Causal event processes and alarm analysis at the European Spallation Source
(2025)Department of Automatic Control
- Abstract
- This thesis investigates the causal relationships among alarms at the European Spallation Source (ESS), focusing on enhancing alarm management through advanced data analytics. The study explores these relationships using Dynamic Bayesian Networks (DBNs) complemented with well-defined Conditional Probability Distributions (CPTs). This approach models the cascading interactions of alarms based on theoretical causal links and leverages simulated datasets with a known causal structure as a baseline for analysis.
By employing both parametric (LASSO logistic regression) and non-parametric (Graph Neural Network) learning algorithms, the graphical structure of causal relationships among alarms, initially modeled using Dynamic Bayesian Networks,... (More) - This thesis investigates the causal relationships among alarms at the European Spallation Source (ESS), focusing on enhancing alarm management through advanced data analytics. The study explores these relationships using Dynamic Bayesian Networks (DBNs) complemented with well-defined Conditional Probability Distributions (CPTs). This approach models the cascading interactions of alarms based on theoretical causal links and leverages simulated datasets with a known causal structure as a baseline for analysis.
By employing both parametric (LASSO logistic regression) and non-parametric (Graph Neural Network) learning algorithms, the graphical structure of causal relationships among alarms, initially modeled using Dynamic Bayesian Networks, is inferred and validated. These relationships are tested against simulated datasets with a predefined causal structure. The effectiveness of these learning algorithms is then assessed using performance metrics to ensure their applicability and accuracy. Following this evaluation, these methods have been successfully implemented on alarm datasets from ESS to establish causal link graphs.
The findings demonstrate that the adopted approach can reduce the frequency of nuisance alarms and enhance the operator’s ability to respond to genuine risks. This investigation contributes to operational safety and efficiency by providing a framework that enhances situational awareness and decision-making through improved analysis and management of alarm data. (Less)
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/student-papers/record/9183988
- author
- Raveendran, Vishnu Pradheep
- supervisor
- organization
- year
- 2025
- type
- H3 - Professional qualifications (4 Years - )
- subject
- report number
- TFRT-6267
- other publication id
- 0280-5316
- language
- English
- id
- 9183988
- date added to LUP
- 2025-02-04 14:06:00
- date last changed
- 2025-02-04 14:06:00
@misc{9183988,
abstract = {{This thesis investigates the causal relationships among alarms at the European Spallation Source (ESS), focusing on enhancing alarm management through advanced data analytics. The study explores these relationships using Dynamic Bayesian Networks (DBNs) complemented with well-defined Conditional Probability Distributions (CPTs). This approach models the cascading interactions of alarms based on theoretical causal links and leverages simulated datasets with a known causal structure as a baseline for analysis.
By employing both parametric (LASSO logistic regression) and non-parametric (Graph Neural Network) learning algorithms, the graphical structure of causal relationships among alarms, initially modeled using Dynamic Bayesian Networks, is inferred and validated. These relationships are tested against simulated datasets with a predefined causal structure. The effectiveness of these learning algorithms is then assessed using performance metrics to ensure their applicability and accuracy. Following this evaluation, these methods have been successfully implemented on alarm datasets from ESS to establish causal link graphs.
The findings demonstrate that the adopted approach can reduce the frequency of nuisance alarms and enhance the operator’s ability to respond to genuine risks. This investigation contributes to operational safety and efficiency by providing a framework that enhances situational awareness and decision-making through improved analysis and management of alarm data.}},
author = {{Raveendran, Vishnu Pradheep}},
language = {{eng}},
note = {{Student Paper}},
title = {{Causal event processes and alarm analysis at the European Spallation Source}},
year = {{2025}},
}