Identification of Distorted Gamma-Ray Signature Patterns in Nuclear Security Applications Using an Autoassociative Memory Implementing Hopfield Artificial Neural Network

Date

2022

Authors

Valdez, Luis

Journal Title

Journal ISSN

Volume Title

Publisher

Abstract

Identification of the characteristic gamma ray radiation spectra of radioactive isotopes, i.e., signatures, that constitute a threat to society has been the focal point of nuclear security in the last decades. Alongside, the field of artificial intelligence with its advancements in the same period has instigated the development of new data analysis methods for isotopic signature identification. In nuclear security, the analysis methods focus on identifying signatures under uncertainty, that stems from the distortion resulted from intentional shielding and/or background radiation. In this paper, a new intelligent method for identifying and extracting distorted signatures from radiation measurements is presented and tested on a set of gamma ray spectra. This work proposes a method that combines image processing techniques with a Hopfield Artificial Neural Network (HANN) to identify the underlying signature pattern that has been distorted. HANN implements an auto-associative memory and can match the distorted input spectrum to one of its memorized patterns. The Hopfield Network proposed here uses the Hebbian learning algorithm to train the network parameters and memorize a set of spectrum signature patterns. The input to the HANN come in the form of a set of values extracted from the image of the measured spectrum, while its output designates the matched pattern. Results are obtained on a set of real-world gamma ray spectra that have been distorted either by extensive background counts or by random noise. Results demonstrate the high potential of the proposed method of identifying the correct signature with high accuracy (above 80%).

Description

This item is available only to currently enrolled UTSA students, faculty or staff. To download, navigate to Log In in the top right-hand corner of this screen, then select Log in with my UTSA ID.

Keywords

Hopfield Neural Networks, Neural Networks, Nuclear Identification, Nuclear Technology

Citation

Department

Electrical and Computer Engineering