Wind Turbine Power Analytics Using ML to Cut False Alarms
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current wind farm monitoring systems face challenges in accurately detecting under-performance in wind turbines due to high false alarm rates and the lack of simultaneous computation of analytic outputs, leading to inefficient decision-making.
Innovation Solution
A machine-learning model-based analytic system that combines multiple wind performance analytics with supervised learning and continuous improvement, using techniques like principal component analysis and power ensemble to reduce dimensions and enhance precision, thereby minimizing missed classifications and false alarms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple individual analytic outputs are used to evaluate wind turbine performance, then the ability to detect performance issues is improved, but the false alarm rate increases and decision-making becomes less efficient
Solution Approach 1:
The patent combines multiple individual analytic outputs into a single integrated machine learning model that processes multiple features simultaneously. This merging approach integrates power curve analysis, availability metrics, and other performance indicators into a unified framework that reduces false alarms while maintaining detection accuracy.
Solution Approach 2:
The machine learning model acts as an intermediary between raw analytic outputs and decision-making. It processes and synthesizes multiple analytic streams through supervised learning, producing a single reliable performance assessment that eliminates the need for operators to interpret multiple conflicting individual analytics.
2Measurement precision
If comprehensive analytic outputs are computed to improve detection accuracy, then measurement precision is improved, but computational complexity and processing time increase
Solution Approach 1:
The system performs preliminary computations by pre-processing and feature-engineering analytic data before main analysis. It prepares cleaned, normalized, and relevant features in advance, reducing the computational burden during actual performance assessment while maintaining high detection accuracy.
Solution Approach 2:
The machine learning model transforms multiple complex analytic parameters into a simplified performance score through supervised learning. This parameter transformation converts complex multi-dimensional analytic outputs into a single interpretable metric that maintains precision while reducing computational complexity.
3Loss of information
If all analytic outputs are made available simultaneously to improve decision-making, then information completeness is improved, but the difficulty of interpreting and acting on the data increases
Solution Approach 1:
The system extracts only the most relevant performance information from multiple analytic streams and presents it through a simplified interface. It identifies and highlights key performance indicators and anomalies while filtering out redundant or less critical data, making decision-making efficient without losing essential information.
Solution Approach 2:
The unified machine learning model serves multiple functions simultaneously: it detects performance issues, ranks turbines by performance level, identifies anomalies, and provides actionable insights. This multi-functional approach consolidates multiple analytic purposes into a single system that improves both information availability and ease of operation.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method for controlling a wind turbine includes detecting a plurality of analytic outputs relating to power performance of the wind turbine from a plurality of different analytics. The method also includes analyzing the plurality of analytic outputs relating to power performance of the wind turbine. Further, the method includes generating at least one computer-based model of the power performance of the wind turbine using at least a portion of the analyzed plurality of analytic outputs. Moreover, the method includes training the computer-based model(s) of the power performance of the wind turbine using annotated analytic outputs relating to the power performance of the wind turbine. In addition, the method includes estimating a power magnitude of the wind turbine using the machine-learned computer-based model(s). As such, the method includes implementing a control action when the power magnitude of the wind turbine is outside of a selected range.