Development of UAV-image Based Wind Turbine Blade Crack Detection Software
Project: Research › ARG
DescriptionWind farms have started the application of the Unmanned Aerial Vehicales (UAVs) in the inspection of wind turbine blade healths. This project aims at developing a software to serve as an artificial intelligence for automatically detecting wind turbine blade surface cracks based on images taken by UAVs. The PI has recently developed a Haar-like feature based data-driven model for rapidly detecting surface cracks of wind turbine blades and validated the capability of the developed model based on a set of blade images taken by UAVs offered by the PI’s industrial partner. In this project, the PI will firstly conduct a 9 months data collection to obtain sufficient blade images taken by UAVs for training a more general data-driven model. Next, the PI will develop the alpha version of the proposed software, a web-based application. After that, a field testing will be conducted to assess the computational efficiency and effectiveness of the alpha version of the software. An improvement will be applied to the software so that an ultimate goal that the detection of blade surface cracks with an accuracy higher than 95% and a computational time less than 10 seconds in each detection can be achieved.
|Effective start/end date||1/06/17 → …|