[3A3] Machine learning to enhance the performance of guided wave SHM
A Croxford
University of Bristol, UK
Guided wave structural health monitoring (SHM) has great potential for the permanent monitoring of structures. The potential to subtract signals recorded in a healthy and damaged state to reveal changes in the structure is attractive and has received significant research effort. However, the presence of environmental changes means that such an approach is often impractical, either preventing damage detection or requiring impractical quantities of data for the required performance levels. This paper will explore the application of a machine learning-based technique to detect changes in signals. The described approach offers excellent performance from reduced levels of data and addresses several of the core challenges for guided wave SHM. This paper will demonstrate the performance on simulated and experimental data illustrating this good performance.
