The wind energy sector has, in recent decades, been expanding by multiple orders of magnitude due to a global desire of replacing finite energy sources with renewable and clean alternatives. The maturity of wind technology, its stable infrastructure and the cost-effectiveness are factors that motivate engineers to research and develop innovative solutions leading to overall improvements. While the volume of installations, power capacities, physical sizes and offshore placements are all increasing, it is obvious, that the ability to ascertain and ensure the structural integrity of wind turbines is becoming more challenging and more valuable. A demand for a more sophistical approach, with higher capability as well as reliability than conventional visual inspection, is clarified. This project focuses on a research to facilitate the development of a remote structural health monitoring system for wind turbines, of which, particularly, wind turbine blades are evaluated to be the preferred candidate of interest.

One major issue, prohibiting the development and implementation of a health monitoring system for wind turbine blades, is the confounding influences of
environmental effects upon the sensitivity to identify the occurrence of damage. A reliable SHM system must be able to distinguish between
changes caused by ambient variations, such as temperature fluctuations, and those caused by damage. This problem of data normalization, which can be ascertained by singling out a damage-sensitive feature from latent
environmental influences, is the main focus point to be addressed in thisparticular thesis context. The employed approach, to remove environmental effects in SHM data, is based on a technique originating from the field of econometrics, namely cointegration. Non-stationary time series with common trends are considered as cointegrated if a linear combination of the series exists to be stationary. This residual linear combination will be purged from all common trends, thus, from an engineering perspective, this technique can be used to analyze whether or not, the environmental effects will be removed in the process of cointegration.

An experimental campaign of a full-scale operational Vestas V27 wind turbine has provided a substantial amount of empirical data. Over approximately a period of three months, this turbine was monitored and subjected to five different structural states including artificially introduced damages. Simultaneously, meteorological data from a nearby weather mast were collected. Hence, this campaign and its results provide the opportunity for the author of this thesis, to perform analyses and validate the inherent algorithms. The damage detection technique, which will be used to validate the robustness of the developed feature against environmental variations, is established by means of outlier analysis based on Mahalanobis metric.