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

Tunable mode-locked fiber lasers are critical for applications in optical communications, fiber sensing, spectral analysis, and biomedical imaging. However, optimizing cavity parameters to achieve personalized output pulses remains challenging due to complex nonlinear relationships between parameters and pulse characteristics. Here, we propose an approach combining ensemble learning (EL) and the Fata Morgana algorithm (FATA) to predict and optimize pulse characteristics. EL integrates seven machine learning algorithms to model cavity parameters and pulse outputs, while FATA identifies optimal pulse configurations. Our results demonstrate precise control over pulse duration (minimum 0.780 ps), peak power (maximum 1.751 W), and energy (maximum 2.032 pJ), validated via split-step Fourier transform simulations. Notably, at 1551 nm, simultaneous optimization of pulse duration and peak power was achieved with specific cavity parameters. This study advances the intelligent design of high-performance tunable lasers, offering significant potential for precision machining, biomedicine, and quantum communication.