Analysis of Soft Phononic Crystals: Using Machine Learning to Predict Compression using Transmission Data

Abstract

Within the broader field of metamaterials, the study of photonic and phononic structures has demonstrated that soft, periodic, porous, phononic materials, which manipulate acoustic waves, exhibit transmission and reflection in the frequency domain that is tunable under compression and tension. However, most of this tunability has been shown either only theoretically, or between two states, buckled and unbuckled. This work aims, through experimentation, to model the effect of compression on the transmission spectrum of these materials, with much higher compression resolution. Using different machine learning techniques, the analysis shows that the transmission behavior of these materials can be modeled using analysis of the transmission intensity of different frequencies, across a continuous compression domain. The applications of this research include, primarily, stress sensors, as well as possibly tunable acoustic filters, and possibly tunable electromagnetic sensors.