Search

This paper has the goal of comparing one and two-sided wake steering, in terms of plant-level power capture and loading on the upstream and downstream turbines. Wind tunnel tests were conducted in a boundary layer wind tunnel in typical offshore conditions, using three scaled controlled and sensorized wind turbines, governed by an open-loop plant controller. Dynamic wind direction changes were obtained by rotating the wind tunnel turntable. Results indicate that one-sided steering achieves less than half the power gain of double-sided steering with respect to the uncontrolled greedy case. Additionally, although loading on the steering turbine is reduced, loading downstream is also significantly higher than for double-sided steering. On the other hand, actuator duty cycle is much reduced for one-sided steering, thanks to a reduced activity around the wind-aligned condition.

The paper studies the effects of model fidelity on the performance of a robust open-loop wake deflection controller. For this purpose, a small cluster of scaled turbines is operated in a boundary layer wind tunnel, where wind direction changes are simulated by a rotating turntable. Results indicate that better models lead to improved power gains and reduced fatigue loading, motivating the development of improved accuracy wind farm flow models.

The standard material fatigue estimation process is implemented in an Economic Nonlinear Model Predictive Controller. The process comprises cycle identification, mean-stress effect, Wohler curves and Linear Damage Accumulation. For cycle identification, this novel implementation uses a standard Rainflow-algorithm which is substituted periodically by linear functions based on its output. The implementation is compared to two straightforward methods for fatigue penalization using a wind turbine model and exhibits higher profit over the entire range of tested reference wind velocities.

An engineering wake model is updated online based on measurements provided by shaded turbines. Departing from other approaches, the measurements include information on the impinging wakes, obtained by a wake detector based on measured rotor loads. The updated model exhibits improved prediction capabilities, and it can be used for implementing a model-based wind farm controller.

This paper considers the preliminary investigation of a passive flap for the mitigation of loads on wind turbines. The flap moves passively in response to blade vibrations, opposing them, thereby yielding an attenuation of loads. In comparison to active flaps, this solution has the advantage of not requiring sensors nor actuators, resulting in a particularly simple implementation, with potential benefits in manufacturing and maintenance costs. The paper first describes the concept of the novel passive flap, highlighting its main characteristics. A proof of concept of the new idea is then given by a simulation study conducted with the combination of a sectional model of the flap and a multibody model of the rest of the machine. Results, obtained for a 10MW wind turbine, indicate the ability of the passive flap in attenuating blade vibrations in a significant frequency range, which in turn yield a reduced fatigue damage to the structure without noticeable effects in terms of power production.

A tightly-coupled vision-aided inertial navigation system (TC-VA-INS) is proposed in this work, as a synergistic incorporation of vision with other sensors. In order to avoid the loss of information possibly resulting by the preprocessing of visual information, a best set of tracked feature points and readings of a low cost IMU are directly fused together within a vehicle state estimator. Instead of using 3D reconstruction, a vision based model is derived by using the trifocal tensor to propagate feature points across time steps, so as to express geometric constraints among three consecutive scenes. A kinematic model is used to account for the vehicle motion, and a Sigma Point Kalman Filter (SPKF) is used to achieve a robust state estimation in the presence of non-linearities. The proposed formulation is tested and demonstrated with a real dynamic indoor dataset. Results show improved estimates than in the case of a classical visual odometry approach, even in GPS-denied conditions and when magnetometer measurements are not reliable.

We describe a novel adaptive non-linear model predictive controller which is based on the idea of neural- augmentation of reference elements, both at the level of the reduced model and at the level of the control action. The new methodology is primarily motivated by the desire to consistently incorporate existing legacy modeling and control techniques into an adaptive non-linear, yet real-time-capable, control framework. The proposed procedures are demonstrated in a virtual environment with the help of the classical model problem of the double inverted-pendulum, and with the more challenging reflexive control of an autonomous helicopter.