Causal Alarm Pattern Analysis by the Integration of Technical Information from Engineering Documents (CausAlITI)

The CausAlITI project adressed alarm overload in industrial automation by automating causal analysis. It combined engineering knowledge from control logic, process diagrams, and docu-mentation with machine learning to identify cause–effect relationships between alarms. The developed methods learn from historical incidents, reason about new ones, and provide prob-abilistic assessments under uncertainty. As part of the project, several simulation datasets of industrial scenarios were produced and published as benchmarks for the scientific community. These were used to evaluate the developed analysis methods, many of which have also been made available as open source tools.