Simulation of Improved Genetic Algorithm for Mode-locked Control

Optical frequency comb technology serves as a powerful tool driving innovative advancements in vacuum metrology. Passively mode-locked fiber optical frequency combs, a primary method for generating femtosecond frequency combs, have gained prominence due to their simplicity in design and ultrashort pulse output characteristics by utilizing the nonlinear polarization rotation （NPR） mechanism for mode-locking. However, due to the high nonlinearity of NPR fiber lasers, traditional control methods struggle to stabilize mode-locking. Additionally, the system is susceptible to environmental disturbances, and mode-locking failure often occurs after initial locking, significantly limiting its practical applications. The mode-locking principles of NPR fiber lasers are briefly introduced. A simulation model of the mode-locked laser is established using coupled nonlinear Schrödinger equations （CNLSE）, and MATLAB software is employed to simulate the evolution of intracavity electric fields from initial white noise in the 0–0.1 normalized amplitude range to Gaussian pulses with peak amplitudes around 5. The characteristics of fundamental mode-locked pulses are analyzed and summarized. Building on this model, a genetic control algorithm is implemented to simulate the automatic mode locking process. The effectiveness of the genetic algorithm is validated by comparing results under different objective functions. Based on the fundamental mode-locked pulse characteristics revealed in simulations, the objective function is optimized. Through parameter traversal, the distribution of mode-locking points in the domain is illustrated, and the proportion of the mode-locking region in the entire domain is evaluated via random sampling to refine genetic algorithm parameters.This methodology holds significant potential for advancing the engineering applications of NPR-based fiber lasers.