Velocity-Based Wind Turbine Blade Deblurring Using Richardson-Lucy Algorithm

In response to current expensive and unsafe wind turbine blade inspection techniques, alternative inspection approaches involving aerial imagery from autonomous drones and turbine blade damage-detecting machine learning algorithms have been explored. These methods require sharp, non-blurry images of wind turbine blades for accurate classification of blades as healthy or damaged. This means both current and newly explored methods can only inspect non-rotating turbines, resulting in turbine power generation losses during inspection time. To implement the proposed machine learning inspection methods during wind turbine operation hours, sharp images of the rotating turbine blades must somehow be acquired. In this research, a velocity-based Richardson-Lucy algorithm method is used to deblur aerial images of rotating wind turbine blades. This deblurring approach is first applied to artificially blurred wind turbine blade images, and then to real images of rotating turbine blades.