Control of damage‐sensitive features for early failure prediction of wind turbine blades

The current study focuses on early prediction of structural failure of a composite wind turbine blade (WTB) using acoustic emission (AE) and strain measurement. The structural response of a 14.3-m blade with embedded artificial defects is investigated under fatigue loading in flapwise direction. The fatigue loading is realized in several successive portions until structural failure.

Strain and acoustic emission signals from each portion are recorded. The goal is to explore damage-sensitive features (DSFs) derived from acoustic emission and strain signals that would be suitable for early indication of blade failure under fatigue. These features include modal characteristics of strain time history, such as natural frequencies, damping ratios, and modal amplitudes.

Acoustic emission features explored in this study comprise average frequency centroids based on an amplitude and absolute energy and gradients of cumulative energy curves. Changes of these features before failure relative to the previous loading portion are calculated and compared among different sensor locations with a twofold goal—firstly, to find the features that are the most sensitive to damage accumulation and, secondly, to find a location with the largest relative changes, thus enabling damage localization.

The results show that strain and AE signals are correlated well in terms of pinpointing to a location of the largest aggregation of defects. This study gives recommendations of the most efficient feature combination of different measurements for reliable structural health monitoring of wind turbine blades.