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 industry and the academic community.
PO196: Edge-enabled multi-modal monitoring and remaining useful life prediction for wind turbine drivetrains
Cédric Peeters, Postdoctoral researcher, Vrije Universiteit Brussel
Abstract
We present a multi-modal condition-monitoring and remaining useful life (RUL) framework for wind-turbine drivetrains that spans end-to-end, from sensing hardware to end-user insight. The system ingests high-sample-rate, long-duration vibration signals, SCADA temperatures and operating data, and online oil debris counts. An in-house data-acquisition (DAQ) platform with edge computing performs local data quality checks, preprocessing, and on-device inference, cutting storage/transfer needs and central compute. These systems have been deployed in field campaigns on more than 25 offshore turbines. Analytics proceed in two stages. First, advanced signal processing compensates speed variations (virtual encoder estimation), cleans signals (adaptive filtering, periodic–stochastic separation), and extracts thousands of indicators in time and spectral domains. Second, an explainable and scalable ML layer performs normal behavior modelling for temperatures, oil-debris trending, and vibration-based features; it then fuses deviations into sensor- and component-level alarm trends that are traceable back to the base indicators. For prognosis, a modular RUL methodology continuously updates forecasts directly from streaming risk scores, even with limited data, and allows the conservatism of predictions to be tuned per component. An ensemble of forecast models provides robustness across failure modes; bootstrap uncertainty quantifies confidence; and time-sensitive weighting adapts to evolving operating conditions. By unifying multi-sensor findings with explainable ML and edge-enabled deployment, the framework turns raw drivetrain data into actionable maintenance intelligence.
No recording available for this poster.
