Experiences from Wind Turbine Pilot Test of a Remote Inspection System

This paper describes a pilot test of a remote inspection prototype system in a wind turbine. This test provided useful experience on how the system operated in this realistic location, and a preview on what kind of data it is capable of collecting. The test lasted for approximately 6 months before the system had a fatal failure in a part that has since been improved for future versions of the system. An exploratory analysis of the collected data has been performed, using principal component analysis and random forest regression, two commonly used machine learning methods. The results demonstrate that it is possible to extract some potentially useful information from the collected data and points to some possible uses of this, e.g. to trigger when to alert a human operator to investigate a set of data or when additional data should be collected.