Single Ended In-Situ Atmospheric Turbulence Strength Characterization Using Deep Neural Networks

Date of Award

5-1-2025

Degree Name

M.S. in Electro-Optics

Department

Department of Electro-Optics and Photonics

Advisor/Chair

Mikhail Vorontsov

Abstract

This thesis explores a AI-based approach for single-ended prediction of atmospheric turbulence strength at high temporal rates using Deep Neural Network (DNN) models trained on computer simulated images of laser beam intensity scintillation patterns. The major objective of this study is to investigate whether a DNN model trained solely on intensity scintillation patterns obtained using unidirectional (single-pass) laser beam propagation can accurately predict strength of atmospheric turbulence values in the target-in-the-loop (double-pass) propagation scenarios with laser beam scattering off a retro-target or a target glint. The obtained results demonstrate ability of accurate prediction of atmospheric turbulence strength by the DNN model trained on the single pass data in double-pass operational engagements for relatively small size retro-targets under wide range of turbulence conditions.

Keywords

Artificial Intelligence, Energy, Engineering, Optics

Rights Statement

Copyright 2025, author.

Share

COinS