant Shrivastava, Tey Kok Soon, Mohd Yamani Idna Bin Idris, Saad Mekhilef, Syed Bahari Ramadzan Syed Adnan

    To build up a proficient battery management system, it is required to accurately estimate the state of charge (SOC) of the electric vehicle (EV) battery. Generally, the accuracy of the conventional extended Kalman Filter (CEKF) algorithm is exceptionally affected by the method used to update the noise covariance matrices under running conditions. In this work, the new adaptive extended Kalman filter (AEKF) algorithm is designed for the SOC estimation. Methods such as forgetting factor method and moving window are used for estimation of measurement noise and sensor noise covariance matrix respectively. Pulse discharge and customized dynamic stress tests are conducted to check the robustness of the proposed algorithm. Experimental results indicated that proposed AEKF has superior performance than CEKF under dynamic load conditions.

Key words:Lithium-ion Battery,State of Charge,Kalman Filter,Battery Management System,Electric Vehicle

DOI:10.1109/IPEMC-ECCEAsia48364.2020.9367743

Date:2020-11-29