Seafloor characterization from free fall penetrometers using machine learning

Abstract

Rapid seafloor surface sediment characterization is important for many naval applications including navigation in areas of active sediment dynamics, mine burial prediction, sensor placements, unexploded ordnance detection and classification, to name just a few. Free fall penetrometers (FFP) offer a means to rapid seafloor sediment characterization from any vessel of opportunity and in a wide range of environmental conditions. FFP seabed profiling has found to be reliable and accurate, and methods are available to derive seabed stratification including layer thicknesses, to classify sediment type, and to estimate geotechnical properties such as undrained shear strength, friction angles, and relative density. However, those data analysis methods can be complex and currently require expert users. The proposed work focuses on using a large existing FFP deployment and sediment information database to develop a machine learning model to facilitate FFP data analysis with high accuracy but without need for expert users. The research tasks include: the preparation of the database, expansion of a current numerical model simulating FFP deployments for sensitivity analysis and investigation of physical processes leading to the seabed specific FFP profiles, development of a machine learning model for FFP data analysis, and finally, assessment of the accuracy of FFP results in comparison with seabed coring. Integration of students into naval research is a core objective of this proposal. At least two graduate students and two undergraduate students will be actively involved in the research. The team includes leading experts in FFP development and data analysis, numerical simulations and probabilistic analysis in geotechnical engineering, and physics-informed machine learning. The proposed work is expected to pave the way for more user friendly, reliable, and accurate data analysis of FFP deployments in the framework of rapid seabed surface investigation for naval applications, and to introduce four students to naval research

Document Details

Document Type
DoD Grant Award
Publication Date
Apr 07, 2025
Source ID
N001742210018

Entities

People

  • Nina Stark

Organizations

  • United States Navy
  • Virginia Tech

Tags

Readers

  • Acoustical Oceanography.
  • Geotechnical Engineering.
  • Research Science/Academic Research

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference
  • AI & ML - Neural Networks