Guided Wave based Damage Characterisation and Localization in Composite Wind Turbine Blades

Abstract

Wind energy, as a major contributor to the renewable energy sector, is receiving increasing attention to meet the growing demand for low-carbon electricity. The performance of a wind turbine can be significantly affected by various types of damage in its components, particularly the rotor blades. If left undetected, damage can develop rapidly under harsh operating conditions, potentially leading to severe failure, reduced power output, and considerable economic losses. However, the anisotropic properties and curved geometry of composite wind turbine blades make conventional time-domain methods less effective for reliable damage identification and localisation. To address this challenge, a methodology is developed that integrates deep learning with ultrasonic guided wave data analysis to enable accurate detection and classification of blade damage. An extensive numerical investigation is conducted, considering a range of damage locations and sizes across the blades. The samples are excited using a five-cycle Hanning-modulated sinusoidal pulse, and the structural responses at selected sensing positions are captured and converted into time-frequency representations to reveal more damage-sensitive features. The image-like data are then analysed using a custom-designed deep learning model to categorise and localise different damage cases. The results demonstrate a high level of accuracy in detecting both the location and severity of damage, indicating the strong potential of this approach for large-scale application in smart structural health monitoring of wind turbines.