Expert System for Analysis of Spectra in Nuclear Metrology

Abstract

In this paper is described an expert system (ES) developed in order to enable the analysis of emission spectra, which are obtained by measurements of activities of radioactive elements, i.e., isotopes, actually cesium. In the structure of those spectra exists two parts: first on lower energies, which originates from the Compton effect, and second on higher energies, which contains the photopeak. The aforementioned ES is made to perform analysis of spectra in whole range of energies. Analysis of those spectra is very interesting because of the problem of environmental contamination by radio nuclides.

Detection of Isotopes in Urban Source Search Gamma Spectra Using Hopfield Neural Network

“…Other statistical-based methods developed for isotope identification include multiple linear regression [18], sequential deconvolution [19], Bayesian statistical inference [20], and physicsbased importance weighting [21]. In addition, several methods utilize tools from artificial intelligence such as particle swarm optimization [22], fireworks algorithm [23], expert systems [24], clustering [25], fuzzy logic [11][26], fuzzy support vector regression [27], wavelet processing [28], and fuzzy-genetic hybrid approaches [29]. Despite the large volume of research, analysis of distorted signals with random count variability in search applications remains an open challenge.…”