Passive Acoustic Measurements Using EM-APEX Profiling Floats - implementation of machine learning algorithms for acoustic environmental characterization

Abstract

Understanding the composition of ocean ambient noise is crucial to Navy operations since it is the baseline of the sonar equation." Also, conventional subsurface profiling instruments are not able to sample surface condition (wind, rain and braking waves). We propose a three-year project for integrate a mid-frequency passive acoustic recorder into the EM-APEX float, conduct field workin the Puget Sound region and study the acoustic characteristics of the environment. The combined autonomous platform will provide ocean temperature, conductivity, pressure, horizontal velocity and ambient noise profiles. The ambient noise data will be labeled with ancillary data and processed utilizing modern machine learning libraries to train predictivemodels for target-oriented classification and provide statistics of acoustic environmental characterization. The goal is to find the optimal composition of power consumption and data telemetry bandwidth for an embedded system on source classification and target detection. Statistics of ocean conditions including the sound channel depth, ambient noise level, traffic noise, and calls of marine mammals will be generated. The long-term goa"l is to estimate andpredict ASW theatre conditions in near real-time.

Document Details

Document Type
DoD Grant Award
Publication Date
Aug 20, 2019
Source ID
N000141912626

Entities

People

  • B. Barry

Organizations

  • Office of Naval Research
  • United States Navy
  • University of Washington

Tags

Readers

  • Marine Mammal Biology
  • Neural Network Machine Learning.
  • Oceanography.

Technology Areas

  • AI & ML
  • AI & ML - Neural Networks