Due to the rapid development and commercial availability of modern computer hardware and storage, there is an increasing demand for automated, intelligent monitoring of civil infrastructures. Nowadays, inspections of structures, like wind turbines, are often done manually on a regular basis. Those visual examinations are costly and can also leave significant damages undetected for a longer period, for example when they occur right after an inspection or if the damage occurs within the structure and is therefore not visible from the outside yet.

For many common damage detection applications, such as acoustic emission testing (AET), piezoelectric sensors are mounted in specific positions on a structures surface and record signals of structure-borne sound in order to identify potential damage scenarios. In this collaborative research project, a monitoring system for automated detection of wire breaks in tendons of wind energy turbines based on the previously described AET shall be developed. For that purpose, laboratory experiments as well as field measurements in a real wind turbine are conducted to record a data base of damage signals, in particular wire breaks, and environmental noise. On the basis of this data base, the capability of modern machine learning techniques for the classification of damage-induced acoustic emission signals will be evaluated. Finally, the most promising technique will be integrated in a monitoring system that will be tested on a real wind turbine.