Posters
Come meet the poster presenters to ask them questions and discuss their work
We would like to invite you to come and see the posters at our upcoming conference. The posters will showcase a diverse range of research topics, and will give delegates an opportunity to engage with the authors and learn more about their work. Whether you are a seasoned researcher or simply curious about the latest developments in your field, we believe that the posters will offer something of interest to everyone. So please join us at the conference and take advantage of this opportunity to learn and engage with your peers in the academic community. We look forward to seeing you there!
PO158: Vibration Anomaly Detection of Wind Turbines Components Using Machine Learning and Statistical Methods.
Thiago Kleis, Global Sales Executive, AQTech Power Prognostics
Abstract
Wind energy and the number of wind turbines in operation have been spreading around the world. This spreading brings the benefits of renewable energies but also comes together with the challenges for Operation and Maintenance (O&M) to handle this number of assets. One way to tackle these problems is to use data-driven approaches, taking advantage of various information that can be inferred from wind turbine data. To obtain this information, the use of statistical methods and machine learning models can be very helpful, they can even be used to infer the health state of the wind turbines. A key question is what data is best suited to analyze the turbine's health. ISO 10816-21 recommends monitoring vibrations in some components of wind turbines to prevent potential component failures. Based on the behavior of the variables during the wind turbine operation, and their intrinsic relationships, it is possible to identify some deviations of the expected behavior or identify when the intrinsic relationships are broken. This can be achieved through statistical inference on historical turbine data and by utilizing machine learning models. The central concept of this work is the integration of statistical methods and machine learning models. Machine learning models are used to predict turbine behavior, and deviations from actual behavior are quantified. Statistical methods are then applied to analyze the distribution of these deviations, identifying outliers in vibration signals that indicate degradation in wind turbine components. The results of this methodology demonstrate its efficiency in identifying component degradation, accurately identifying outliers associated with severe degradation that could lead to failures. The methodology was tested with real wind turbine data and successfully pinpointed components requiring human intervention to prevent breakdowns, underscoring its practical value in ensuring the reliability and longevity of wind power generation systems.
