Efficient Inversion in Underwater Acoustics with Analytic, Iterative, and Sequential Bayesian Methods.

摘要

The long-term goal of this project is to develop efficient inversion algorithms for successful geoacoustic parameter estimation, inversion for sound- speed in the water-column, and source localization, exploiting (fully or partially) the physics of the propagation medium. Algorithms are designed for inversion via the extraction features of the acoustic field and optimization. The potential of analytic approaches also is investigated. Our specific objectives are as follows: (1) Achieve accurate and computationally efficient inversion for propagation medium parameters and source localization by designing estimation schemes that combine acoustic field and statistical modeling, (2) Develop methods for passive localization and inversion of environmental parameters that select features of propagation that are essential to model for accurate inversion, (3) Implement Bayesian filtering methods that provide dynamic and efficient solutions for the first two objectives, and (4) Develop analytic techniques for sediment sound speed estimation. Continuing efforts from previous years, we worked with Bayesian approaches applied to sound signals for the extraction of acoustic features using a combination of physics and statistical signal processing. One of the topics approached this past year was source localization, bathymetry, and water column sound speed estimation using arrival time estimates for propagation in multipath environments with sequential Monte Carlo methods, tied with a linearization method with novel features. The initial goal is to estimate accurately the arrival times of sound paths in shallow water environments. Then, we propagate these arrival times and their posterior PDFs through a quasilinear model for source location, bathymetry, and water column sound speed profile estimation. Finally, we worked on a new sediment sound speed estimation scheme based on Stickler's inverse problem approach.