基于LightGBM 和遗传算法的二维层状声子 晶体结构设计.

Phononic crystals are periodically artificial composite structures that have garnered widespread attention for their bandgap properties and potential applications in wave propagation control, offering novel solutions for engineering vibration and noise reduction. The design of these materials is a current research focus. This study takes two-dimensional layered phonon crystal structure as an example, presents an innovative method for designing two-dimensional layered phononic crystal structures, utilizing LightGBM and an improved genetic algorithm. Initially, the LightGBM algorithm was employed for bandgap prediction of the phononic crystals, with the structural arrangement vectors subjected to categorical feature processing, and hyperparameter tuning conducted through the simulated annealing algorithm, achieving a prediction accuracy with an overall error of no more than 2% within 1/3134 of the traditional finite element method computation time. Subsequently, a design method integrating a genetic algorithm based on an elite preservation strategy was introduced. Using the environmental vibration issue caused by subway operation as an example, a fitness function optimized for bandgap width was established, resulting in a population with structural bandgaps covering the 30 ~40 Hz frequency range. Finally, finite element analysis of one structure verified its bandgap characteristics align closely with expectations. This research provides a new solution for the efficient optimization design of multiple phononic crystal structure schemes. [ABSTRACT FROM AUTHOR]