Intelligent health monitoring of aerospace composite structures based on dynamic strain measurements

This work presents a study on an intelligent system for structural health monitoring of aerospace structures based on dynamic strain measurements, in order to identify in an exhaustive way the structural state condition. Four fiber Bragg grating (FBG) optical sensors were used for collecting strain data, representing the dynamic response of the structure and the expert system that was developed was based on the collected response data. Multi-sensor data fusion in a feature-level approach was followed. Advanced signal processing and pattern recognition techniques such as discrete wavelet transform (DWT) and support vector machines (SVM) were used in the system. For the current analysis, independent component analysis (ICA) was additionally used for the reduction of feature space. The results showed that SVMs using non-linear kernel is a powerful and promising pattern recognition scheme for damage diagnosis.The system was developed and experimentally validated on a flat stiffened composite panel, representing a section of a typical aeronautical structure. Within the frame of the present work the flat stiffened panel was manufactured using carbon fiber pre-pregs. Damage was simulated by slightly varying the mass of the panel in different zones of the structure by adding lumped masses. The analysis of operational dynamic responses was employed to identify both the damage and its position. Numerical simulation with finite element analysis (FEA) was also used as a support tool.

► Fiber Bragg gratings were successfully utilized for dynamic strain measurements.
► Support vector machines (SVMs) were successfully utilized for damage identification.
► Independent component analysis enhanced the efficiency of SVM-based classification.
► Feature-level data fusion was performed on the measurements of four FBG sensors.