Homepage of DataDriven@DCSC

For the wind energy community, model-based controller design becomes more and more important. Model-based controller synthesis necessitates a nominal description of the real plant. Nominal description of the plant can be derived from physical principles or using measurement data, respectively. Therefore, the latter is considered as a preliminary phase on the way towards a controller design. system identification of wind turbines is not only important at the design of a new turbine setup, but also when existing devices has to be re-identified in order to create a more up-to-date accurate model than the existing one. In this specific case, wind turbine system identification has to be performed in taking consideration the existing controller as well. This project is motivated by the closed-loop system identification problem of wind turbine systems. The data source of the system identification is based on nonlinear controller-in-the-loop simulations of a typical multi-MW wind turbine.

Active control is becoming more and more important for the wind energy community. If we compare the `old' stall turbines with today's individual pitch controlled turbines we see that the loads can be considerably reduced, leading to lighter or larger turbines. However, limited actuator bandwidth and component fatigue impose  significant constraints on the pitch system. Furthermore, with the trend to go offshore it is of interest to increase the rotor diameter as much as possible because the foundation costs of offshore wind turbines amount to a large part of the total costs. Due to the increasing size of wind turbines and the limitations of individual pitch control it is thus necessary to look ahead to new control concepts which can impose a force profile matching the distributed nature of turbulence, and  guarantee an economic lifetime of 20 years for the next generation of offshore wind turbines (diameter over 150 meters).

One novel concept is to use a number of control devices that locally change the force profile on the wind turbine blade to copy the spatially distributed nature of turbulence. The success of distributed load reduction greatly depends on the selection of appropriate sensors that measure the loads and a controller that manipulates the measured signals and generates an actuation signal.  This overall combination of sensors, actuators, and control is defined as the `smart' rotor concept. We contributed to the development of this new concept by showing the feasibility of the `smart' rotor under realistic wind turbine conditions (e.g. realistic disturbances, feedback control, and load measurements). On the designed experimental setup we showed that when the disturbance is known, perfect cancelation of the disturbance can be realized. However, under realistic circumstances the disturbance is not known and feedback control is required. For this situation we showed the broadband load reduction capabilities of the `smart' rotor for different load cases (e.g. with a sinusoidal disturbance we showed that we can reduce the amplitudes by 90% at the first eigenfrequency of the blade, see the movie below). The success of this proof of concept should help aerodynamicists and structural experts to embrace control engineering much earlier in their design cycle, and utilize it as a lever to create additional design freedom.

In this project an innovative approach of model based robust controller design will be developed based on the strong past performance of the members of the DCSC. The development of system identification tools for robust controller design will be extended towards Linear Parameter Varying (LPV) systems. Further, this fundamental experience will be combined with the application expertise the DCSC  has acquired in vibration reduction using "Smart" materials.

For more information in Dutch, go to Kennislink.

The trend in offshore wind turbines is to increase the rotor diameter as much as possible to decrease the cost per kWh. The increasing dimensions have led to a relative increase of the loads on the wind turbine structure, thus it is necessary to react to disturbances in a more detailed way: each blade separately. The load disturbances acting on an individual wind turbine blade are to a large extent deterministic, such as the tower shadow, wind shear, yawed error and gravity, and they depend on the rotation angle and speed. A repetitive controller can learn these periodic disturbances for fixed-speed wind turbines and variable-speed wind turbines operating aboverated. For relatively slow changing periodic disturbances it is expected that this control method can significantly reduce the vibrations in the wind turbine structure.

In many motion and control systems the controlled process generally consists of: an actuator, a control loop (e.g. software) and feedback information (sensors). Failures in one of these components normally lead to unacceptable performance of the system. In many cases though, the failure itself is not that critical, it is the combination with the control loop or the feedback elements that might lead to an unstable system or unacceptable system performance. By detecting the failure and subsequently adaptation of the control loop in the process, the performance might be adjusted to a level that is still within acceptable limits. The detecting of failure as well as the adjustment of the control loop is defined as 'Fault Tolerant Control'. The goal of fault-tolerant control is to prevent those simple faults in a system or its sensors and actuators develop into serious failures. Fault-tolerant control increases the availability of the system and reduces the risk of safety hazards. To achieve fault-tolerant control, intelligent methods needed to be developed for on-line fault detection and diagnosis, automatic condition assessment and calculation of remedial actions or controller reconfiguration. In this research topic the focus will be on novel fault-tolerant control methods and to apply and test them on the drive-by-wire systems of SKF.