This paper reports on a novel metamodel for impact detection, localization and characterization of complex composite structures based on Convolutional Neural Networks (CNN) and passive sensing. Methods to generate appropriate input datasets and network architectures for impact localization and characterization were proposed, investigated and optimized. The ultrasonic waves generated by external impact events and recorded by piezoelectric sensors are transferred to 2D images which are used for impact detection and characterization. The accuracy of the detection was tested on a composite fuselage panel which was shown to be over 94%. In addition, the scalability of this metamodelling technique has been investigated by training the CNN metamodels with the data from part of the stiffened panel and testing the performance on other sections with similar geometry. Impacts were detected with an accuracy of over 95%. Impact energy levels were also successfully categorized while trained at coupon level and applied to sub-components with greater complexity. These results validated the applicability of the proposed CNN-based metamodel to real-life application such as composite aircraft parts.

In this work, a new methodology is presented for reconstruction of the impact force history using artificial neural networks (ANNs) and spectral components of sensor data recorded by piezoceramic sensors. A large set of data, required for training the ANNs, was generated by using an efficient nonlinear finite element (FE) model of a sensorised composite stiffened panel. Impact experiments were performed on a composite plate equipped with surface-mounted piezoceramic sensors to validate the numerical modelling approach. Using the FE model of the panel, data were generated for impacts that are likely to occur during the life-time of an aircraft, consisting of large mass (e.g. dropping tool) and small mass (e.g. debris) impacts at various locations, i.e. in the bay, on the foot of a stringer and over/under a stringer. Even though the panel undergoes large deformation during impact (nonlinear response), the established networks predict the impact force history and its peak with reasonable accuracy.

In this paper, a low-power high-response wireless structural health monitoring system (WSHMS) is designed, implemented, and experimentally evaluated for impact detection in composite airframes. Due to the rare, random, and transitory nature of impacts, an event-triggered mechanism is adopted for allowing the system to exhibit low power consumption when no impact occurs and high performance when triggered. System responsiveness, robustness, and energy efficiency are considered and modeled. Based on system requirements and functions, several modules are designed, including filtering, impact detecting, local processing, and wireless communicating modules. The system was implemented on a printed circuit board. The response time is about 12 μs with an average current lower than 1 mA when the impact activity is lower than 0.1%. The system exhibits high robustness to ambient vibration noises and is also capable of accurately and responsively capturing multiple sensing input channels (up to 24 channels). This paper presents a low-latency energy-aware WSHMS for impact detection of composite structures. It can be adapted to monitor of other rare, random, and ephemeral events in many Internet of Things applications.