Predicting mode-locked fiber laser output using feed-forward neural network (2311.18385v1)

Abstract: With a great ability to solve regression problems, the artificial neural network has become a powerful tool to facilitate advancing ultrafast laser research. In this contribution, we demonstrate the capability of a feed-forward neural network (FNN) to predict the output parameters of a mode-locked fiber laser, which mutually depend on multiple intracavity parameters, with high speed and accuracy. A direct mapping between cavity parameters and laser output is realized through the FNN-trained models, bypassing tedious iterative numerical simulation as a common approach to get a converged solution for a laser cavity. We show that the laser output spectrum and temporal pulse profiles can be accurately predicted with the normalized root mean square error (NRMSE) of less than 0.032 within only a 5 ms time frame for scenarios inside and outside the training data. We investigate the influence of FNN configuration on prediction performance. Both gain and spectral filter parameters are explored to test the prediction capability of the trained FNN models at high speed. Straightforward and fast prediction of the laser output performance for varying laser intracavity parameters paves the way to intelligent short-pulsed lasers with inversed design or autonomous operation maintenance.