ABSTRACT
In recent years, structural health monitoring (SHM) has advanced significantly, providing insights about structures with the integration of Big Data and Machine Learning tools. Modern SHM involves extracting features to capture structural behavior, providing valuable insights. The Random Decrement Technique (RDT) is crucial in estimating free responses and frequency-features through System Identification (SID) techniques. Despite the effectiveness of RDT and SID in SHM, a comprehensive sensitivity analysis of parameters and resulting frequency-features is still a topic of research. This article analyses the parameters for RDT and Hankel alternative view of Koopman (HAVOK) approach in extracting modal parameters from a real structural bridge in Andoain, Spain. The study reveals that extending the Hankel horizon minimizes damping factor variance, and increasing averaged segments or the time duration of these segments ensures stable numerical values of the estimated modal parameters. Ultimately, the algorithm parametrization depends on the resource constraints in desired monitoring deployments.
