Abstract
This paper deals with linear parameter-varying (LPV) modeling and output-feedback H1 control design for an offshore wind turbine. The controller is designed with consideration that not all the information in the feedback loop will be used. This constraint is incorporated into the design procedure. Constrained information means that a special zero-nonzero pattern is forced upon the
gain matrix. The constrained controller is obtained based on parameter-dependent Lyapunov functions and formulated in terms of linear matrix inequalities (LMIs). Since the functions are dependent on the wind speed and accurate wind speed measurements are rarely available in practice, an Extended Kalman filter (EKF) is used to estimate the wind speed. The controller is designed in such a way that it should maintain its stability and performance even if one of the sensors in the feedback loop should malfunction. The control objectives are to mitigate oscillations in the structure and drivetrain, to smoothen power/torque output in addition to keep the closed loop system stable. This should be achieved by means of individual blade pitch. A traditional procedure for designing a controller for such a system is to choose an operating point and assume it works in a suitable way under the influence of turbulent wind. In this paper, the wind turbine model is obtained from the software FAST. To design the controller, the model is linearized about several operating points. The degrees of freedom in the linearized model are chosen according to the controller objectives.
The linear models are valid within the span of operating points. Finally, the controller is tested on the fully nonlinear system under the influence of turbulent wind and a scenario where one of the sensors in the feedback loop is malfunctioning. The closed loop response of the presented controller is compared to the closed-loop response of the baseline controller included in the FAST package along with a controller designed based on a single linearized model.
gain matrix. The constrained controller is obtained based on parameter-dependent Lyapunov functions and formulated in terms of linear matrix inequalities (LMIs). Since the functions are dependent on the wind speed and accurate wind speed measurements are rarely available in practice, an Extended Kalman filter (EKF) is used to estimate the wind speed. The controller is designed in such a way that it should maintain its stability and performance even if one of the sensors in the feedback loop should malfunction. The control objectives are to mitigate oscillations in the structure and drivetrain, to smoothen power/torque output in addition to keep the closed loop system stable. This should be achieved by means of individual blade pitch. A traditional procedure for designing a controller for such a system is to choose an operating point and assume it works in a suitable way under the influence of turbulent wind. In this paper, the wind turbine model is obtained from the software FAST. To design the controller, the model is linearized about several operating points. The degrees of freedom in the linearized model are chosen according to the controller objectives.
The linear models are valid within the span of operating points. Finally, the controller is tested on the fully nonlinear system under the influence of turbulent wind and a scenario where one of the sensors in the feedback loop is malfunctioning. The closed loop response of the presented controller is compared to the closed-loop response of the baseline controller included in the FAST package along with a controller designed based on a single linearized model.
| Original language | English |
|---|---|
| Journal | IET Control Theory & Applications |
| Volume | 8 |
| Issue number | 1 |
| Pages (from-to) | 22-29 |
| Number of pages | 8 |
| ISSN | 1751-8644 |
| DOIs | |
| Publication status | Published - 1 Jan 2014 |
Keywords
- LPV modeling, constrained information, H1 control, wind turbine.