Title: Unscented Kalman filter-based control strategy for wind turbine systems
Authors: Abdelkader Garmat; Kamel Guesmi
Addresses: Department of Electrical Engineering, University of Djelfa, Djelfa, Algeria ' CReSTIC, University of Reims, Reims, France
Abstract: This paper proposes a new control strategy based on an improved linear quadratic multimodel optimal controller for a wind turbine system. The proposed structure is formed of a multimodel base with eight models and an unscented Kalman filter estimator. This study is performed to allow the wind speed accurate estimation, improve the controller performance, maximise the delivered power and reduce its fluctuation. Furthermore, the proposed approach provides a power reserve to ensure the wind turbines' smooth participation in the network frequency regulation. This paper demonstrates, through a comparative study, that the unscented Kalman filter is an excellent alternative and gives better results than the Kalman filter and the extended Kalman filter in the case of highly nonlinear systems. The simulation results validated the proposed approach and showed its efficiency. They confirmed that the proposed controller, based on the unscented Kalman filter, can ensure the best results and the lowest fluctuation of the power than the controller based on the Kalman filter and the one without wind speed estimation. This allows the proposed approach to maximise the generated electric power and helps in the network frequency regulation.
Keywords: wind turbine; variable-speed; linear quadratic multimodel optimal control; wind speed estimator; Kalman filter; extended Kalman filter; unscented Kalman filter; Newton-Raphson.
DOI: 10.1504/IJAAC.2023.134552
International Journal of Automation and Control, 2023 Vol.17 No.6, pp.613 - 634
Received: 08 Sep 2022
Accepted: 28 Jan 2023
Published online: 27 Oct 2023 *