TITLE

A neural network model for non invasive subsurface stratigraphic identification

AUTHOR(S)
Sullivan, John M.; Ludwig, Reinhold; Lai, Qiang
PUB. DATE
May 2000
SOURCE
AIP Conference Proceedings;2000, Vol. 509 Issue 1, p683
SOURCE TYPE
Academic Journal
DOC. TYPE
Article
ABSTRACT
Ground-Penetrating Radar (GRP) is a powerful tool to examine the stratigraphy below ground surface for remote sensing. Increasingly GPR has also found applications in microwave NDE as an interrogation tool to assess dielectric layers. Unfortunately, GPR data is characterized by a high degree of uncertainty and natural physical ambiguity. Robust decomposition routines are sparse for this application. We have developed a hierarchical set of neural network modules which split the task of layer profiling into consecutive stages. Successful GPR profiling of the subsurface stratigraphy is of key importance for many remote sensing applications including microwave NDE. Neural network modules were designed to accomplish the two main processing goals of recognizing the “subsurface pattern” followed by the identification of the depths of the subsurface layers like permafrost, groundwater table, and bedrock. We used an adaptive transform technique to transform raw GPR data into a small feature vector containing the most representative and discriminative features of the signal. This information formed the input for the neural network processing units. This strategy reduced the number of required training samples for the neural network by orders of magnitude. The entire processing system was trained using the adaptive transformed feature vector inputs and tested with real measured GPR data. The successful results of this system establishes the feasibility the feasibility of delineating subsurface layering nondestructively. © 2000 American Institute of Physics.
ACCESSION #
6029063

 

Related Articles

Share

Read the Article

Courtesy of THE LIBRARY OF VIRGINIA

Sorry, but this item is not currently available from your library.

Try another library?
Sign out of this library

Other Topics