Prediction and parameter inversion of wavelength-tunable fiber laser based on machine learning

Area of Science:

• Photonics and Laser Technology
• Artificial Intelligence in Engineering

Background:

• Tunable mode-locked fiber lasers are essential for diverse fields like optical communications and biomedical imaging.
• Optimizing these lasers is complex due to nonlinear relationships between cavity parameters and pulse output.

Purpose of the Study:

• To develop an intelligent approach for predicting and optimizing pulse characteristics in tunable mode-locked fiber lasers.
• To overcome the challenges of complex nonlinear parameter relationships in laser design.

Main Methods:

• Combined ensemble learning (EL) with seven machine learning algorithms to model laser parameters and outputs.
• Utilized the Fata Morgana algorithm (FATA) to identify optimal laser pulse configurations.
• Validated results using split-step Fourier transform simulations.

Main Results:

• Achieved precise control over pulse duration (min 0.780 ps), peak power (max 1.751 W), and energy (max 2.032 pJ).
• Demonstrated simultaneous optimization of pulse duration and peak power at 1551 nm.
• Identified specific cavity parameters for optimized pulse performance.

Conclusions:

• The proposed EL and FATA approach enables intelligent design of high-performance tunable lasers.
• This method offers significant potential for applications in precision machining, biomedicine, and quantum communication.