Estimation of SoC for Real Time EV Drive Cycle using Kalman Filter and Coulomb Counting
Deepesh Kanchan, Nihal Nihal, Avinash Paul Fernandes
Abstract
Research and development to determine State of Charge(SoC) has taken a new turn ever since the electric vehicle technology has entered the market. The current SoC determination methods employed gives us results that are far from perfect, so a lot of new technologies and methods have emerged which allows us to determine the SoC of a battery with lesser errors compared to the previous ones, yet there is a lot of space for improvement. The Battery Management Systems in electric vehicle domain mostly employ the Coulomb counting method of estimation to determine the state of charge and the subsequent range of the vehicle. Kalman filter is a method which can give better estimates of a state and in energy storage systems that of SoC. In recent Battery Management Systems the Kalman Filter is proposed to estimate SoC with better accuracy. This paper compares the use of Coulomb counting and Kalman filter techniques to estimate State of Charge of EV battery for the Mangaluru drive cycle. The Mangaluru drive cycle is transient and gives better estimate of SoC in comparison with Standard drive cycles for Mangaluru region. The Kalman filter technique was found to be more accurate compared to coulomb counting method due to the inclusion of uncertainties in measurement of current and voltage in the simulation model. Set with the same initial conditions, coulomb counting estimated the end state to be around 81% while Kalman filter around 71%. The Kalman filter gave an RMS error of 2.5067% which was obtained by comparing the model output.