Crack damage identification of a thick composite sandwich structure based on Gaussian Processes classification

Structural damage of composite materials used in aeronautics and aerospace has attracted increasing attention. Efficient and reliable Structural Health Monitoring (SHM) systems are required to provide a probabilistic interpretation of diagnostics. In this study, crack damage identification of a thick composite sandwich structure based on Gaussian Processes (GP) classification is reported by numerical simulations. The goal of the study is to obtain a data-driven probabilistic interpretation of damage detection. The investigation is carried out based on healthy and damaged status of a sandwich panel with a honeycomb core modelled in ANSYS. Instantaneous signals with different frequencies are applied to the structure and finite element analysis is performed to obtain vibration responses in both statuses. Features extracted by Discrete Wavelet Transform (DWT) are used to train and test the GP model to assess the health status of the structure. Impacts of mother wavelet in DWT, likelihood function and inference method, as well as iteration numbers are investigated on the classification accuracy. The pertinence of sensors located at different positions is also investigated. This proposed method is effective for crack-type damage detection in the studied composite sandwich structure. It is expected to be suitable for damage detection of more complex structures.

Automated summary

The study focuses on crack damage identification of thick composite sandwich structures using Gaussian Processes classification based on numerical simulations. Data-driven probabilistic interpretation of damage detection is the main goal. Features extracted by Discrete Wavelet Transform are used to train and test the GP model for assessing the health status of the structure.