Radiation Protection Dosimetry Vol. 110 Nos. 1-4 © Oxford University Press 2004; all rights reserved
ANDI-03: a genetic algorithm tool for the analysis of activation detector data to unfold high-energy neutron spectra
Deutsches Elektronen-Synchrotron (DESY), Notkestrasse 85, D-22607 Hamburg, Germany
Corresponding author: mukherjee{at}ieee.org
The thresholds of (n,xn) reactions in various activation detectors are commonly used to unfold the neutron spectra covering a broad energy span, i.e. from thermal to several hundreds of MeV. The saturation activities of the daughter nuclides (i.e. reaction products) serve as the input data of specific spectra unfolding codes, such as SAND-II and LOUHI-83. However, most spectra unfolding codes, including the above, require an a priori (guess) spectrum to starting up the unfolding procedure of an unknown spectrum. The accuracy and exactness of the resulting spectrum primarily depends on the subjectively chosen guess spectrum. On the other hand, the Genetic Algorithm (GA)-based spectra unfolding technique ANDI-03 (Activation-detector Neutron DIfferentiation) presented in this report does not require a specific starting parameter. The GA is a robust problem-solving tool, which emulates the Darwinian Theory of Evolution prevailing in the realm of biological world and is ideally suited to optimise complex objective functions globally in a large multidimensional solution space. The activation data of the 27Al(n,
)24Na, 116In(n,
)116mIn, 12C(n,2n)11C and 209Bi(n,xn)210-xBi reactions recorded at the high-energy neutron field of the ISIS Spallation source (Rutherford Appleton Laboratory, UK) was obtained from literature and by applying the ANDI-03 GA tool, these data were used to unfold the neutron spectra. The total neutron fluence derived from the neutron spectrum unfolded using GA technique (ANDI-03) agreed within ±6.9% (at shield top level) and ±27.2% (behind a 60 cm thick concrete shield) with the same unfolded with the SAND-II code.