Automated Operational Modal Analysis of a Helicopter Blade with a Density-Based Cluster Algorithm

Abstract

Automated operational modal analysis (AOMA) is a common standard for unsupervised, data-driven, and output-only system identification, utilizing ambient vibrations as an environmental input source. However, conventional AOMA approaches apply the -means clustering algorithm (with ) to discern possibly physical and certainly mathematical modes. That is not totally appropriate due to the intrinsic tendency of -means to produce similarly sized clusters, as well as its limitation to approximately normally distributed variables. Hence, a novel approach, based on the density-based clustering algorithm Density-Based Spatial Clustering of Applications with Noise (DBSCAN), is introduced here. Among other technical advantages, this enables to automatically detect and remove outliers. A data-driven strategy for the DBSCAN parameter selection is proposed as well, to make the whole procedure fully automated. This methodology is then validated on a case of aeronautical interest, an Airbus Helicopter H135 bearingless main rotor blade, and compared to more classic strategies for the same case study.