EdgeVISS Local
Alarm Management
Threshold evaluation built for real plants: noise filtering, ISA-18.2 suppression, and a return-to-normal event that fires once — not a flood of alarms on every recovery.
1
Threshold Load
Setpoints load from an external IT threshold API (setOT) or manual fallback thresholds — whichever is more current.
2
Evaluate
HA → HW → LA → LW, in that order, against the latest reading.
3
Filter
Deadband and rate-limiting suppress noise; RBE (report-by-exception) fires only on real state changes.
4
Suppression Check
ISA-18.2 shelving — 15min/1h/4h/8h/24h or indefinite — silences a known issue without deleting the setpoint.
5
Publish
One MQTT alarm event, with priority (1–999). Return-to-normal publishes exactly one clear event, not a flood.
Beyond a threshold check
Device offline detection
A 60-second background check flags a device that stopped reporting as a High Alert, and clears it automatically on recovery — no separate "is it alive" tooling needed.
ISA-18.2 priority and shelving
Numeric priority (1=highest, 999=lowest) orders the active-alarm view; shelving suppresses a known nuisance alarm for a fixed window without silencing the whole setpoint permanently.
External threshold sync
Pull thresholds from your own IT system via the setOT API — EdgeVISS Local caches them locally so evaluation keeps working even if that system is briefly unreachable.