AI Agent for Alarm Rationalization

Objective

Develop an agent that continuously tracks a plant's alarms, computes alarm KPIs, detects bad actors and identifies related alarm groups. The agent will draft a rationalization sheet proposing deadband, delay, limit and priority changes, with a written rationale and supporting evidence. Before implementation, historical data will be replayed to estimate the number of alarms that would be removed. Engineers will review and approve every proposal. After implementation, the system will re-measure performance and produce a before-and-after report for the client.

Scope

• Study alarm management principles and alarm KPIs.
• Compute floods, chattering and stale alarms from alarm history.
• Detect bad actors and groups of related alarms.
• Draft a rationalization sheet with proposed settings, rationale and evidence.
• Replay historical data to estimate alarm reduction before any change.
• Generate a reviewable change set and a before-and-after report.

Technologies and keywords

Alarm Rationalization, ISA-18.2, AI, Ignition, KPIs.

Position details

Engineering internship; 1 intern; end-of-studies project lasting 4–6 months.