Unsupervised and supervised machine learning techniques for slamming classification from SHM data

In this paper, the problem of identifying slamming impacts for monohulls and catamarans is addressed by using Machine Learning techniques. To highlight differences and similarities, two test cases, separately dealt with in previous work [1]-[2], are here considered: (i) the classification of slamming events based on clustering analysis from data collected on board a fast catamaran during full-scale trials, and (ii) the classification of different types of slams on a fast monohull from data collected from an experimental campaign carried out in the towing tank on an elastic segmented model. The analysis shows the generality and versatility of the ML approach to slamming identification, as well as its robustness in terms of accuracy