A Kalman Filter Based Battery State of Charge Estimation MATLAB Function

Exponential Moving Average Extended Kalman Filter for Robust Battery State-of-Charge Estimation

“…The considered model is given in Fig. 1 and this can also be found in [6], [18]. In this kind of modeling approach, open circuit voltage of the battery is denoted by V oc while the terminal voltage that is available to measure is denoted by V t , internal resistance is denoted by R 0 .…”

“…Tested temperatures are: −10 • , 0 • , 10 • ,25 • , and 40 • . Then, by using parameter optimization approach in Matlab [6], battery parameters are obtained at different SOC levels and temperatures. Then, using interpolation, temperature and SOC dependent RC circuit parameters are obtained and used for calculation of the system and input matrices.…”

“…As highlighted in the previous Section, measurement noise and unavailability of the initial SOC make it difficult to apply Coulomb counting method in practice. Extended Kalman filter (EKF) [6]- [9] is a popular choice in the literature to overcome these issues. EKF can easily be applied to the nonlinear system ( 14) and ( 15) and based on the estimated states, battery SOC and filtered version of the terminal voltage can be obtained.…”

“…Although this method is relatively easy to implement, however, it is not very suitable for dynamic applications and/or cases where the SOC-OCV curve is very nonlinear. To overcome these issues, researchers have combined various techniques such as coulomb counting, SOC-OCV curve together with model-based approaches such as extended Kalman filter (EKF) [6]- [13]. In the case of EKF, mathematical model of the battery equivalent circuit is required.…”

“…Although extended Kalman filter can provide robust estimation, however, certain input parameters can significantly affect the performance of EKF. To overcome this issue, numerous modifications of the EKF structure are proposed in the literature [6]. However, these solutions often are very complicated, requires lot of real-time computationally expensive computation that can be prohibitive for low-cost energy management systems.…”