Wind Turbine Wake Prediction Using Data-Driven Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for controlling wind turbines in a wind farm to avoid wake effects are inefficient due to the stochastic nature of wind, leading to sub-optimal annual energy production (AEP) as they do not effectively account for the wake impact on downstream turbines.
Innovation Solution
A computer-implemented method using a trained data-driven model, such as a Deep Neural Network, processes measurement values from downstream turbines to determine the wind speed profile and wake center, enabling precise prediction of the wake's horizontal and vertical position, which can be used to optimize yaw angles and control upstream turbines to steer the wake away from downstream turbines.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If simulations and engineering models are used to predict wake behavior, then wake prediction capability is improved, but model development difficulty increases due to the stochastic nature of wind
Solution Approach 1:
The patent replaces complex physics-based simulations and engineering models with a data-driven machine learning approach. Instead of using traditional mechanical/fluid dynamics models that are difficult to develop due to wind's stochastic nature, the invention uses measured data from anemometers and LIDAR to train predictive models, substituting the complex modeling process with data-driven pattern recognition.
2Ease of operation
If each turbine operates independently to face the wind, then individual turbine control simplicity is improved, but overall wind farm energy production decreases due to wake effects
Solution Approach 1:
The patent implements a feedback mechanism where wake predictions from the data-driven model are used to adjust upstream turbine yaw angles. The system continuously monitors wake conditions and adjusts turbine orientations based on predicted wake trajectories, creating a closed-loop control system that optimizes overall farm production while maintaining operational simplicity through automated decision-making.
3Productivity
If upstream turbines adjust yaw angle to steer wake away from downstream turbines, then downstream turbine performance is improved, but upstream turbine individual energy capture may be reduced
Solution Approach 1:
The patent uses preliminary wake prediction through the trained data-driven model to anticipate wake trajectories before they reach downstream turbines. By predicting wake behavior in advance, the system can proactively adjust upstream turbine yaw angles to steer wakes away from downstream turbines, preventing energy loss before it occurs rather than reacting after the fact.
Data Source
Figure 1~2
AI summary
A method for computer-implemented determination of a wind speed profile information (WSPI) of a wind field (WF) approaching a turbine (1, 2) of a wind farm is described. The wind farm comprises a first upstream wind turbine (1) and a second downstream wind turbine (2), wherein at each time point of one or more time points during the operation of the wind farm the following steps are performed: obtaining a number of measurement values (MV) of the second wind turbine (2), the number of measurement values (MV) being captured by respective sensors of the second wind turbine (2) to monitor and/or control the second wind turbine (2); and determining a wind speed profile information (WSPI) by processing the number of measurement values (MV) by a trained data driven model (MO), where the number of measurement values (MV) is fed as a digital input to the trained data driven model (MO) and the trained data driven model (MO) provides the wind speed profile information (WSPI) as a digital output, where the wind speed profile information (WSPI) at least comprises a horizontal and/or vertical position parameter of a wake center (WC) of the wind field in direction to the second wind turbine (2) and caused by the first wind turbine (1).