Share Email Print
cover

Proceedings Paper

Characterization of transuranic waste using artificial intelligence techniques
Author(s): Charles A. Sparrow; Susan M. Bridges; Julia E. Hodges; Jun Chen
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

Characterization of containers of transuranic waste is accomplished by passive and active neutron interrogation. In the passive approach, neutrons from spontaneous fission of 240Pu are detected by coincidence counting. An inherent limitation of this measurement is the presence of alpha particles from 239Pu, and from other radionuclides, which induce neutron emission by way of the ((alpha) ,n) reaction. Since detection of neutrons does not necessarily imply the presence of 240Pu, the content of plutonium is not uniquely determined. In the active mode, neutron pulses induce fission events, producing neutrons. Gated counting permits some discrimination between the pulse and the fission neutrons. Several different artificial intelligence techniques for representing uncertainty have been investigated to determine which ones might yield additional insight in a manner similar to the judgment of experts. Three experts were asked to evaluate two sets of data, a training set and a test set, and to provide an estimate of their confidence in the results. A neural-genetic optimizer was employed to evolve a neural network to mimic each expert's characterization. The conclusions of this study are useful in refinement of the logic for both Bayesian and fuzzy techniques.

Paper Details

Date Published: 29 January 1999
PDF: 11 pages
Proc. SPIE 3536, Nuclear Waste Instrumentation Engineering, (29 January 1999); doi: 10.1117/12.339064
Show Author Affiliations
Charles A. Sparrow, Mississippi State Univ. (United States)
Susan M. Bridges, Mississippi State Univ. (United States)
Julia E. Hodges, Mississippi State Univ. (United States)
Jun Chen, Mississippi State Univ. (United States)


Published in SPIE Proceedings Vol. 3536:
Nuclear Waste Instrumentation Engineering
David E. Robertson, Editor(s)

© SPIE. Terms of Use
Back to Top