Abstract
Operational modal analysis (OMA) can be considered one of the most important processes in structural health monitoring (SHM) owing to its capacity to accurately estimate modal parameters which have a physical nature and are highly correlated with damage occurrence. This paper proposes a generic automatic OMA strategy with the ability to efficiently estimate modal parameters in complex structures with high repeatability and multiple symmetries. The strategy is based on an efficient version of the covariance-driven stochastic subspace identification (SSI-COV) method, combined with pattern recognition based on clustering analysis and on silhouette validity applied sequentially in a moving windows procedure across the frequency domain under analysis. In addition, procedures for estimating the best performant dissimilarity measures and clustering methods are proposed, along with a new procedure for estimating the most accurate number of natural modes in OMA. Application of the methods to the data collected from a suspension bridge demonstrates the effectiveness and accuracy of the proposed methodology for automatic OMA and estimation of the number of natural modes. Modal assurance criterion (MAC)-based dissimilarity and k-medoids are shown to be the best set of dissimilarity measures and best clustering method.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call