Design of extended Kalman filter for SEPIC converter and comparison to Kalman filter

Abstract

Single ended primary inductance converter (SEPIC) is a type of DC–DC converter whose industrial applications include maximum power point tracking and Active power factor correction. Owing to present nonlinear circuit components, SEPIC has discontinuous nonlinearity and for the mathematical model determination, state space averaging or more advanced modeling techniques are required. Once the approximated nonlinear model is obtained, the states of the plant can be estimated using a nonlinear state estimator. Nonlinear estimators are preferred mostly when the real trajectory of the states cannot be estimated accurately by a linear state estimator and the deviation from the linear behavior and nonlinear one is above what is tolerable for the applications. Extended Kalman filter (EKF), being such a state estimator is widely employed in control and fault monitoring applications where a complete or an approximated nonlinear model of the plant is available. EKF especially operates well for the plants where the nonlinearity does not cause the assumptions behind EKF to be violated. In this study, first an approximated nonlinear model for SEPIC is determined. Using the model, EKF is designed to estimate the states of the plant. The efficacy of the designed EKF is demonstrated in a series of numerical MATLAB based simulations where the both open loop and closed loop performances are analyzed, and the results are detailed in the paper.