Sparse identification of nonlinear dynamics applied to the levitation of acoustically large objects

TL;DR

This paper applies Sparse Identification of Nonlinear Dynamical Systems (SINDy) to derive accurate equations of motion for acoustically large objects, improving understanding where traditional Gorkov formulations are inadequate.

Contribution

It introduces a novel application of SINDy to experimentally derive nonlinear dynamics of large objects under acoustic radiation forces, surpassing the limitations of existing analytical models.

Findings

SINDy accurately recovers governing equations with less than 0.05% error for small objects.

Experimental data shows external excitation amplitude significantly affects large object dynamics.

SINDy effectively models complex acoustic levitation phenomena.

Abstract

Many studies on acoustic radiation forces focus on characterizing the behavior of acoustic fields. However, the dynamic response of objects, particularly those larger than the wavelength, remains underexplored. Here we bridge this gap by deriving nonlinear equations of motion for a trapped spherical object under acoustic radiation forces and external excitation, where the Gorkov formulation fails to provide accurate results. Using Sparse Identification of Nonlinear Dynamical Systems (SINDy), we derive the corresponding nonlinear equation of motion from analytical time series data obtained through the Gorkov formulation and external excitation for acoustically small objects, and which recovers the governing equation with less than 0.05% error in coefficient values compared to the analytical solution.. We conduct experiments with the TinyLev levitator with external excitation to generate time series for acoustically large particles. Then, SINDy is applied to reconstruct governing equations from experimental data to see how external excitation amplitude influences the dynamics of acoustically large objects. Our findings demonstrate that the SINDy can effectively be used as a tool for deriving governing equations from complex data to improve and refine theoretical developments; in the present case, for acoustically large objects, where the Gorkov formulation fails to provide an accurate prediction.