Wind Farm Yaw Consensus Control Using Distributed Reinforcement Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Wind turbines in a wind farm typically operate independently, leading to suboptimal performance due to lack of coordination and reliance on local wind measurements, which can be noisy and unreliable, resulting in dynamic yaw misalignment and reduced power production.
Innovation Solution
A system and method using an augmented Lagrangian method to calculate a consensus wind estimate from measurements across multiple wind turbines, allowing for improved wind direction estimation and adjustment of operating parameters to enhance yaw alignment and reduce unnecessary movements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If wind turbines operate independently using local wind measurements, then each turbine can control its own yaw direction, but the wind direction estimates become noisy and unreliable leading to dynamic yaw misalignment
Solution Approach 1:
The patent combines wind direction measurements from multiple independent turbines into a consensus estimate. Each turbine shares its local measurements with neighboring turbines, and through iterative consensus algorithms, they collectively produce a more reliable wind direction estimate that reduces noise and improves accuracy compared to individual measurements.
Solution Approach 2:
The patent introduces a consensus algorithm as an intermediary process that mediates between individual turbine measurements and the final wind direction estimate. This intermediary layer processes and reconciles measurements from multiple turbines, producing a consensus value that is more reliable than any single measurement while maintaining distributed operation.
2Productivity
If wind turbines use noisy local wind measurements for yaw control, then they can respond quickly to local conditions, but they experience dynamic yaw misalignment and reduced power production
Solution Approach 1:
The patent merges wind measurement data from multiple turbines to create a consensus estimate with higher precision. By combining measurements from neighboring turbines through iterative consensus algorithms, the system produces a more accurate wind direction estimate that reduces yaw misalignment and improves power production.
Solution Approach 2:
The patent implements a feedback mechanism where turbines continuously share their measurements and update their estimates based on consensus values from neighboring turbines. This feedback loop allows the system to converge on a more accurate wind direction estimate while maintaining responsiveness to changing conditions.
3Productivity
If wind turbines operate independently without coordination, then the system is simpler to implement, but the overall wind farm performance is suboptimal
Solution Approach 1:
The patent segments the wind farm control into distributed autonomous turbines that each perform local processing and decision-making. Each turbine independently executes the consensus algorithm using only measurements from neighboring turbines, eliminating the need for centralized control while achieving coordinated optimization of wind farm performance.
Solution Approach 2:
The patent enables each turbine to serve itself by autonomously participating in the consensus algorithm. Each turbine independently processes its own measurements, communicates with neighbors, and adjusts its yaw control based on the consensus estimate, without requiring external centralized control.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Disclosed herein are methods, systems, and devices for utilizing distributed reinforcement learning and consensus control to most effectively generate and utilize energy. In some embodiments, individual turbines within a wind farm may communicate to reach a consensus as to the desired yaw angle based on the wind conditions.