This paper presents an analysis of state-of-the-art of impact detection techniques for aerospace structural components as well as a study about the combination of two promising approaches for localizing an incidental impact event on a typical metallic aerospace structural component as test article. In the aeronautical scenario, some typical damaging events that may occur during service life are runway bird-strike, tool drop and debris impact. The last two cases produce generally high-frequency vibrations that are usually well predicted by ultrasonic techniques. The impacts from birds on the other hand produces vibrations in the lower or modal frequency range. The present work is focused on the possible combination of two methodologies: the first one, related to impacts inducing low-frequency vibrations, is based on the implementation of a Neural Network, while the second one, related to impacts inducing higher-frequency stress waves, is based on an acoustic source localization approach. Both numerical and experimental analyses were implemented on the same isotropic aluminum flat panel, and a possible combination of the experimental sensors arrangement will be discussed within the paper. The results have confirmed the positive performance of the neural network, opening to a more extended experimental campaign mainly oriented to the definition of the system precision, possible fault reconstruction and optimization in the data handling and reduction of computational effort. On the other hand, the main advantage of the acoustic emission formulation is that it does not require the knowledge of the wave velocity profile in the panel. Dependence of the guided wave velocity on the signal frequency for isotropic plates and, also on the wave propagation direction for anisotropic plates are the two major obstacles for acoustic source localization in a plate. Both these obstacles are avoided in this latter formulation.

本文分析了最先进的航空航天结构部件碰撞检测技术，并研究了两种有前途的方法的组合，用于定位典型金属航空航天结构部件上的偶然碰撞事件作为测试物品。在航空场景中，在使用寿命期间可能发生的一些典型破坏事件是跑道鸟击、工具掉落和碎片撞击。后两种情况通常会产生高频振动，超声波技术通常可以很好地预测这些振动。另一方面，来自鸟类的撞击会在较低或模态频率范围内产生振动。目前的工作侧重于两种方法的可能组合：第一种方法与引起低频振动的冲击有关，是基于神经网络的实现，而第二个与诱发高频应力波的冲击有关，基于声源定位方法。数值和实验分析都在同一个各向同性铝平板上进行，本文将讨论实验传感器布置的可能组合。结果证实了神经网络的积极性能，开启了更广泛的实验活动，主要面向系统精度的定义、数据处理中可能的故障重建和优化以及减少计算工作量。另一方面，声发射公式的主要优点是它不需要面板中波速分布的知识。导波速度对各向同性板的信号频率以及各向异性板的波传播方向的依赖性是板中声源定位的两个主要障碍。在后一种表述中避免了这两个障碍。