METHODS FOR DETERMINING THE TRAJECTORIES OF THE DRONE FLIGHT AND ASSESSING THE CRITICALITY OF DEFECTS DETECTED FROM IMAGES OF WIND TURBINE COMPONENTS

Abstract

This study introduces a novel method for assessing the criticality of defects detected on wind energy facility components using unmanned aerial vehicles (UAVs). It integrates automated analysis of multispectral images and fuzzy logic techniques to incorporate expert domain knowledge effectively. The method comprises three main stages: extracting physical dimensions and thermal characteristics of defects, formalizing expert-defined criticality criteria, and aggregating these parameters using fuzzy logic. Final numerical criticality scores are calculated via centroid-based defuzzification. Experimental validation conducted on defects such as blade cracks, tower corrosion, and motor overheating demonstrated high consistency between automated assessments and expert evaluations, with an average deviation of approximately 0.15. This approach significantly enhances accuracy, objectivity, and efficiency in criticality assessment, aiding proactive maintenance and operational safety management of wind turbines. Moreover, the experiments revealed specific insights into the relationship between defect characteristics and criticality scores. For instance, blade cracks exceeding certain dimensions or curvature parameters directly correlated with elevated criticality levels, requiring prompt intervention. Similarly, the presence of extensive corrosion or significant thermal anomalies was accurately captured and quantified by the method, enabling precise maintenance recommendations. In conclusion, the designed method improves the accuracy and objectivity of criticality assessments and significantly contributes to optimizing the overall maintenance strategy for wind turbines. Its practical applicability has been validated through rigorous experimental studies, proving its efficacy in diverse, real-world operational conditions.