Improved online localisation of CANDU duel defects using ancillary data sources and neural networks

Research output: Contribution to journalArticle

Abstract

This paper summarizes a novel approach to improved localization of fuel defects by fusing existing data sources and methods within a neural network model to make accurate and quantifiable identification earlier than existing processes. The approach is demonstrated through application to a CANDU reactor and utilizes a small, manually labeled set of delayed neutron data augmented with neutronic power data to train a neural network to estimate the probability of a fuel channel containing a defect. Results demonstrate that the model is often capable of identifying likely defects earlier than existing methods and could support earlier decision making to enable a reduction in cost and time required to localize defects. The approach described has broader application to other reactor types given the general difficulty of detecting fuel defects via fission product measurement and the large quantities of ancillary parameters normally already recorded that can be leveraged using machine learning techniques.
Original languageEnglish
JournalNuclear Technology
Early online date27 Jan 2020
DOIs
Publication statusE-pub ahead of print - 27 Jan 2020

Fingerprint

Neural networks
Defects
defects
reactors
machine learning
fission products
Fission products
decision making
Learning systems
Identification (control systems)
Neutrons
Decision making
costs
neutrons
estimates
Costs

Keywords

  • fuel reliability
  • condition monitoring
  • machine learning
  • decision support

Cite this

@article{ab3c366a2c5b41a7a645c086ef311caf,
title = "Improved online localisation of CANDU duel defects using ancillary data sources and neural networks",
abstract = "This paper summarizes a novel approach to improved localization of fuel defects by fusing existing data sources and methods within a neural network model to make accurate and quantifiable identification earlier than existing processes. The approach is demonstrated through application to a CANDU reactor and utilizes a small, manually labeled set of delayed neutron data augmented with neutronic power data to train a neural network to estimate the probability of a fuel channel containing a defect. Results demonstrate that the model is often capable of identifying likely defects earlier than existing methods and could support earlier decision making to enable a reduction in cost and time required to localize defects. The approach described has broader application to other reactor types given the general difficulty of detecting fuel defects via fission product measurement and the large quantities of ancillary parameters normally already recorded that can be leveraged using machine learning techniques.",
keywords = "fuel reliability, condition monitoring, machine learning, decision support",
author = "Christopher Wallace and Curtis McEwan and Graeme West and William Aylward and Stephen McArthur",
year = "2020",
month = "1",
day = "27",
doi = "10.1080/00295450.2019.1697174",
language = "English",
journal = "Nuclear Technology",
issn = "0029-5450",

}

TY - JOUR

T1 - Improved online localisation of CANDU duel defects using ancillary data sources and neural networks

AU - Wallace, Christopher

AU - McEwan, Curtis

AU - West, Graeme

AU - Aylward, William

AU - McArthur, Stephen

PY - 2020/1/27

Y1 - 2020/1/27

N2 - This paper summarizes a novel approach to improved localization of fuel defects by fusing existing data sources and methods within a neural network model to make accurate and quantifiable identification earlier than existing processes. The approach is demonstrated through application to a CANDU reactor and utilizes a small, manually labeled set of delayed neutron data augmented with neutronic power data to train a neural network to estimate the probability of a fuel channel containing a defect. Results demonstrate that the model is often capable of identifying likely defects earlier than existing methods and could support earlier decision making to enable a reduction in cost and time required to localize defects. The approach described has broader application to other reactor types given the general difficulty of detecting fuel defects via fission product measurement and the large quantities of ancillary parameters normally already recorded that can be leveraged using machine learning techniques.

AB - This paper summarizes a novel approach to improved localization of fuel defects by fusing existing data sources and methods within a neural network model to make accurate and quantifiable identification earlier than existing processes. The approach is demonstrated through application to a CANDU reactor and utilizes a small, manually labeled set of delayed neutron data augmented with neutronic power data to train a neural network to estimate the probability of a fuel channel containing a defect. Results demonstrate that the model is often capable of identifying likely defects earlier than existing methods and could support earlier decision making to enable a reduction in cost and time required to localize defects. The approach described has broader application to other reactor types given the general difficulty of detecting fuel defects via fission product measurement and the large quantities of ancillary parameters normally already recorded that can be leveraged using machine learning techniques.

KW - fuel reliability

KW - condition monitoring

KW - machine learning

KW - decision support

U2 - 10.1080/00295450.2019.1697174

DO - 10.1080/00295450.2019.1697174

M3 - Article

JO - Nuclear Technology

JF - Nuclear Technology

SN - 0029-5450

ER -