Offshore Wind Turbine Blade Monitoring Using Computer Vision and AI
Lead Participant:
CRACK MAP LTD
Abstract
Commercial Wind Turbines (WT), situated in remote and challenging environments, endure fluctuating loads, causing significant mechanical stresses and frequent failures. The reported Operations and Maintenance (O&M) cost, comprising 25-30% of the total wind power generation cost, underscores the critical role of O&M in addressing the growing needs of the wind energy industry.
Aging wind turbines face a rising risk of blade failures, leading to significant financial losses and downtime, especially for offshore assets. Thus, monitoring and promptly identifying blade defects are crucial.
This project focuses on developing physics-enhanced machine learning algorithms to automate the detection, localisation, and classification of damage as well as quantifying their severity and extent from internal and external images taken from offshore wind turbine blades in operation, using physics-based model, removing human interaction for damage evaluation and making decision. The project will also focus on developing a method comprehensive internal blade detection and flight control of an Unmanned Aerial Vehicle (UAV) within the blade. One of the scopes of this project is to develop a navigation and sensing approach to operate the drone in a confine, GPS-denied environment and perform the inspection autonomously.
The objectives of this project are:
\*To develop external drone-equipped with a super-high-resolution camera for external inspection of operational wind turbine blades suitable for offshore wind.
\*To develop internal drone-equipped with a hybrid sensors fusion (i.e., camera, lidar, altimeter, etc) for internal GPS-denied flight control and inspection.
\*To develop advance machine vision systems for damage detection/classification for external and internal blade.
\*To develop a digital twin to create a repository in which the historical damage occurring on a blade is recorded and monitored.
\*To industrially validate the developed in-house algorithm for inspecting operational external blade.
\*To perform field tests at the Wind Technology Testing Center on a utility-scale wind turbine blade to demonstrate the accuracy of the proposed solution for internal inspection of wind turbine blades.
\*To provide more accurate guidance on whether a same damage evaluation (physics-based model) procedure should be used for UK & US, as the environmental/operational conditions may change.
To complete the novel objectives, the project brings together a multidisciplinary team of researchers in robotics/drone, AI/machine learning, physics modelling, and digital twin, from Crack Map Ltd (Ilosta(tm)), Queen's University Belfast (QUB), and Air Tech Integrity Ltd, and the US partners: University of Massachusetts Lowell (UML) and Wind Technology Testing Center (WTTC).
Aging wind turbines face a rising risk of blade failures, leading to significant financial losses and downtime, especially for offshore assets. Thus, monitoring and promptly identifying blade defects are crucial.
This project focuses on developing physics-enhanced machine learning algorithms to automate the detection, localisation, and classification of damage as well as quantifying their severity and extent from internal and external images taken from offshore wind turbine blades in operation, using physics-based model, removing human interaction for damage evaluation and making decision. The project will also focus on developing a method comprehensive internal blade detection and flight control of an Unmanned Aerial Vehicle (UAV) within the blade. One of the scopes of this project is to develop a navigation and sensing approach to operate the drone in a confine, GPS-denied environment and perform the inspection autonomously.
The objectives of this project are:
\*To develop external drone-equipped with a super-high-resolution camera for external inspection of operational wind turbine blades suitable for offshore wind.
\*To develop internal drone-equipped with a hybrid sensors fusion (i.e., camera, lidar, altimeter, etc) for internal GPS-denied flight control and inspection.
\*To develop advance machine vision systems for damage detection/classification for external and internal blade.
\*To develop a digital twin to create a repository in which the historical damage occurring on a blade is recorded and monitored.
\*To industrially validate the developed in-house algorithm for inspecting operational external blade.
\*To perform field tests at the Wind Technology Testing Center on a utility-scale wind turbine blade to demonstrate the accuracy of the proposed solution for internal inspection of wind turbine blades.
\*To provide more accurate guidance on whether a same damage evaluation (physics-based model) procedure should be used for UK & US, as the environmental/operational conditions may change.
To complete the novel objectives, the project brings together a multidisciplinary team of researchers in robotics/drone, AI/machine learning, physics modelling, and digital twin, from Crack Map Ltd (Ilosta(tm)), Queen's University Belfast (QUB), and Air Tech Integrity Ltd, and the US partners: University of Massachusetts Lowell (UML) and Wind Technology Testing Center (WTTC).
Lead Participant | Project Cost | Grant Offer |
---|---|---|
CRACK MAP LTD | £135,683 | £ 94,978 |
  | ||
Participant |
||
QUEEN'S UNIVERSITY OF BELFAST | £84,878 | £ 84,878 |
AIR TECH INTEGRITY LIMITED | £71,272 | £ 49,890 |
People |
ORCID iD |
Saber Khayatzadeh (Project Manager) |