Acoustic emission monitoring and automated characterization of low-velocity impacts on composite components

The low-velocity impacts caused by debris and hail can be one of the main threats to the integrity of the composite components of aircraft such as elevators. Recent improvements in sensor technology have enabled acoustic emission sensors to be applied to composite impact monitoring. However, due to power, weight, and environmental limitations during flight, the challenge lies in minimizing the number of sensors for real-time impact monitoring while still accurately identifying impact energy and location. This paper develops a comprehensive impact monitoring framework using a single acoustic emission sensor. In this framework, a deep residual convolutional neural network is used to identify the impact energy, achieving an accuracy of 98.6 %. In addition, a heterogeneous ensemble network is proposed to localize impacts at light, moderate, and severe energy levels. The proposed framework is validated by an impact experiment performed on an aircraft elevator. The localization accuracies for light, moderate, and severe impacts are 95.0 %, 98.3 %, and 99.2 % respectively. The results demonstrate the effectiveness and potential of the proposed framework.