Automated Detection and Localization of Wind Turbine Deformations Through YOLO11 and UAV Systems

This study presents an advanced model for the automated detection of turbine blade deformations, leveraging Industry 4.0 technologies such as machine learning, computer vision, and GPS-based localization. The majority of the dataset used for training the model was created with a UAV system equipped with a high-resolution camera, enabling precise and comprehensive data collection from turbine blades during routine inspections. YOLO11 model was trained to identify and classify common turbine blade deformations, including cracks, surface wear, blade bending, and thermal deformations. By integrating GPS support, the UAV system detects deformations and accurately pinpoints their locations, facilitating efficient and targeted maintenance interventions. The model achieved an overall precision of over 80%, demonstrating robust performance in detecting visible deformations such as cracks and blade bending. However, challenges persist in identifying smaller deformations, such as fine cracks and surface roughness, which underline the need for dataset augmentation and enhanced training methods. This study underscores the transformative potential of AI-driven UAV systems in predictive maintenance, contributing to the digital evolution of the wind energy sector. By integrating automation, real-time data processing, and advanced analytics, the approach aligns with Smart Industry and Industry 4.0 principles, promoting operational efficiency, cost savings, and sustainability.