Discretization Order Influences on Extended Kalman Filter Estimation for Doubly-Fed Induction Generator

Abstract

The main objective of this paper is to analyze the influence of the discretization step on the estimated states of the Doubly-Fed Induction Generator (DFIG). Although the Extended Kalman Filter (EKF) has been widely used for such systems, the discretization process is conventionally ensured by the first-order Forward Euler method. Therefore, the effects of the discretization order of the discrete state-space representation on the Extended Kalman Filter estimation have not been studied before. In this paper, we combine the Extended Kalman Filter with two second-order discretization methods: Central Difference and Adams-Bashforth methods, to estimate the states of a Doubly-Fed Induction Generator and improve the estimation precision of the rotor speed and the Flux of the generator. A comparative study has been conducted to analyze the qualitative and quantitative responses of the estimator for different cases. The obtained results have demonstrated the significance of the discretization order on the estimation process of the two states of the DFIG.