A Histogram-Difference Method (HDM) for Neutron/Gamma Discrimination Using Liquid and Plastic Scintillators

Abstract

This paper introduces a novel and extremely simple -and therefore, easily implemented in a Field-Programmable Gate Array (FPGA) -method of neutron/gamma pulse discrimination called the Histogram-Difference Method (HDM). Crucially, this method relies on the use of a "reference set" of γonly pulses from 22 Na, 137 Cs and 60 Co sources and on features extracted from the pulses. The pulses from this reference set are compared to a set of an identical number of n+γ pulses from a 252 Cf source. Not only can HDM determine the presence of neutrons in the 252 Cf source, but it can also estimate the approximate percentage of neutron pulses from that source. The results reported in this paper are based on data from different radioactive sources and a standard scintillator for this type of research -namely, BC-501A. These results were confirmed using three other scintillators, one made in-house, and two others that are commercially available, EJ-299-33 and EJ-200. Pulses were encoded using Q-values, which are features extracted from the pulses, i.e., (Q total , Q tail ) pairs in IR 2 . As a control, we also encoded pulses as the Root-Mean-Square (RMS) distances from the average pulse of all known γ-only pulses and ran HDM using these pulse representations. The results are very similar to those obtained using a Q-value encoding of the pulses.

FPGA Implementation of MLP, 1D-CNN and TTTratio algorithms for Neutron/Gamma-ray Discrimination using Plastic Scintillator

“…The dataset used to train and assess the proposed MLP and 1D-CNN models are the same as those used in [12]. They were prepared at 250 MHz using EJ276 plastic scintillator, where number of neutron and gamma-ray samples are 40600 and 66800, respectively.…”

“…In our recent work we showed that a two hidden layer MLP model can outperform the T T T ratio algorithm for neutron/gamma-ray discrimination in EJ276 plastic scintillator, especially for low energy radiations ([100, 250] keV) [12]. In some applications, the classification should be achieved online, requiring the implementation of a discrimination approach on an embedded system.…”

Recent Developments in Count Rate Processing Associated with Radiation Monitoring Systems

“…All of these algorithms will be implemented in a dedicated DSP card [3] and the compliance of the RPM system will the standard ANSI42-35 will be tested in due course [31]. Figure 13.…”

“…This chapter will not address pulse processing techniques, for which details can be found in [1][2][3], but presents some recently developed techniques to process count rate signal using frequentist inference. Bayesian inference can also be implemented to process count rate as for instance for gamma spectrum unfolding or photon-limited imaging filtering [4,5].…”

Model-Based Deep Learning Algorithm for Pulse Shape Discrimination in High Event Rates

“…Various PSD techniques were proposed in the literature. Traditional digital signal processing PSD methods include charge integration [9], slice-fitting algorithms [10], histogramdifference methods [11], and discrete wavelet transforms [12], [13]. PSD methods based on traditional machine learning techniques include Gaussian mixture models [14], support vector machines [15], and k-means [16].…”