An Approach on How to Determine Key Performance Indicators for Guided Wave Based SHM Systems Based on Numerical Simulation

Abstract

Realization of an efficient SHM system for an arbitrarily shaped damage tolerant structure requires a clear understanding of what needs to be monitored. This may be determined through key performance indicators (KPI) being characteristics with respect to a tolerable damage to be monitored. Those characteristics can vary due to different reasons such as a structure’s geometry, the shape, size or type of the damage, the type of material and possibly other factors as well. To obtain an appropriate understanding of what a physical principle such as mechanical and hence guided waves does when travelling through a structure, numerical simulation can be of invaluable help. Starting from such a simulation the time domain signals to be monitored in practice can be generated that will subsequently be processed in a sequence of steps ending up in an artificial neural network (ANN) such that KPIs are derived that will then be used to realize an appropriate SHM system in practice. The sequence developed will be demonstrated along two examples, a first one being a plate with a notch grove through which guided waves are sent and the resulting signals are processed considering different algorithms looking at the time domain signal first and then moving onwards to difference of signals, Hilbert transform, oblique polarization filtering, and statistical methods like Auto-Covariance Function (ACF), Linear correlation coefficient, root mean square deviation and mean absolute error. Features of those algorithms are then used to train an ANN, which is finally due to provide the KPIs for signals recorded. The procedure is then applied to a more complex component being a riveted patched repair where it will be shown if the features determined from the simple plate with the notch grove can be simply transferred to the patch repair or what additional simulation work has to be done such that the KPIs can be identified accordingly. Based on these simulations an SHM system is then realized used for validation.