Abstract:In the field of nuclear science, obtaining and utilizing nuclear data, including nuclear reaction data, nuclear structure information, and radioactive decay data, is crucial. Neutron-induced nuclear reactions, particularly nuclear cross sections data, are essential for various applications, including reactor design. The EXFOR database is the only international repository for storing nuclear reaction experimental measurement information and data. However, experimental measurement data are often scarce, subject to discrepancies, or even errors, requiring human evaluation. This process can be prone to biases and significant uncertainties. To address these challenges, this study proposes a novel framework, Feature Engineering for Nuclear Reaction Cross Section Generation using Machine Learning (FECSG-ML), which employs machine learning methods to generate nuclear cross sections data, serving as a substitute for evaluating nuclear databases. Given the limited size of the EXFOR database, training a model solely on EXFOR data could lead to underfitting. Therefore, the proposed approach utilizes transfer learning, initially pre-training the model using the ENDF/B-VIII.0 dataset and subsequently fine-tuning it with the EXFOR database. This approach ensures high accuracy where real data are available and enables the learning of characteristics of the evaluation dataset where real data are lacking. Moreover, machine learning techniques are employed to transform discrete nuclear cross sections data into a continuous format, accommodating various isotopes and predicting multiple sets of cross sections data. The framework integrates various machine learning methods and utilizes ensemble learning for result optimization. Experimental results demonstrate that the regression curves generated by the FECSG-ML model align well with EXFOR data points, outperforming the ENDF/B-VIII.0 evaluation database. Furthermore, the nuclear cross sections data generated by the FECSG-ML model are applied in the OpenMC Monte Carlo simulation program to simulate pin fuel assemblies and CANDU reactors, confirming the effectiveness of the model. This study underscores the importance of accurate and reliable nuclear cross sections data and provides a method for substituting the evaluation of nuclear databases.

Comprehensive estimation of nuclide production cross sections using a phenomenological approach

Methodology for physics-informed generation of synthetic neutron time-of-flight measurement data

Gaussian Processes for Surrogate Modeling of Discharged Fuel Nuclide Compositions

FECSG-ML: Feature Engineering for Nuclear Reaction Cross Sections Generation Using Machine Learning

Neutron-Induced Nuclear Cross-Sections Study for Plasma Facing Materials via Machine Learning: Molybdenum Isotopes

Improving nuclear data evaluations with predictive reaction theory and indirect measurements

Quantification of nuclear uncertainties in nucleosynthesis of elements beyond Iron

Proton and Neutron Induced SEU Cross Section Modeling and Simulation: A Unified Analytical Approach

Illuminating Systematic Trends in Nuclear Data with Generative Machine Learning Models

Nuclear mass predictions using machine learning models

A data-driven machine learning approach for electron-molecule ionization cross sections

Cosmic Ray particle production

Machine learning in nuclear materials research

Production mechanism of neutron-deficient actinide isotopes in complete fusion reactions and multinucleon transfer reactions

Uncertainty Quantification for Nuclear Density Functional Theory and Information Content of New Measurements

Predictions for the ( ) reaction cross section based on a Bayesian neural network approach

Production cross sections of superheavy elements: insights from the dinuclear system model with high-quality microscopic nuclear masses

Building Surrogate Models of Nuclear Density Functional Theory with Gaussian Processesand Autoencoders

Neutron-Alpha Reaction Cross Section Determination by Machine Learning Approaches

Reliable Calculations of Nuclear Binding Energies by the Gaussian Process of Machine Learning