Neutron spectrum unfolding using three artificial intelligence optimization methods.

Area of Science:

• Nuclear Physics
• Computational Physics

Background:

• Neutron spectrum unfolding is crucial for nuclear applications.
• Existing methods require significant computational resources and time.

Purpose of the Study:

• To compare three artificial intelligence optimization methods for neutron spectrum unfolding.
• To evaluate the performance of genetic algorithms, radial basis function neural networks, and generalized regression neural networks.

Main Methods:

• Developed and applied genetic algorithms, radial basis function neural networks, and generalized regression neural networks.
• Unfolded 63 neutron spectra using identical response functions.
• Assessed performance using mean squared error, spectral quality, and sphere reading quality indexes.

Main Results:

• All three methods produced acceptable neutron spectra, even without initial guess spectra.
• Generalized regression neural networks demonstrated superior speed and accuracy.
• Generalized regression neural networks exhibited the most powerful generalization ability.

Conclusions:

• Artificial intelligence methods offer viable alternatives for neutron spectrum unfolding.
• Generalized regression neural networks are recommended for efficient and accurate neutron spectrum retrieval.