Acoustic Emission Source Location Using Bayesian Optimisation for a Composite Helicopter Blade

Abstract

It has been demonstrated that acoustic-emission (AE) based inspection of structures can offer advantages over other types of monitoring techniques in the detection of damage; namely, an increased sensitivity to damage, as well as an ability to localise its source. These advantages have resulted in the use of AE-based systems in Structural Health Monitoring (SHM) applications becoming an increasingly popular research topic. The strong correlation between features of AE-waves and the sources that generate them can be exploited to reveal information of the health-state of a structure in a non-invasive framework. This idea has been demonstrated successfully in the early detection of crackgrowth, nucleation, corrosion, dislocation slips, and cavitation, among others. There are, however, numerous challenges associated with the analysis of AE-data. Some of these challenges include sensor placement, calibration, operational variations, sources of noise, and manipulation of large-size datasets. These challenges must be addressed carefully if one wishes to implement an AE-based system in SHM applications. Although the analysis of AE-data can be complex, its advantages can be meaningful in the development of an optimised maintenance schedule, reducing costs associated with unnecessary repairs, or by preventing potentially catastrophic outcomes from overlooking an emerging defect that can be detrimental to the safe operation of the structure. In this paper, AE data recorded from a full-scale helicopter blade are processed to develop a probabilistic model designed to predict the location of potential sources of damage. The following work addresses some of the challenges related to the construction of ΔT maps and proposes a novel strategy for identifying optimal sampling points, eliminating the need for extensive data collection for training.