Farm-wide dynamic drivetrain event and diagnosis tracking using multi-modal data

This work presents an advanced study on improving condition monitoring systems or CMS for wind turbine drivetrains through multi-modal data. We present an innovative AI-based CMS framework that integrates temperature monitoring, oil debris analysis, and vibration data into a comprehensive assessment of drivetrain health. By employing physics-based data-driven techniques to analyze the various data streams, our approach enables more reliable and precise detection and diagnosis of incipient and evolving drivetrain issues. Additionally, vibration data allows for the characterization of the effects of dynamic events such as wind gusts on the drivetrain throughout the farm. This work demonstrates the AI-based monitoring and event-tracking approach on an extensive experimental field dataset measured on wind turbines on two offshore wind farms. The findings illustrate the added value of multi-modal condition monitoring and the benefit of tracking the drivetrain response to farm-wide events.