Structural Health Monitoring for Jacket-Type Offshore Wind Turbines: Experimental Proof of Concept

Area of Science:

• Engineering
• Structural Health Monitoring
• Renewable Energy

Background:

• Offshore wind turbines are increasingly deployed at greater depths, necessitating robust structural integrity solutions.
• Jacket foundations are crucial for supporting these turbines in challenging marine environments.
• Monitoring the structural health of these underwater foundations is essential for operational safety and longevity.

Purpose of the Study:

• To develop and validate a data-driven methodology for diagnosing structural damage in offshore wind turbine jacket foundations.
• To address the challenge of unmeasurable wind excitation by utilizing vibration-response-only data.
• To compare the effectiveness of machine learning classifiers for fault diagnosis in these structures.

Main Methods:

• Simulating wind excitation as Gaussian white noise and collecting accelerometer data.
• Pre-processing data using group-reshape and column-scaling techniques.
• Employing Principal Component Analysis (PCA) for dimensionality reduction and feature extraction.
• Classifying damage using k-Nearest Neighbor (k-NN) and Support Vector Machine (SVM) algorithms, validated by 5-fold cross-validation.

Main Results:

• The study successfully developed a reliable fault diagnosis method for jacket foundations.
• The Support Vector Machine (SVM) classifier demonstrated superior performance compared to the k-NN algorithm.
• Experimental validation on a small-scale structure confirmed the reliability of the proposed approach.

Conclusions:

• The developed methodology provides a reliable means for structural health monitoring of offshore wind turbine foundations.
• The data-driven approach, utilizing vibration-response-only data, is effective even with unmeasurable excitation.
• The SVM classifier is recommended for its high accuracy in diagnosing structural damage in these critical infrastructure components.