Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

Alarm Philosophy and Novelty Detection for the ESS Fast Beam Interlock System

Utsi, Joel (2025)
Department of Automatic Control
Abstract
Machine Protection Systems (MPS) at the European Spallation Source (ESS) aim to protect equipment from damage while minimizing accelerator downtime with respect to safety and availability requirements. Main Control Room operators and MPS experts rely on alarm systems to effectively detect faults and guide intervention actions. Current implementations in the Fast Beam Interlock System (FBIS) are prone to cascading nuisance alarms which impede the operator’s response by causing unwanted alarms.

This report attempts to tackle the problems of nuisance alarms by developing a specific FBIS alarm philosophy. In addition to this, detection of abnormalities are not implemented in certain FBIS hardware components. Novelty detection techniques are... (More)
Machine Protection Systems (MPS) at the European Spallation Source (ESS) aim to protect equipment from damage while minimizing accelerator downtime with respect to safety and availability requirements. Main Control Room operators and MPS experts rely on alarm systems to effectively detect faults and guide intervention actions. Current implementations in the Fast Beam Interlock System (FBIS) are prone to cascading nuisance alarms which impede the operator’s response by causing unwanted alarms.

This report attempts to tackle the problems of nuisance alarms by developing a specific FBIS alarm philosophy. In addition to this, detection of abnormalities are not implemented in certain FBIS hardware components. Novelty detection techniques are therefore utilised to detect anomalies. This report limited the investigation of novelty detections schemes to the Median Absolute Deviation (MAD) and the One-Class Support Vector Machines (OCSVMs). Directed Acyclic Graphs (DAGs) were employed to model alarm propagation in the FBIS for the purpose of alarm revision.

Application of the MAD and OCSVMs to normal states of the FBIS hardware, and tuning boundary thresholds and bias respectively enabled the detection of outlier anomalies. The implementation of the FBIS alarm philosophy coupled with MAD for novelty detection showed that nuisance alarms could be significantly mitigated in the alarm service. (Less)
Please use this url to cite or link to this publication:
author
Utsi, Joel
supervisor
organization
year
type
H3 - Professional qualifications (4 Years - )
subject
report number
TFRT-6288
other publication id
0280-5316
language
English
id
9208534
date added to LUP
2025-08-08 15:07:13
date last changed
2025-08-08 15:07:13
@misc{9208534,
  abstract     = {{Machine Protection Systems (MPS) at the European Spallation Source (ESS) aim to protect equipment from damage while minimizing accelerator downtime with respect to safety and availability requirements. Main Control Room operators and MPS experts rely on alarm systems to effectively detect faults and guide intervention actions. Current implementations in the Fast Beam Interlock System (FBIS) are prone to cascading nuisance alarms which impede the operator’s response by causing unwanted alarms.

This report attempts to tackle the problems of nuisance alarms by developing a specific FBIS alarm philosophy. In addition to this, detection of abnormalities are not implemented in certain FBIS hardware components. Novelty detection techniques are therefore utilised to detect anomalies. This report limited the investigation of novelty detections schemes to the Median Absolute Deviation (MAD) and the One-Class Support Vector Machines (OCSVMs). Directed Acyclic Graphs (DAGs) were employed to model alarm propagation in the FBIS for the purpose of alarm revision.

Application of the MAD and OCSVMs to normal states of the FBIS hardware, and tuning boundary thresholds and bias respectively enabled the detection of outlier anomalies. The implementation of the FBIS alarm philosophy coupled with MAD for novelty detection showed that nuisance alarms could be significantly mitigated in the alarm service.}},
  author       = {{Utsi, Joel}},
  language     = {{eng}},
  note         = {{Student Paper}},
  title        = {{Alarm Philosophy and Novelty Detection for the ESS Fast Beam Interlock System}},
  year         = {{2025}},
}