State of Charge Estimation of Lithium Ion Battery using Adaptive Extended Kalman Filter and Artificial Neural Network /

Abstract

With Electric vehicles (EVs) are gaining traction amid global warming, air pollution and other environmental concerns arising from the use of Internal Combustion (IC) engine vehicles. As batteries are main power sources for EVs, so to ensure optimal operation of EVs, accurate estimation of batteries’ state of charge (SOC) is crucial. In this research work, a hybrid method for SOC estimation is proposed that employs both adaptive extended Kalman filter (AEKF) and artificial neural network (ANN) which is then further used to reserve a charging slot in a charging station in advance. A real time embedded hardware is also implemented for SOC estimation keeping in mind its on-board implementation and practicality. In addition to this, to save the time of EV users when their EV is being charged, a real time IoT based monitoring of charging status is also implemented via ThingSpeak. The proposed technique is validated under five different EV drive cycles and four temperature values. The RMSE calculated for the proposed technique under five drive cycles proves the effectiveness of the proposed method over existing methods like CC, EKF, AEKF and ANN.