Phononic Bandgap
Predictor

ML-driven prediction of phononic bandgaps in rhombic unit cells with air hole inclusions. Select geometric parameters below to visualize the unit cell, view COMSOL-computed band structures, and get ML predictions for bandgap presence and width — trained on 3,000 COMSOL simulations.

Geometric Parameters

Select parameters and click Analyze.

ML Model Info

Classification

Loading...

Regression

Loading...

Band Structure NN

Loading...

Dirac Cone Classifier

Loading...

Dataset

Loading...

3D Unit Cell Geometry

a01.000 mm
a = a0√3
r0 (hole radius)
h (thickness)

Band Structure — Dispersion Relation

ML Bandgap Prediction Map

ML Pipeline Architecture

Input ThicknessRatio, FillingFraction
Feature Engineering 16 derived features
Classifier Has bandgap? (Y/N)
Regressor Bandgap width (kHz)
Prediction Bandgap + Band Structure

COMSOL FEA Data

3,000 rhombic unit cell simulations with air hole inclusions. 61 k-points along the irreducible Brillouin zone path, 15 frequency bands per sample.

Feature Engineering

16 features from 2 params

Polynomial (x², x³), interactions (x*y, x²*y), ratios (x/y), and transforms (√x, log(1+x)) capture non-linear physics.

Geometric Parameters

a = a0√3, r0 = √(f·√3·a²/2π), h = ThicknessRatio × a. Rhombic unit cell with 2 cylindrical air inclusions.

← All ML/AI Projects PhononicNet Demo →