Predicting particle accelerator failures using binary classifiers

Miha Rescic, Rebecca Seviour, Willem Blokland

Research output: Contribution to journalArticle

Abstract

Particle accelerator failures lead to unscheduled downtime and lower reliability. Although simple to mitigate while they are actually happening such failures are difficult to predict or identify beforehand. In this work we propose using machine learning approaches to predict machine failures via beam current measurements before they actual occur. To demonstrate this technique in this paper we examine beam pulses from the Oakridge Spallation Neutron Source (SNS). By evaluating a pulse against a set of common classification techniques we show that accelerator failure can be identified prior to actually failing with almost 80% accuracy. We also show that tuning classifier parameters and using pulse properties for refining datasets can further lead to almost 92% accuracy in classification of bad pulses. Most importantly, in the paper we establish there is information about the failure encoded in the pulses prior to it, so we also present a list of feasible next steps for increasing pulse classification accuracy.

Original languageEnglish
Article number163240
Number of pages8
JournalNuclear Instruments and Methods in Physics Research, Section A: Accelerators, Spectrometers, Detectors and Associated Equipment
Volume955
Early online date16 Dec 2019
DOIs
Publication statusE-pub ahead of print - 16 Dec 2019

Fingerprint

particle accelerators
classifiers
Particle accelerators
Classifiers
pulses
Neutron sources
Electric current measurement
Refining
Learning systems
Tuning
downtime
machine learning
refining
spallation
neutron sources
beam currents
lists
accelerators
tuning

Cite this

@article{89ff80fd9a614c7789ed0813ec0f14af,
title = "Predicting particle accelerator failures using binary classifiers",
abstract = "Particle accelerator failures lead to unscheduled downtime and lower reliability. Although simple to mitigate while they are actually happening such failures are difficult to predict or identify beforehand. In this work we propose using machine learning approaches to predict machine failures via beam current measurements before they actual occur. To demonstrate this technique in this paper we examine beam pulses from the Oakridge Spallation Neutron Source (SNS). By evaluating a pulse against a set of common classification techniques we show that accelerator failure can be identified prior to actually failing with almost 80{\%} accuracy. We also show that tuning classifier parameters and using pulse properties for refining datasets can further lead to almost 92{\%} accuracy in classification of bad pulses. Most importantly, in the paper we establish there is information about the failure encoded in the pulses prior to it, so we also present a list of feasible next steps for increasing pulse classification accuracy.",
keywords = "Accelerator reliability, Binary classifier, Failure prediction, Machine learning, Particle accelerator",
author = "Miha Rescic and Rebecca Seviour and Willem Blokland",
year = "2019",
month = "12",
day = "16",
doi = "10.1016/j.nima.2019.163240",
language = "English",
volume = "955",
journal = "Nuclear Instruments and Methods in Physics Research, Section A: Accelerators, Spectrometers, Detectors and Associated Equipment",
issn = "0168-9002",
publisher = "Elsevier",

}

TY - JOUR

T1 - Predicting particle accelerator failures using binary classifiers

AU - Rescic, Miha

AU - Seviour, Rebecca

AU - Blokland, Willem

PY - 2019/12/16

Y1 - 2019/12/16

N2 - Particle accelerator failures lead to unscheduled downtime and lower reliability. Although simple to mitigate while they are actually happening such failures are difficult to predict or identify beforehand. In this work we propose using machine learning approaches to predict machine failures via beam current measurements before they actual occur. To demonstrate this technique in this paper we examine beam pulses from the Oakridge Spallation Neutron Source (SNS). By evaluating a pulse against a set of common classification techniques we show that accelerator failure can be identified prior to actually failing with almost 80% accuracy. We also show that tuning classifier parameters and using pulse properties for refining datasets can further lead to almost 92% accuracy in classification of bad pulses. Most importantly, in the paper we establish there is information about the failure encoded in the pulses prior to it, so we also present a list of feasible next steps for increasing pulse classification accuracy.

AB - Particle accelerator failures lead to unscheduled downtime and lower reliability. Although simple to mitigate while they are actually happening such failures are difficult to predict or identify beforehand. In this work we propose using machine learning approaches to predict machine failures via beam current measurements before they actual occur. To demonstrate this technique in this paper we examine beam pulses from the Oakridge Spallation Neutron Source (SNS). By evaluating a pulse against a set of common classification techniques we show that accelerator failure can be identified prior to actually failing with almost 80% accuracy. We also show that tuning classifier parameters and using pulse properties for refining datasets can further lead to almost 92% accuracy in classification of bad pulses. Most importantly, in the paper we establish there is information about the failure encoded in the pulses prior to it, so we also present a list of feasible next steps for increasing pulse classification accuracy.

KW - Accelerator reliability

KW - Binary classifier

KW - Failure prediction

KW - Machine learning

KW - Particle accelerator

UR - http://www.scopus.com/inward/record.url?scp=85076704576&partnerID=8YFLogxK

U2 - 10.1016/j.nima.2019.163240

DO - 10.1016/j.nima.2019.163240

M3 - Article

AN - SCOPUS:85076704576

VL - 955

JO - Nuclear Instruments and Methods in Physics Research, Section A: Accelerators, Spectrometers, Detectors and Associated Equipment

JF - Nuclear Instruments and Methods in Physics Research, Section A: Accelerators, Spectrometers, Detectors and Associated Equipment

SN - 0168-9002

M1 - 163240

ER -