On the data-driven damage detection of offshore structures using statistical and clustering techniques under various uncertainty sources: An experimental study

Damage detection procedure of offshore structures based on data-driven techniques is of paramount importance to ensure their safety and integrity, especially in real-world applications. This procedure become more challenging in case of long-term structural health monitoring of offshore jacket platforms, which are inevitably subjected to the marine environmental conditions and various uncertainty sources the leads to variability in their dynamic characteristics. Due to the importance of the mentioned challenge and enhance damage detectability in these structures, the main aim of this article is to propose an improved density-based clustering method for data-driven damage detection with the aid of a distance scaling technique. The major contribution of the proposed method is to deal with the problem of finding clusters among large datasets with varied densities based on a multi-dimensional scaling technique. This method not only suppresses the effect of uncertainty sources always available in offshore structures but also enhances the performance of data-driven damage detection methodology. The feasibility and reliability of the method presented in this study is exemplified by application in a laboratory jacket-type offshore platform under different damage scenarios along with several comparative studies. Results are demonstrated to be effective and successful in detecting early damage of the offshore structure in the presence of various uncertainty sources.