ORION-AE: Multisensor acoustic emission datasets reflecting supervised untightening of bolts in a jointed vibrating structure

Monitoring loosening in jointed structures during operation is challenging because contact and friction, in bolted joints, induce a nonlinear stochastic behavior. The purpose of work is to provide datasets from acoustic emission (AE) sensors obtained during harmonic vibration tests. These datasets have been used to validate the development of a new clustering method for time-series, based on an original loss function that includes parameters on clusters onset, kinetics and accumulation. The datasets are originating from a jointed structure called ORION, which is an original test-rig designed for vibration study. It integrates several sensors that provide the vibration velocity from a laser vibrometer and the acoustic signals from three different AE sensors. A design of experiment allowed to acquire data streams in different operating conditions by varying the tightening configurations. These datasets can be used to challenge machine learning and signal processing methods.