Modelling Techniques for Optimal Wind Farm Performance

Automatic identification of parameters for wind turbine generators to improve control performance, safety and reduce maintenance costs. The method involves controlling the generator to no-load start and shut down by adjusting blade pitch. During this transient, voltages and flux linkage can be measured to determine generator parameters like rotor angle, pole pairs, and flux linkage. Closing the circuit breaker at shutdown allows measuring stator resistance and inductance. This avoids manual parameter entry errors and reduces software versions compared to static tables.

Neural network based control of wind turbines to optimize power extraction and robustness. The control uses a trained neural network to estimate the optimal rotor speed and maximum power for a wind turbine given the wind speed. This is done by feeding wind speed and tip speed ratio into the network and outputting the optimal rotor speed and maximum power. The wind turbine is then operated at the estimated optimal speed determined by the neural network for any wind condition. This allows the turbine to track maximum power and efficiently adjust rotor speed in response to wind changes.

Optimizing power output of multi-rotor wind turbines by accurately determining the optimal yaw angle for the turbines. The method involves measuring wind power parameters over a range of relative wind directions for each rotor nacelle assembly. The maximum power points are determined for each nacelle. The average of these maximum points is used as the control wind direction for the common yaw system. This aligns the rotors with the wind for maximum power.

Predictive monitoring of wind turbine components like chopper resistors to prevent overheating and failures. It uses two separate thermal models, one to estimate internal temperatures based on external conditions and power loss, and another to predict future surface temperatures using the internal temperatures. By running these models in sequence, it allows detecting potential overheating issues before they become critical. If the predicted surface temperature exceeds limits, action can be taken to shut down the turbine before component failure or safety risks occur.

Neural network based control of wind turbines that efficiently tracks and adjusts maximum power as wind speed changes. A neural network model is trained using wind speed and tip speed ratio samples to output maximum power and optimum rotor speed. This model is used to control the wind turbine's reference angular speed instead of directly tracking maximum power. This allows the turbine to rapidly adapt to changing wind speeds and extract maximum power. The neural network is trained using a dataset of averaged wind speeds, which is then used to test the model's accuracy. The neural network output is used to control the wind turbine's rotor speed for optimal power extraction.

Vertical wind turbine with a blade pitch motor that allows optimal blade angle adjustment for maximum efficiency and longevity. The blade pitch motor is mounted between the upper and lower blade sections, allowing symmetric torque distribution along the blade span. The pitch motor also supports the blade weight. This avoids external actuators and guys for blade angle control. The pitch motor can have absolute and relative position sensors. The turbine also has a compact transmission with planetary stages. The control calculates optimal blade angles based on wind speed and direction. This enables continuous, smoothest blade pitch control compared to discrete steps.

Method and arrangement for determining pitch speeds of wind turbine blades to reduce bearing damage while allowing rapid and reliable blade pitching. The method involves calculating the pitch speed based on the blade bearing moment. If the moment is below a reference, the pitch speed exceeds it. This prevents high moments causing damage. If the moment increases, the pitch speed decreases. If the moment decreases, the pitch speed increases to catch up. This prevents overshooting. It balances speed to minimize damage while allowing quick pitching.

Optimizing power generation in a wind farm by dynamically adjusting the operating parameters of individual wind turbines based on local wind shear conditions. The method involves measuring or predicting the vertical wind shear profile above the wind farm. This profile is then used to determine optimal adjustments to turbine settings like blade pitch, rotor speed, and yaw angle. These adjustments are made in real-time to optimize overall farm power production considering the non-standard wind shear profile and turbine interactions.

Optimizing the operation of wind turbines to maximize lifetime or energy production by automatically determining the best combination of control features based on user-selected targets. The optimization involves estimating the optimization parameter (lifetime, energy, power demand satisfaction) for different combinations of activated control features, and selecting the combination that best meets the target while considering boundary conditions. The method considers the impact of all possible feature combinations to find the optimal strategy.

Monitoring air flow over a wind turbine blade to improve blade design and control by placing a simple device on the blade surface. It has two air inlets facing opposite directions along an axis. A sensor module with two pressure sensors, one connected to each inlet, outputs signals. Processing determines flow direction based on pressure difference sign and speed based on magnitude. Additional inlets and sensors can provide additional flow data. The device is integrated in the blade and wirelessly communicates.

