Section 40 of 102
39. Anomaly Detection
Stable section ID: S05-CON-010-SECTION-40 · 28 content blocks
Anomaly Detection identifies conditions that differ materially from approved limits, expected behavior, established baselines, or related system observations.
Anomalies may include:
- Out-of-range measurements.
- Unexpected rate of change.
- Unusual energy or water consumption.
- Abnormal vibration, displacement, sound, or temperature.
- Conflicting sensor readings.
- Loss of communication.
- Repeated control failure.
- Unexpected equipment cycling.
- Interface lock-state changes.
- Moisture migration.
- Unrecognized configuration changes.
- Behavioral patterns associated with degradation.
The agent shall evaluate anomalies in context. A condition may be normal during maintenance or emergency operation but abnormal during ordinary occupancy.
Detected anomalies should be classified according to:
- Affected system.
- Location.
- Severity.
- Confidence.
- Urgency.
- Potential consequences.
- Duration.
- Recurrence.
- Required verification.
- Recommended response.
Anomaly detection shall not be treated as fault confirmation. The agent shall distinguish between a detected deviation, a suspected fault, a diagnosed cause, and a verified physical condition.
False alarms, missed detections, and threshold performance should be reviewed over time without weakening protection against low-frequency, high-consequence events.