Synthetic data for damage assessment in aircraft turbines

Bachelor thesis (2021)

Authors

E.T. Klein Onstenk Electrical Engineering, Mathematics and Computer Science

Contributors

J.C. van Gemert Pattern Recognition and Bioinformatics - EEMCS (mentor)
B. Yildiz Pattern Recognition and Bioinformatics - EEMCS (graduation committee member)
Faculty
Electrical Engineering, Mathematics and Computer Science, Electrical Engineering, Mathematics and Computer Science
Copyright: © 2021 Eduard Klein Onstenk
More Info
expand_more

Published Date

28-06-2021

Language

English

Reuse Rights

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.

Faculty

Electrical Engineering, Mathematics and Computer Science

Abstract

This paper discusses possible ways to generate synthetic data and its use cases for damage assessment in aircraft turbines. Synthetic data has many advantages such as exact ground truth and scalable data sets. Using SLAM and SfM, which are 3D construction tools, 3D models can be constructed from 2D monocular borescope videos. A 3D reconstruction of parts of the engine allows us to measure the damage if it exists. But when is the synthetic data "good"? Using the methods SLAM and SfM synthetic data could possibly be evaluated by comparing it to real data. Using Blender, synthetic borescope videos are generated and the performance of SLAM and SfM on these videos is compared to the real videos. In general, there are many different use cases for synthetic data in damage assessment and there are multiple ways to generate the right data set. Evaluating synthetic data shows that synthetic data that qualitatively looks closer to real data does not perform closer when running SfM or SLAM on it.