### Abstract

The determination of the reactivity level in ADS represents nowadays an important open issue for the development of subcritical systems. Several approaches were developed in the past based on simplified models, such as point kinetics, for which an analytical inversion is feasible. On the other hand, numerical simulations shows that the evolution of the power and detector signals are not point-like. In the present work, we propose a neural-based inversion that can invert more realistic physical models. As a result, the spatial and spectral effects connected to the position and the energy selectivity of detectors are directly accounted for. An outline of the theory is presented in the following and some preliminary results show the performance of the technique proposed in the interpretation of pulsed experiments.

Original language | English (US) |
---|---|

Title of host publication | American Nuclear Society - International Conference on Mathematics, Computational Methods and Reactor Physics 2009, M and C 2009 |

Pages | 3049-3058 |

Number of pages | 10 |

State | Published - Dec 1 2009 |

Event | International Conference on Mathematics, Computational Methods and Reactor Physics 2009, M and C 2009 - Saratoga Springs, NY, United States Duration: May 3 2009 → May 7 2009 |

### Publication series

Name | American Nuclear Society - International Conference on Mathematics, Computational Methods and Reactor Physics 2009, M and C 2009 |
---|---|

Volume | 5 |

### Other

Other | International Conference on Mathematics, Computational Methods and Reactor Physics 2009, M and C 2009 |
---|---|

Country | United States |

City | Saratoga Springs, NY |

Period | 5/3/09 → 5/7/09 |

### Keywords

- Accelerator-driven systems
- Inverse techniques
- Neural networks
- Reactivity measurement

### ASJC Scopus subject areas

- Nuclear Energy and Engineering
- Computational Mathematics
- Nuclear and High Energy Physics

## Fingerprint Dive into the research topics of 'Subcriticality determination by neural-based inversion of space-energy neutron kinetic equations'. Together they form a unique fingerprint.

## Cite this

*American Nuclear Society - International Conference on Mathematics, Computational Methods and Reactor Physics 2009, M and C 2009*(pp. 3049-3058). (American Nuclear Society - International Conference on Mathematics, Computational Methods and Reactor Physics 2009, M and C 2009; Vol. 5).