|
چکیده
|
The purpose of the present paper is to compare the various computer codes used in fast-neutron spectroscopy for spectrum unfolding and calculation of the response of the organic scintillation detectors. For neutron spectrum unfolding, the Fredholm integral equation is solved using iterative and soft computing algorithms. In the present paper, the Modified Least SQuaRe (MLSQR) method (iterative algorithm) and some soft computing algorithms are presented. The latter comprise Artificial Neural Network (ANN), Support Vector Machine (SVM) and Adaptive Group of Ink Drop Spread (AGIDS) methods. Furthermore, a MCNPX based algorithm for calculating the neutron response of organic scintillation detectors is presented which is used for Time-Of-Flight based fast-neutron spectroscopy (nTOF) . To this end, the ability for simulating the emission and tracking of two simultaneously emitted particles as well as the light production and light transport in the scintillator were added to the MCNPX computer code. Also, a post-processing software was developed to analyze the massive amounts of data in the output of the PTRAC card. The presented techniques are benchmarked with a published spectrum from an 241Am-9Be neutron source. The results obtained from the calculation using the proposed methods have an acceptable agreement with the standard ISO-8592 spectrum of the 241Am-9Be neutron source. However, the accuracy of the calculation using the SVM and AGIDS is better than the accuracy of the other presented methods.
|