Characterizing composites with acoustic backscattering: Combining data driven and analytical methods

Acoustic wave measurements are quick and non-invasive. They can be ideal to characterize composites, when we can accurately interpret the acoustic signals. However, for a wide range of frequencies, interpreting the signal is complicated as the field will be subjected to multiple scattering. We introduce a new data driven approach to characterize composites via acoustic backscattering. There are two major challenges in a data driven approach. First, we need a large amount of data of backscattered waves from well characterized composites. Second, is how best to use this data to characterize unknown composites. In this talk, we address both these challenges, and use simulated 2D data to demonstrate our method. Our methodology can be used to characterize many features of the medium in question, but for simplicity, we make a number of assumptions: the medium is formed from two materials; one material forms a large number of scatterers (all with approximately the same radius) embedded in a uniform background material. Characterizing this composite is now equivalent to measuring the volume fraction and radius of the scatterers. To simulate a wide range of volume fractions and scatterer sizes, we use the exact theory for acoustic scattering by identical circular cylinders.