Exploratory Dissertation: Machine learning for health monitoring of guided wave tested CFRP’s: Towards integrated feature extraction and domain adaptations with deep learning
S.R.M. van Baars (TU Delft - Aerospace Engineering)
Rene C. Alderiesten – Mentor (TU Delft - Structural Integrity & Composites)
Olga Fink – Graduation committee member (ETH Zürich)
Gabriel Michau – Graduation committee member (ETH Zürich)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
The goal of this dissertation was to integrate the models developed by the Intelligent Maintenance Systems (IMS) group of ETH Zürich, with a dataset typical to the Structural Integrity and Composites (SI&C) group of the TU Delft. To achieve this objective, the open source Guided wave dataset from the NASA Prognostics Center of Excellence repository was selected.Three objectives have been studied:Integrated feature extraction. Here the goal was to use artificial intelligence (AI) to automatically extract features from raw audio signals.Predictive modelling. The goal was to use heterogeneous modelling to predict the remaining useful lifetime and the delamination growth in carbon fibre reinforced polymers (CFRP's).Domain adaptation. Here, the objective was to converge delamination predictions of one boundary condition such that a model could make predictions for a second boundary condition. The approach is novel to the SHM datasets used by the SI&C group and is first introduced in this dissertation.