The September issue of the Journal of the Acoustical Society of America featured a new publication by UBC Mechanical Engineering researchers. Postdoctoral researcher Dr. Wrik Mallik, Associate Professor Rajeev Jaiman and Assistant Professor Jasmin Jelovica are the authors of the recent publication, “Predicting transmission loss in underwater acoustics using convolutional recurrent autoencoder network.”
Their work explores using a data-driven deep-learning technique called a convolutional recurrent autoencoder network (CRAN) architecture to model underwater signal transmission loss within the complex ocean environment, in which acoustic transmission is affected by geometric spreading, refraction, and reflection from the surface and bottom. The team demonstrate that the CRAN is successful at learning these physical elements, and its ability to predict this complex acoustic behavior could be developed into real-time far-field underwater noise prediction with potential applications for marine vessel decision-making and online control.
Read “Predicting transmission loss in underwater acoustics using convolutional recurrent autoencoder network.” at DOI: https://doi.org/10.1121/10.0013894.