A data-driven framework for predicting the acoustic performance of metamaterial arrays

Acoustic metamaterials can reduce noise, for example in ventilated sound barriers or Helmholtz resonator designs. Prior work suggests that using arrays of metamaterial units with varying geometry can improve broadband noise reduction, but predicting performance becomes difficult because elements interact through complex coupling. The study proposes a data-driven framework to estimate how an entire metamaterial panel performs acoustically. It does this by using existing surrogate models for each unit’s behavior and then using a limited set of high-fidelity simulation results from full arrays to train a predictor for array-level transmission loss. The stated goal is to make design iteration faster and performance estimation more accurate for complex metamaterial geometries, while avoiding the high computational and financial cost of scaling full numerical/experimental studies to large arrays.