Damage detection in composite aerostructures from strain and telemetry data fusion by means of pattern recognition techniques

Fiber Optic Sensors (FOS) have been used for Structural Health Monitoring (SHM) in aerostructures since they offer attractive advantages over standard electric gauges such as electromagnetic immunity, embedded ability, and small size. Damage detection performed directly from data acquired by this kind of sensors (in particular Fiber Bragg Gratings or FBGs) cannot be achieved since such sensors can only measure strain and temperature and are not able to measure damage by themselves. Different alternatives have been proposed in order to develop a global, automatic damage detection technique based on strain data. One of those techniques consists of studying the correlations among different sensors in a sensor network by using pattern recognition techniques, with the aim of unveiling changes in the global stiffness of a structure promoted by damage occurrence. However, in real-world aerostructures, variations in the operations conditions may also change the strain field, affecting the performance of these techniques. In this paper, a novel methodology based on strain and telemetry data fusion is proposed as an improvement of the global damage detection technique based on strain field pattern recognition. The technique was validated by using data a from autopilot telemetry and inflight strain data of an Unmanned Aerial Vehicle (UAV) instrumented with 20 FBGs embedded in its wing structure. Different artificial damages were induced into the wing’s main beam (made of Carbon Fiber Reinforced Polymer, or CFRP) in order to test the whole methodology using different raw sensor and feature-level data fusion techniques. The results demonstrated the capability of the methodology for detecting damages during UAV operation and aim to provide solutions for a practical implementation of FOS-based SHM in real-world composite aerostructures.