Real-Time Structural Health Monitoring of Composite Wing in a Wind Tunnel Test using PCA-Based Statistics

Structural health monitoring could prove particularly valuable in the case of composite aerostructures, where traditional inspection methods for critical components face challenges due to limited accessibility. An effective online SHM system can be achieved by employing fiber optic sensors, facilitating the integration of a 'fiber optic nervous system' into the composite structure, enabling the detection, measurement, and communication of a number of critical structural parameters. While the success of such systems relies on selecting the right parameters to be monitored and the right sensing technology, they can also greatly benefit from on-board, real-time processing of the rather large amount of streamed data that ends with some data-driven real-time and continuous assessment as to the current state of the structure: Normal/Abnormal (based, of course, upon previously selected criteria). This work covers fiber-optic-based structural health monitoring of a healthy/damaged composite airplane wing located inside a wind tunnel. Real-time health assessment of the wing is based on the Principal Component Analysis (PCA) algorithm. Two PCA-based measures are utilized as fault indicators: the Hotelling's T-squared distribution and the Q-statistic. These statistics are very sensitive to structural anomalies. A PCA model assisted by these statistics together form a decision support tool that offers dependable real-time information regarding the health of the airborne structure, including real-time indications on the structural deviations from a prescribed flight envelope. The performance of the method is shown on real-time structural data, where the test specimen consists of a composite aircraft wing located inside a wind tunnel. A PCA model was built based on a healthy wing state under a nominal loading regime, where a single Principal Component (PC) covers most of the data variance. During the damage identification test, the system processing of the data collected from 10 FBGs, successfully identified structural deviations, such as overload and initial damage.