Application of artificial neural networks to infer subcriticality level through kinetic models

Paolo Picca, Roberto Furfaro, Barry D Ganapol, Sandra Dulla, Piero Ravetto

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Citations (Scopus)

Abstract

The paper presents some recent advances in the study of the inverse kinetics for subcritical systems. A neural-based approach is adopted to predict the reactivity of the multiplying medium through the analysis of the reactor response to a source pulse. An artificial neural network is designed to infer the subcriticality level through the analysis of power evolution. The training set is computed using an approximate model and its performances are then tested directly on experimental measures, here simulated through a detailed space-energy kinetic model. In order to improve the accuracy of the reactivity estimation, various strategies are proposed and compared, including a multi-transient inversion and the use of different kinetic models for the training. The issue of robustness of the inversion scheme to experimental noise is also addressed.

Original languageEnglish (US)
Title of host publicationInternational Conference on the Physics of Reactors 2010, PHYSOR 2010
Pages397-407
Number of pages11
Volume1
StatePublished - 2010
EventInternational Conference on the Physics of Reactors 2010, PHYSOR 2010 - Pittsburgh, PA, United States
Duration: May 9 2010May 14 2010

Other

OtherInternational Conference on the Physics of Reactors 2010, PHYSOR 2010
CountryUnited States
CityPittsburgh, PA
Period5/9/105/14/10

Fingerprint

Neural networks
Kinetics
kinetics
education
reactivity
inversions
Kinetic energy
kinetic energy
reactors
pulses

Keywords

  • Accelerator-driven system
  • Artificial neural network
  • Inverse problem
  • Neutron kinetics

ASJC Scopus subject areas

  • Nuclear Energy and Engineering
  • Nuclear and High Energy Physics

Cite this

Picca, P., Furfaro, R., Ganapol, B. D., Dulla, S., & Ravetto, P. (2010). Application of artificial neural networks to infer subcriticality level through kinetic models. In International Conference on the Physics of Reactors 2010, PHYSOR 2010 (Vol. 1, pp. 397-407)

Application of artificial neural networks to infer subcriticality level through kinetic models. / Picca, Paolo; Furfaro, Roberto; Ganapol, Barry D; Dulla, Sandra; Ravetto, Piero.

International Conference on the Physics of Reactors 2010, PHYSOR 2010. Vol. 1 2010. p. 397-407.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Picca, P, Furfaro, R, Ganapol, BD, Dulla, S & Ravetto, P 2010, Application of artificial neural networks to infer subcriticality level through kinetic models. in International Conference on the Physics of Reactors 2010, PHYSOR 2010. vol. 1, pp. 397-407, International Conference on the Physics of Reactors 2010, PHYSOR 2010, Pittsburgh, PA, United States, 5/9/10.
Picca P, Furfaro R, Ganapol BD, Dulla S, Ravetto P. Application of artificial neural networks to infer subcriticality level through kinetic models. In International Conference on the Physics of Reactors 2010, PHYSOR 2010. Vol. 1. 2010. p. 397-407
Picca, Paolo ; Furfaro, Roberto ; Ganapol, Barry D ; Dulla, Sandra ; Ravetto, Piero. / Application of artificial neural networks to infer subcriticality level through kinetic models. International Conference on the Physics of Reactors 2010, PHYSOR 2010. Vol. 1 2010. pp. 397-407
@inproceedings{b38ad83702334b198f092e2c31af2449,
title = "Application of artificial neural networks to infer subcriticality level through kinetic models",
abstract = "The paper presents some recent advances in the study of the inverse kinetics for subcritical systems. A neural-based approach is adopted to predict the reactivity of the multiplying medium through the analysis of the reactor response to a source pulse. An artificial neural network is designed to infer the subcriticality level through the analysis of power evolution. The training set is computed using an approximate model and its performances are then tested directly on experimental measures, here simulated through a detailed space-energy kinetic model. In order to improve the accuracy of the reactivity estimation, various strategies are proposed and compared, including a multi-transient inversion and the use of different kinetic models for the training. The issue of robustness of the inversion scheme to experimental noise is also addressed.",
keywords = "Accelerator-driven system, Artificial neural network, Inverse problem, Neutron kinetics",
author = "Paolo Picca and Roberto Furfaro and Ganapol, {Barry D} and Sandra Dulla and Piero Ravetto",
year = "2010",
language = "English (US)",
isbn = "9781617820014",
volume = "1",
pages = "397--407",
booktitle = "International Conference on the Physics of Reactors 2010, PHYSOR 2010",

}

TY - GEN

T1 - Application of artificial neural networks to infer subcriticality level through kinetic models

AU - Picca, Paolo

AU - Furfaro, Roberto

AU - Ganapol, Barry D

AU - Dulla, Sandra

AU - Ravetto, Piero

PY - 2010

Y1 - 2010

N2 - The paper presents some recent advances in the study of the inverse kinetics for subcritical systems. A neural-based approach is adopted to predict the reactivity of the multiplying medium through the analysis of the reactor response to a source pulse. An artificial neural network is designed to infer the subcriticality level through the analysis of power evolution. The training set is computed using an approximate model and its performances are then tested directly on experimental measures, here simulated through a detailed space-energy kinetic model. In order to improve the accuracy of the reactivity estimation, various strategies are proposed and compared, including a multi-transient inversion and the use of different kinetic models for the training. The issue of robustness of the inversion scheme to experimental noise is also addressed.

AB - The paper presents some recent advances in the study of the inverse kinetics for subcritical systems. A neural-based approach is adopted to predict the reactivity of the multiplying medium through the analysis of the reactor response to a source pulse. An artificial neural network is designed to infer the subcriticality level through the analysis of power evolution. The training set is computed using an approximate model and its performances are then tested directly on experimental measures, here simulated through a detailed space-energy kinetic model. In order to improve the accuracy of the reactivity estimation, various strategies are proposed and compared, including a multi-transient inversion and the use of different kinetic models for the training. The issue of robustness of the inversion scheme to experimental noise is also addressed.

KW - Accelerator-driven system

KW - Artificial neural network

KW - Inverse problem

KW - Neutron kinetics

UR - http://www.scopus.com/inward/record.url?scp=79952417014&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=79952417014&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:79952417014

SN - 9781617820014

VL - 1

SP - 397

EP - 407

BT - International Conference on the Physics of Reactors 2010, PHYSOR 2010

ER -