Our recent work with Claudio Sbarufatti, Nikos Dervilis and Keith Worden, “On a meta–learning population-based approach to damage prognosis” is now available in Mechanical Systems and Signal Processing 😀

In this work, we introduce a population-based structural–health–monitoring approach to damage prognosis. We consider a training population of structures with recorded data of damage evolution and define a population model to perform damage prognosis for new testing structures. More specifically, we follow two approaches. The first is based on meta–learning and deep kernel transfer to define the model, and the second is based on functional–PCA and normalising flows to define a functional subspace exclusively based on the acquired data. We test our methodology on an experimental dataset of six aluminium plates with a crack growing in them and the results show that the population model provides accurate predictions for the testing structures which are properly informed by the so–far recorded data.

If you are interested, you can find the paper here