Machine learning-based pulse characterization in figure-eight mode-locked lasers

Area of Science:

• Optics and Photonics
• Machine Learning Applications
• Laser Physics

Background:

• Accurate measurement of ultrashort laser pulse duration is crucial for advanced optical systems.
• Traditional methods for picosecond pulse characterization can be complex and expensive.
• Developing compact and cost-effective diagnostic tools remains a significant challenge.

Purpose of the Study:

• To demonstrate a novel technique for determining picosecond laser pulse durations.
• To utilize machine learning and dispersive Fourier transform for pulse duration measurement.
• To enable the use of a nanosecond photodetector for picosecond pulse analysis.

Main Methods:

• Combining machine learning algorithms with dispersive Fourier transform (DFT).
• Utilizing a fiber laser system to generate picosecond pulses (28-160 ps) with varying spectral widths (0.75-12 nm).
• Training an artificial neural network (ANN) for pulse duration prediction.

Main Results:

• Successfully determined the temporal duration of picosecond laser pulses using a nanosecond photodetector.
• Achieved a mean agreement of 95% in pulse duration prediction using the trained ANN.
• Demonstrated the feasibility of measuring pulse durations from 28 to 160 ps.

Conclusions:

• The proposed technique offers a new, accurate, and potentially low-cost method for laser pulse characterization.
• This approach paves the way for compact and affordable feedback systems in complex laser setups.
• The combination of machine learning and DFT provides a powerful tool for optical diagnostics.