Identifying wind component failures before they occur

Predict uses statistical models, artificial neural networks and machine learning to identify wind turbine component failures before they occur.

The feature alarms users on changes in temperature that indicate need for maintenance. Predict’s advanced statistical models developed by Greenbyte’s Head of Research, Dr. Pramod Bangalore have been optimised for high accuracy and in collaboration with Greenbyte’s Head of Technology, Mikael Baros, been put to vigorous testing to ensure high accuracy.

Predict estimates the expected temperature for critical components, compares that estimated data to the actual measured values, and enables intelligent and early detection of developing failures. The pilot study on Predict detected faults 2-9 months in advance, achieved 94% accuracy and showed a 23% reduction of cost, and the software keeps learning and outperforming itself.

Multiple benefits accrue from this heavily researched feature. Early indication for component failure can reduce down-time, maintenance cost and increase component life. It enables operators and managers to act with a plan instead of acting within a crisis, and allows them agency on making informed maintenance decisions. 

Developing Predict has been a journey of knowledge for Greenbyte and an evidence of innovation for the industry. Director of Technology, Mikael Baros has been describing the Artificial Intelligence and machine learning part of the journey in a thrilling blog series The Greenbyte recipe for Artificial Intelligence in renewable energy.

More specifically in the first article, he narrates the imminence of component failures in the lifetime of a wind turbine: We expect turbines to operate 24 hours a day, 7 days a week. If we did the same with a car it would only last us 8 months! Hence it is not surprising that these poor turbines fail (too) often. It is estimated that up to 30% of the total life-cycle cost of a wind farm is due to failure and maintenance activities. 

The rest of the series continues to unravel how the data crunching process was applied to the first test customer. Stay tuned for the big and final reveal of Greenbyte’s Predict recipe, published on 4th September.

In the meantime, Head of Research at Greenbyte, Dr. Pramod Bangalore is holding a compelling webinar on Predict on 29th August, where he unveils the science behind the technology. This webinar is a valuable source of knowledge for users of Greenbyte Energy Cloud, industry professionals and data scientists alike. Interesting parties can sign up to attend the webinar here.