The continuous monitoring of wind turbine rotor blades is of great significance. An important aspect regarding monitoring on wind energy applications is the need for data classification, due to the change of the structure’s response depending on the varying environmental and operational conditions (EOCs). Within this work, a fatigue test of a 34 m rotor blade, at the end of which a significant damage occurred, is described. The rotor blade was excited by a harmonic load in edgewise direction for over 1 million cycles, leading to the initiation of damage at the bondline of the trailing edge. A residue based on the stochastic subspace identification (SSI) approach was used for the definition of a condition parameter, in order to monitor structural changes during the experiment. At the same time, manual data classification was performed on the data by taking into account the varying testing conditions, such as the load level, which causes changes in the structural response. The derivation of the condition parameter and the manual classification are performed by means of implementing a modular SHM-Scheme that includes the steps of machine learning for data classification, the definition of condition parameters and finally hypothesis testing for the validation of damage existence.
doi: 10.12783/SHM2015/347