Interpreting environmental variability from damage sensitive features

Mitigation of Environmental and Operational Variabilities (EOVs) remains one
of the main challenges to adopt Structural Health Monitoring (SHM) technologies. Its implementation in wind turbines is one of the most challenging due to the adverse weather and operating conditions these structures have to face. This work proposes an EOV mitigation procedure based on Principal Component Analysis (PCA) which uses EOV-Sensitive Principal Components (PCs) as a surrogate of EOVs, which may be hard to measure or correctly quantify in real-life structures. EOV-Sensitive PCs are conventionally disregarded in an attempt to mitigate the effect of environmental variability. Instead, we postulate to use these
variables as predictors in non-linear regression models, similar to how Environmental and Operational Parameters (EOPs) are used in explicit EOV mitigation procedures. The work results are validated under an experimental dataset of a small-scale wind turbine blade with various cracks artificially introduced. Temperature conditions are varied using a climate chamber. The
proposed method outperforms the conventional-PCA based approach, implying that directly disregarding Sensitive-EOV PCs is detrimental in the decision-making within a SHM methodology. In addition, the proposed method achieves similar results to an equivalent explicit procedure, suggesting that EOV-Sensitive PCs can replace directly measured EOVs.