A stability evaluation method and system for direct-drive wind turbine generators that allows online assessment and parameter adjustment to improve stability of the system. The method involves calculating the energy gradient at the wind turbine terminal using voltage, current, and angle measurements. A negative energy gradient indicates instability. The gradient is influenced by factors like PLL parameters, wind turbine current levels, and transmission line resistance. By understanding these relationships, the method proposes adjustments to critical parameters like PLL gains and wind turbine current limits to improve stability.

Accurately determining the angular position of a wind turbine generator using an encoder sensor, even when the encoder has imperfections that distort the position signal. The method involves compensating for the distortions to improve position determination accuracy. The compensation signal is calculated based on the imperfection characteristics and applied to the raw position signal to correct for the distortions. This modified position signal is then used for tasks like wind turbine control.

Operating wind turbines more efficiently by using real-time loading and travel data to estimate component damage accumulation and make adjustments to preventive maintenance and turbine operation. Instead of relying on simulated loading histories, the method involves generating actual cumulative loading histograms based on the measured or estimated loading and travel metrics during turbine operation. This allows accurate damage tracking and decision making based on the true fatigue experience of the components. The histograms are applied to life models to determine current damage levels and actions like shutdown, power reduction, or maintenance scheduling are implemented based on that.

Neural network based control of wind turbines using permanent magnet synchronous generators (PMSG) that can efficiently and robustly track maximum power at varying wind speeds. The control involves training a neural network with wind speed and other parameters to output optimum rotor speed and maximum power. This network is then used to control the PMSG turbine based on real-time wind speed measurements, allowing the turbine to operate at maximum power in varying winds.

A method for controlling wind turbines in the presence of solar heating. The method involves determining the thermal gradient of the tower due to solar heating, calculating a displacement of the turbine reference point from nominal position due to tower thermal expansion, and deriving a correction factor to mitigate the impact of the displacement. This correction factor is used to generate setpoints for turbine components to compensate for the displacement effects. This helps accurately control the turbine despite solar heating displacements.

Modifying the natural frequency of a wind turbine tower to prevent resonance with blade passing and rotor frequencies. The modification involves suspending weights from the upper flange of the tower using rigid supports. By adding weights to the tower top, it increases the natural frequency. This provides a way to customize the tower frequency to separate it from the blade passing and rotor frequencies if they become too close due to variation in tower mass or foundation stiffness. It avoids the need for large design tolerances or safety margins for eigenfrequency separation.

Wind turbine drag device to reduce blade rotation and fatigue during idling and improve aerodynamic damping. The device has a perpendicular planar surface that increases drag and stops airflow when attached to blades, nacelle, tower, or rotating element. It can be a foldable sail or kite with tightening attachment to fix position. Attaching between blades increases drag and balances forces. This reduces rotational speed and oscillations without generator power.

Operating wind turbines based on actual load histories rather than simulations to optimize maintenance and lifespan. The method involves tracking component loading and travel metrics during turbine operation. Cumulative load histograms are generated from the actual data. A life model is applied to these histograms to determine actual component damage accumulation. Corrective actions are implemented based on the damage levels.

Data-driven wind farm frequency control method based on dynamic mode decomposition for high-rate wind power integration into grids. The method allows wind farms to participate in grid frequency response without needing complete models of the turbines. It uses dynamic mode decomposition to find accurate low-dimensional models of the wind turbine dynamics. These models are then used in a frequency control algorithm to calculate optimal frequency corrections for the wind farm. This allows fast, data-driven frequency response without relying on complex turbine models.

Monitoring wind turbines in a wind farm using a holistic view of the whole farm to detect anomalies in individual turbine behavior. The method involves building a global model of turbine relationships using reference data from a fault-free period. Local models are built for clusters of turbines based on their similarity. Test data is projected onto the local models to derive nonconformity indices. Turbines with high indices are flagged as critical. This holistic approach leverages the similarity between turbines to improve fault detection compared to just monitoring individual turbines.