Wind Turbine Blade Fault Detection Method Based on TROA-SVM

Area of Science:

• Engineering
• Renewable Energy Systems
• Artificial Intelligence

Background:

• Wind turbines operate in harsh environments, leading to blade fatigue damage.
• Effective monitoring and maintenance are crucial for turbine longevity and operation.
• Existing fault detection methods may lack accuracy in complex conditions.

Purpose of the Study:

• To develop a novel fault detection method for wind turbine blades using acoustic signals.
• To enhance the accuracy and reliability of fault detection through advanced algorithms.
• To address the challenges posed by environmental factors on wind turbine blades.

Main Methods:

• Signal preprocessing of raw noise data from operational wind turbines.
• Feature extraction from time, frequency, and cepstral domains.
• Integration of Tyrannosaurus Optimization Algorithm (TROA) with Support Vector Machine (SVM) for classification.

Main Results:

• A comprehensive feature matrix was constructed, capturing multi-dimensional characteristics.
• Feature selection was performed to retain the most significant data.
• The TROA-SVM model achieved a recognition accuracy rate of 98.7%.

Conclusions:

• The proposed TROA-SVM method demonstrates superior effectiveness in wind turbine blade fault detection.
• This approach significantly improves upon traditional methods like SVM, KNN, and random forest.
• The method offers a reliable solution for monitoring and maintaining wind turbine health.