Combining a vibration-based SHM-scheme and an airborne sound approach for damage detection on wind turbine rotor blades

In the current work, a vibration-based SHM-scheme and an acoustic emission (AE) approach based on airborne sound are tested for damage detection at wind turbine rotor blades. The vibration-based approach includes the estimation of condition parameters (CPs), machine learning by means of data classification for changing environmental and operational conditions (EOCs) and hypothesis testing by using the acceleration signals of six measurement positions that are distributed over the blade length. A residue from the stochastic subspace identification (SSI) method and a residue from a vector autoregressive (VAR) model were used, in order to obtain two CPs. These are used as indicators for changes in the response of the structure. The airborne sound acoustic mission damage detection approach monitors the blade with three fiber optical microphones. A model of the cracking sound was developed, which describes characteristics of these sounds in the time-frequencypower domain. A detection algorithm uses these characteristics to detect damages, to estimate their significance and to handle environmental noise. Both methods were applied on data from a fatigue test of a 34 m rotor blade, which was harmonically excited for over one million load cycles in edgewise direction, leading to a significant damage at the trailing edge. Further, the potential of combining the two complementary approaches is investigated.