Wind Turbine Control Using Inferred Wind Shear Profiles
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing wind turbine control systems face challenges in efficiently managing wind shear variations, leading to suboptimal energy production and increased mechanical loads due to the high costs and performance limitations of sensors like LiDAR and met-mast sensors, which are sensitive to weather conditions.
Innovation Solution
A control system that includes measurement sensors, a computing device, and a regulation device to classify wind shear profiles using machine learning algorithms, such as random forests, to infer wind characteristics and generate command signals for adjusting blade pitch and torque, thereby optimizing wind turbine operation based on real-time wind conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If LiDAR or met-mast sensors are used to measure wind characteristics, then wind measurement capability is improved, but system cost increases and reliability deteriorates due to weather sensitivity
Solution Approach 1:
The wind turbine uses its own operational data (power output, rotor speed, blade pitch angles) to infer wind characteristics through machine learning algorithms, eliminating the need for external sensors. The system serves itself by deriving wind information from its own performance measurements.
Solution Approach 2:
The patent introduces an intermediary computational layer (machine learning model) that translates readily available operational data into wind characteristic information. This intermediary process converts power output and operational parameters into inferred wind speed and wind shear values without direct sensor measurement.
2Measurement precision
If traditional sensors are deployed to capture wind characteristics, then measurement accuracy is improved, but system complexity and cost increase
Solution Approach 1:
The patent replaces mechanical sensor systems (LiDAR, met-mast) with a computational/data-driven system. Instead of physical sensors measuring wind directly, the system uses machine learning algorithms processing operational data to infer wind characteristics, substituting mechanical measurement with computational analysis.
Solution Approach 2:
The system creates a virtual model or copy of wind characteristics through computational inference rather than physical measurement. The machine learning model generates a digital representation of wind conditions based on patterns in operational data, effectively copying wind information without direct sensing.
3Productivity
If wind characteristics are not accurately measured, then system cost is reduced, but energy production decreases and mechanical loads increase
Solution Approach 1:
The system implements feedback by continuously monitoring operational data (power output, rotor speed, pitch angles) and using this information to infer wind characteristics that feed back into control decisions. This closed-loop approach enables real-time adaptation to wind conditions without external sensors.
Solution Approach 2:
The machine learning model is pre-trained with historical data to recognize patterns and infer wind characteristics before they are directly observed. This preliminary preparation allows the system to quickly and accurately estimate wind conditions based on operational patterns, enabling proactive control adjustments.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A control system for a dynamic system including at least one measurement sensor. The system includes at least one computing device configured to generate and transmit at least one regulation device command signal to at least one regulation device to regulate operation of the dynamic system based upon at least one inferred characteristic.