Research Article Real time simulation of nonlinear generalized predictive control for wind energy conversion system with nonlinear observer Kamel Ouari a , Toufik Rekioua a,n , Mohand Ouhrouche b a Laboratory LTII, University of Bejaia, Algeria b Laboratory LICOME, University of Quebec at Chicoutimi, Canada article info Article history: Received 21 February 2011 Received in revised form 16 December 2012 Accepted 8 August 2013 Keywords: Nonlinear generalized predictive control (NGPC) Doubly-fed induction generator (DFIG) DFIG-based wind turbine Real-time simulation abstract In order to make a wind power generation truly cost-effective and reliable, an advanced control techniques must be used. In this paper, we develop a new control strategy, using nonlinear generalized predictive control (NGPC) approach, for DFIG-based wind turbine. The proposed control law is based on two points: NGPC- based torque-current control loop generating the rotor reference voltage and NGPC-based speed control loop that provides the torque reference. In order to enhance the robustness of the controller, a disturbance observer is designed to estimate the aerodynamic torque which is considered as an unknown perturbation. Finally, a real-time simulation is carried out to illustrate the performance of the proposed controller. & 2013 ISA. Published by Elsevier Ltd. All rights reserved. 1. Introduction Recently, many new wind farms have employed wind turbines based on doubly-fed induction generator (DFIG) [1], due to their full power control capability, variable speed operation, low con- verter cost and reduced power loss [2,3]. However, DFIG constitu- tes a challenging control problem, because of its fast dynamics, and being a highly coupled and nonlinear multi-variable system. The field-oriented vector control using cascaded PI controllers is widely used, in DFIG-based wind turbines, for reasons of simplicity and applicability [4]. However, PI-type control methods are not effective when the system to be controlled is characterized by strong nonlinearity and external disturbances. To overcome these drawbacks, various approaches have been proposed to replace PI-type controllers. Some examples are the H1 control theory, neural networks and sliding mode control [5-7]. On the other hand, model predictive control (MPC) is now regarded as one of the most robust control strategies because of its advantages, such as insensitivity to parameter variations, external disturbance rejection and fast dynamic responses [8]. Therefore, it has been applied in the industry as a promising form of advanced control [9]. Several control strategies using linear and nonlinear MPC have been proposed in the technical literature [10,15]. In [10] Chen et al. have designed a nonlinear predictive control law for multi- variable nonlinear systems based on Taylor series expansion, where the same relative degree of multi-input and multi-output system is considered. In [11,12], the principle of the MPC is combined with the well-known direct torque control (DTC) and applied to an induction motor (IM) for a fast torque response. Hedjar et al. [13] have proposed a cascaded NGPC based on Taylor series expansion for an IM. However, in these studies, the load torque is considered as a known disturbance. MPC techniques have also been proposed for DFIG-based wind turbines. In [14], the GPC has been used to control the pitch angle of windmill blades in order to reduce power fluctuations. Multi- variable control strategy based on MPC techniques has been proposed for wind turbines based on DFIG in [15]. A predictive current control (PCC) strategy for doubly fed induction generators has been treated in [16]. Nevertheless, all these methods achieve a high performance only when the external disturbance (aerody- namic torque) is known. In order to provide powerful abilities in handling system disturbances and improving robustness, the nonlinear generalized predictive control must be combined with a disturbance observer [17,18]. For improvement, in this paper, the aerodynamic torque observer is integrated into the control law in order to enhance the robustness of the controller. The combination of this observer with NGPC works as a nonlinear controller. In this paper, the performance index proposed in [18] is considered to improve the performances of the DFIG-based wind turbine under unknown disturbance and parameter variations. To this end a cascaded NGPC and stator-field-oriented control are investigated. In addition, an aerodynamic torque observer, defined from predictive control, is integrated into the control law. Contents lists available at ScienceDirect journal homepage: www.elsevier.com/locate/isatrans ISA Transactions 0019-0578/$ - see front matter & 2013 ISA. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.isatra.2013.08.004 n Corresponding author. Tel./fax: þ213 34205090. E-mail addresses: Ouari.Kamel@ymail.com (K. Ouari), to_reki@yahoo.fr, t_rekioua@yahoo.fr (T. Rekioua), Mohand_ouhrouche@uqac.ca (M. Ouhrouche). Please cite this article as: Ouari K, et al. Real time simulation of nonlinear generalized predictive control for wind energy conversion system with nonlinear observer. ISA Transactions (2013), http://dx.doi.org/10.1016/j.isatra.2013.08.004i ISA Transactions ∎ (∎∎∎∎) ∎∎∎–∎∎∎