| KEYWORD |
Shape-sensing based on strain-data and video-data
Thesis abroad
keywords EXPERIMENTAL TESTS, FINITE ELEMENT METHOD, SHAPE-SENSING, STRAINS
Reference persons MARCO GHERLONE, CECILIA SURACE
External reference persons Marco Esposito, Rinto Roy
Research Groups 02-AESDO
Thesis type NUMERICAL AND EXPERIMENTAL
Description Premise:
Shape-sensing is the reconstruction of the deformed configuration of a structure starting from some discrete measurements
Technical Problem:
Video data can provide dense 2D displacement measurements, but these measurements are noisy (uncertain) and sensitive to algorithmic hyperparameters. Sensor installations (ie. strain gauges) have lower noise, but require dense installations to sufficiently quantify displacements
Research Question:
How can we combine image (video) based displacement measurements with traditional sensor installations to better quantify strain and displacement fields?
Proposed Technical Approach:
• Evaluate two forms of data fusion to combine these sensor measurements
- Gaussian process interpolants: kriging, Bayesian filtering
- Machine learning: shallow and deep neural networks
• Two rounds of experiments
- Simple structures with well constrained 1D load-displacement relationships
- More complex structures with 2D load-displacement relationships
See also gmu-polito.pdf https://www.dimeas.polito.it/en/research/research_groups/aesdo_aircraft_and_engine_structural_design_and_optimization
Required skills Structural analysis, programming (MATLAB), FEM commercial codes
Deadline 12/04/2023
PROPONI LA TUA CANDIDATURA