<p>Phononic crystals are periodic artificial composite structures that have gained significant attention for their potential to control wave propagation, offering solutions for vibration and noise reduction. This study proposes an innovative optimal design method for the two-dimensional layered phononic crystal structure, combining light gradient-boosting machine (LightGBM) and an improved genetic algorithm. Firstly, LightGBM is employed to predict the band gap of the phononic crystal, treating the structural arrangement vectors as categorical features. Hyperparameter tuning is performed using simulated annealing, achieving a prediction accuracy with an error of less than 2% and requiring only 1/3134 of the computational time compared to traditional finite element methods. Secondly, a genetic algorithm with an elite retention strategy is integrated into the design method. Taking the environmental vibration caused by subway operation as a case study, the fitness function is optimized based on the bandgap width. A dataset is generated with the corresponding structural bandgap covering the 30 ~ 40 Hz frequency range. Finally, finite element analysis is conducted on the optimized structure, confirming that the bandgap characteristics are consistent with expectations. This approach offers an efficient and intelligent solution for the optimization of multiple phononic crystal designs.</p>