Predictive correction method based on deep learning for a phase compensation system with frozen flow turbulence.

Area of Science:

• Optics and Photonics
• Artificial Intelligence
• Atmospheric Physics

Background:

• Atmospheric turbulence distorts optical beams, degrading the performance of free-space optical systems.
• Accurate phase compensation is crucial for mitigating these distortions.
• Existing methods often struggle with real-time compensation and prediction of turbulence dynamics.

Purpose of the Study:

• To develop a deep learning framework for real-time atmospheric turbulence compensation.
• To accurately predict and correct phase distortions in optical beams.
• To enhance the effectiveness of Gaussian and vortex beams through compensation.

Main Methods:

• A deep learning method combining Convolutional Neural Network (CNN) and Convolutional Long Short-Term Memory (ConvLSTM) models was employed.
• The CNN model learned to generate equivalent turbulent compensation phase screens.
• The ConvLSTM model predicted atmospheric turbulence evolution based on the Taylor frozen hypothesis to address time delays.

Main Results:

• The trained CNN model successfully generated accurate phase compensation screens for distorted Gaussian beams.
• The ConvLSTM model effectively predicted turbulence evolution, enabling more precise phase correction.
• Experimental validation demonstrated effective and accurate compensation of both distorted Gaussian and vortex beams.

Conclusions:

• The proposed deep learning method offers a robust solution for atmospheric turbulence compensation.
• The integration of CNN and ConvLSTM models addresses challenges in real-time correction and prediction.
• This approach significantly improves the quality of optical beams affected by atmospheric turbulence.