Net Results

by Scott Millhouse, Civil Engineer; U.S. Army Corps of Engineers Engineering and Support Center, Huntsville, AL,
Matthew Gifford, Geological Engineer; Sanford Cohen and Associates, McLean, VA,


Serial Information: Civil Engineering—ASCE, 1997, Vol. 67, Issue 1, Pg. 58-60


Document Type: Feature article

Abstract: On any sort of cleanup, digging is expensive and intrusive, but when the cleanup involves unexploded artillery shells and other munitions, digging can also be dangerous. Understandably, one of the goals of the U.S. Army Corps of Engineers efforts to clean up buried ordnance has been to keep digging to a minimum, and then only with the best information possible. One way the Corps is making the most of its data is through neural networks, a form of artificial intelligence that is leading the way to more efficient ordnance detection and removal. The term neural nets describe a broad category of analytic tools that mimic the way humans seem to receive, process, store, and communicate data. Neural nets differ from standard database analysis methods in that they can be trained, literally, to examine complicated sets of data and draw from them useful, generalized conclusions. Currently, the Corps, with Parsons Engineering Science, Fairfax, Va., is developing a net that reads the input of non-intrusive detection devices and recognizes which signals indicate the presence of buried armaments. When combined with geographic information systems that link maps to databases, images and other site-relevant information, neural nets are a powerful tool for speeding cleanups.

Subject Headings: Neural networks | Artificial intelligence | Site investigation | Weapons | Geographic information systems

Services: Buy this book/Buy this article

 

Return to search