Flexible, monolithic piezoelectric sensors for large-area structural impact monitoring via MUSIC-assisted machine learning

Abstract

Aircraft smart skin requires the integration of large-scale, ultrathin, and high-sensitive sensor network on the surface for structural health monitoring (SHM). However, it is fairly difficult to fabricate such large-area flexible sensor networks with low cost and high reliability. Although flexible and monolithic sensors are easy to be fabricated, their applications in large-area impact monitoring have yet to be developed and revealed. In this study, an impact monitoring and localization technology based on monolithic small-area sensors is proposed for large-area structures. The pivotal principle lies in the combination of the flexible piezoelectric sensor array and the Multiple Signal Classification (MUSIC)-assisted Artificial Intelligence Network (ANN) algorithm, where the key sorted feature matrix from the output signals to ANN can be captured by the MUSIC algorithm to achieve the spatial location estimation. The results show that the flexible piezoelectric sensors demonstrate excellent performance to monitor structural impacts in a region of more than 7500% of its area. Secondly, excellent conformation between the sensors and complex surfaces is achieved by the ultrathin thickness almost without affecting the surficial flow field. The overall recognition accuracy and robustness of impact location of machine learning is greatly increased by the integration of MUSIC algorithm, which is immune to the shape, material, thickness, or opening holes of the structure. Except for the impact location, impact energy, impact frequency, impact hardness and other structural health parameters can also be monitored. Therefore, a great prospect in the integrated monitoring of the aircraft’s structural and environmental parameters is expected.