Deep learning for simultaneous phase and amplitude identification in coherent beam combination.

Area of Science:

• Optics and Photonics
• Laser Physics
• Artificial Intelligence

Background:

• Coherent beam combination (CBC) enhances fibre laser power but requires precise phase control.
• Traditional phase retrieval algorithms struggle with power fluctuations common in operational fibre lasers.
• Existing methods often assume stable power levels, limiting practical CBC applications.

Purpose of the Study:

• To develop a robust phase retrieval method for CBC systems that accommodates power fluctuations.
• To enable simultaneous identification of phase and amplitude directly from combined beam intensity.
• To assess the performance and scalability of the proposed deep learning approach.

Main Methods:

• A deep learning algorithm was designed for single-step phase and amplitude retrieval.
• The algorithm analyzes intensity patterns from a single camera observation of the combined beam.
• Simulations using a spatial light modulator investigated power fluctuation effects and scalability with increasing beamlets.

Main Results:

• The deep learning algorithm accurately disentangles phase and power, even with significant power fluctuations.
• Systematic investigation confirmed the impact of power-level fluctuations on phase retrieval accuracy.
• Scalability analysis demonstrated the approach's ability to maintain precision with more beamlets.

Conclusions:

• Deep learning offers a robust solution for phase and amplitude retrieval in CBC systems facing power degradation.
• The proposed method overcomes limitations of traditional algorithms under non-ideal power conditions.
• This approach enhances the practicality and scalability of high-power fibre laser systems.