Archives and Documentation Center
Digital Archives

Noise robust real-time focus detection with deep learning for ultra-fast laser micromachining

Show simple item record

dc.contributor Graduate Program in Physics.
dc.contributor.advisor Elahi, Parviz. Polat, Can. 2023-10-15T08:03:07Z 2023-10-15T08:03:07Z 2022
dc.identifier.other PHYS 2022 P66
dc.description.abstract In this thesis, different types of machine learning models are provided for ultra-fast laser micromachining system to actively control the focusing of light on the processing material by detecting the reflected light. These different types of models are tested for both experimental and theoretical approaches. For the experimental approach, four different machine learning models are explored. These models were tested for mirror, silicon, steel, and copper samples. The proposed machine learning models offer real-time control with over 90\% accuracy. For the simulation, noise at the material surface and the detection system are considered. The noise simulation, including the laser micromachining system, is done using Fourier optics and signal processing. Noise levels at the material surface are determined by laser scanning microscope measurements of experimental samples, and the commercial detection camera noises are considered for the detection noise. Convolutional neural network models are used for focus control in the simulation. Depending on the noise level, the proposed model achieves above 95\% accuracy.
dc.publisher Thesis (M.S.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 2022.
dc.subject.lcsh Lasers -- Industrial applications.
dc.subject.lcsh Deep learning (Machine learning)
dc.title Noise robust real-time focus detection with deep learning for ultra-fast laser micromachining
dc.format.pages xiv, 96 leaves

Files in this item

This item appears in the following Collection(s)

Show simple item record

Search Digital Archive


My Account