The application of Deep Learning methods for breakthrough noise and vibration control using metamaterials

Project Team

Principal Investigator:

Dr. John Kennedy

Graduate Student:

Oluwaseyi Ogun

Overview

Environmental sustainability demands innovative approaches to noise and vibration control - especially as urban soundscapes evolve with new technologies. Acoustic metamaterials (AMs), with their ability to manipulate sound in ways not possible with conventional materials, offer a powerful solution. However, their complex design space exceeds the capabilities of traditional human-led methods.

This research project leverages deep learning to enable the rapid, low-cost design of advanced AMs. High-fidelity numerical simulations generate training datasets that capture the topological features needed to produce the desired system response. Once trained, these networks can generate optimised metamaterial designs without the need for repeated modelling or simulation.

Prototypes are fabricated using the groups additive manufacturing facilities and experimentally tested to validate the acoustic performance. Any manufacturing-induced deviations are fed back into the network during a second training phase, enabling practical, high-performance designs optimized for real-world production.

Research Publications:

1. Ogun O, Rice H, Kennedy J. Bridging model and experiment in the design and validation of a sub-wavelength acoustic metamaterial. Journal of Applied Physics. 2025 May 21;137(19).
2. Ogun O, Kennedy J. Comparison of traditional and deep learning optimisation for the design of acoustic metamaterials. InINTER-NOISE and NOISE-CON Congress and Conference Proceedings 2024 Oct 4 (Vol. 270, No. 10, pp. 1352-1362). Institute of Noise Control Engineering.
3. Muhammad, Kennedy J, Ogun O. Design and fabrication of 3D-printed composite metastructure with subwavelength and ultrawide bandgaps. New Journal of Physics. 2023 May 18;25(5):053015.

Funding Agency

Provost’s PhD Project Awards