ABSTRACT
ABSTRACT The increasing need of extending the structure lifespan for sustainability and economic reasons has motivated the rapid development of Structural Health Monitoring systems. Changes of the modal properties over time are often used as damage features, but they are also influenced by environmental and operational variables (EOVs).
EOVs can affect the monitored features in different ways, and direct modeling is usually very challenging. Different compensation methods have been proposed over the years. The use of nonlinear system identification procedures, such as kernel PCA, has been also recently proposed. However, principal components as well as residues are expressed in the feature space. The present paper presents a procedure with automatic selection of the optimal analysis parameters for the reconstruction of the influence of EOVs on modal characteristics and the computation of the corresponding residue in the input space. Application to selected case studies provided encouraging results.
