Singular value decomposition and information entropy theory for designing Bonner spheres spectrometers

The design theory for Bonner spheres spectrometers was proposed based on the singular value decomposition (SVD) and information entropy theory. The response functions of Bonner spheres were calculated by modeling software and Monte Carlo codes, and then, according to the IAEA neutron spectra, the Bonner spheres data matrix was calculated. The SVD method was used to classify the Bonner spheres, and then the rationality of the design was verified. The predictive model employing artificial neural network (ANN) was built based on the Bonner spheres registers and corresponding neutron spectra. Based on the built ANN model, the information entropy sensitivities of Bonner spheres were analyzed, and a suggestion on optimizing the design of Bonner spheres spectrometer was offered. The results showed that this methodology was effective in designing Bonner spheres spectrometers based on qualitative and quantitative assessment.