Efficient and scalable detection of acoustic emission multiplets for fatigue crack growth monitoring

The presence of repeating self-similar acoustic emission signals, known as multiplets, has been linked to the phenomenon of fatigue crack growth, making their detection an extremely useful way to assess damage. However, the high computational cost and memory requirements of existing multiplet detection algorithms restrict their applicability to only relatively small datasets, posing a strong limitation to the exploitation of multiplet detection to fatigue monitoring. This work proposes two complementary strategies to increase the computational efficiency of multiplet detection algorithms. It is shown that they can be easily implemented with no loss to multiplet detection accuracy. The strategies are then applied to a large fatigue dataset consisting of more than one million acoustic emission signals. Overall, the proposed strategies enable efficient and scalable multiplet detection, extending its applicability to large-scale and potentially real-time fatigue monitoring scenarios.