Plant health: replay
Recorded streams of plant measurements (two episodes), replayed through the detector sample by sample (every 3 minutes of plant time).
Advisory only. This page shows the detector's status. It never changes plant controls, and nothing here is an instruction to act.
Connecting…
Status band
—
Alert ratio (limit = 1)
—
01 (limit)3+
Plant time
—
Equipment groups (Watch above 1)
Diagnosis of the alert (advisory)
Press Play: a diagnosis appears 30 minutes (plant time) after an alert starts.
The explanations were computed once, in advance; this page never contacts a language model. The operator notes are fixed examples: this page takes no typed text. Nothing here approves or carries out an action.
Normal
Watch
Alert
Unknown (warm-up or missing data)
Limit (ratio = 1)
- The Watch band isn't built yet: it needs equipment-group attribution.
- During an alert, the marked group is where the detector's reconstruction of the data points when the alert began. It shows where the symptoms appear, not what caused them, and it isn't an instruction to act.
- The ratio is how far the plant-wide statistic is from its alert limit; above 1 is outside normal operation. An alert needs several samples in a row above 1, and it stays on for a hold time after the ratio drops back, so one disturbance gives one alert.
- Each point is computed only from data up to that moment: nothing later changes what was shown earlier.
- The first samples of a stream are Unknown: the detector needs a short warm-up before it scores.