Vibration Analysis for Fault Detection of Wind Turbines by Combining Machine-Learning Techniques and 3D Scanning...

Area of Science:

• Mechanical Engineering
• Electrical Engineering
• Data Science

Background:

• Wind turbine operational efficiency relies on the integrity of mechanical and electrical components.
• Component degeneration can lead to costly failures and downtime.
• Predictive maintenance strategies are crucial for optimizing wind turbine lifespan and performance.

Purpose of the Study:

• To apply range-resolved interferometry (RRI) and machine learning (ML) techniques for wind turbine maintenance.
• To develop an automated system for predicting, detecting, and anticipating component degeneration.
• To validate the accuracy of RRI measurements against traditional methods.

Main Methods:

• Utilized range-resolved interferometry (RRI) with laser scanning to detect vibrations.
• Applied relevant machine learning (ML) techniques for autonomous learning and analysis.
• Collected in-process measurements during two distinct failure states.

Main Results:

• Successfully detected vibrations indicative of two different failure states in wind turbine components.
• Demonstrated that RRI in-process measurements align with manual, laser scanning, and hand measurements.
• Validated the potential for automatic and autonomous learning for component degeneration analysis.

Conclusions:

• The proposed RRI and ML method is highly effective for monitoring and diagnosing faults in wind turbines.
• This system offers a pathway to low-cost, in-process measurements for predictive maintenance.
• Automated fault detection enhances the reliability and efficiency of wind energy infrastructure.