Improving Image Resolution for Drone-Borne Inspection of Wind Turbine Blades

Abstract

Inspections of wind turbine blades, both exterior and interior, present significant challenges due to limited accessibility as well as physical and measurement constraints. Key barriers to effective monitoring include the lack of suitable high-resolution distributed sensors, high cost and safety risks associated with up-tower technician deployment, limited availability of skilled personnel, and confined spaces involved in internal inspections. Advanced technologies such as drones, computer vision, and artificial intelligence offer transformative potential by enabling faster, more accurate, and cost-efficient inspections, leading to improved reliability and reduced overall energy costs. However, drone-based inspections typically require shutting down turbines to ensure the safe operation of unmanned aerial vehicles (UAVs) near the assets. To address this limitation, this research introduces a novel method referred to as Stack-Average (SA), which combines multiple images of a wind turbine to produce a super-resolved image with higher resolution. The SA method facilitates damage detection from greater distances (50+ meters) and eliminates the need to shut down turbines during inspections. The performance of the SA method was evaluated on a 1.5- meter section of a wind turbine blade. Results demonstrated the method’s accuracy in identifying various damage types as a function of image capture distance and the number of images used for super-resolution. This approach has the potential to improve image resolution for a given working distance and thereby enhance remote monitoring of wind turbine blades or other large-scale structures.