Prediction Monitoring, Telemetry & Alarms

A trained prediction model is a snapshot. Monitoring keeps it current: Trendz refreshes the forecasts on a schedule, writes them to ThingsBoard as telemetry, and raises alarms from event rules and targets.

Every periodic refresh runs the same steps for each item of the model. Four switches decide which steps run:

Switch	What the model does with each refresh
Periodic refresh	Runs the refresh itself, every hour, day, or whatever period you set
Scheduled retrain	Refits the model on the latest data before it predicts
Save telemetry	Writes the forecast, event countdowns, and target progress to ThingsBoard
Alarm creation	Lets event rules and targets raise ThingsBoard alarms

Click Monitoring in the page header (1), or Configure on the Continuous monitoring card (2). Both open the same dialog with the four switches.

Switch Periodic refresh on and set how often it runs and for which items.

Field	Description
Frequency	How often the refresh runs: a count and a unit, Minute, Hour, Day, or Week
Align to time unit start	Starts each window on a whole unit: a 6-hour refresh at 00:00, 06:00, 12:00, and 18:00
Scope	All profile items covers every item of the profile, All trained items only the trained ones, Specific the items you pick

A refresh that finds no new data for an item skips it.

Switch Scheduled retrain on. Each refresh then first learns from the newest readings, and only then predicts.

Switch Save telemetry on and enter a Telemetry key. Trendz writes the forecast to each device in ThingsBoard under that key with an _EPD_ prefix, and updates it at every refresh. You can then show it on dashboards or use it in rule chains.

This switch is also the master switch for the telemetry of event rules and targets. To learn what those write, see Time to Threshold and Hit Target.

Switch Alarm creation on to let the model raise alarms. This switch is shown only for Time to Threshold and Hit Target models, the two intents that raise alarms. It is the master switch for every event rule and target of the model: with it off, no alarm is raised, whatever the rules and targets say.

Click Save changes. The header badge turns to Enabled, and the Continuous monitoring card shows the new settings.

Alarms come from the rules and targets of a model, and each intent sets them up on its own page:

• Time to Threshold: event rules

  Raise an alarm before a metric crosses a limit, with the severity rising as the moment gets closer.

• Hit Target: targets

  Raise an alarm while a period is heading to miss its goal, with the severity following the risk.

• Forecast

  Predict the next values of a metric, like the energy of the next 7 days.

• Time to Threshold

  Predict when a metric crosses a limit, and raise an alarm before it happens.

• Period Total

  Predict the total, peak, or average of a period, such as the energy of this month.

• Hit Target

  Predict whether a goal is reached by the end of a period, and how far off it is.

• Explore Interface

  Find your way around the models list, the model page, and its tabs.

• Models & accuracy

  How a model learns, its settings, and how accurate its forecasts are.

• Show on dashboards

  Chart history and forecast together in Trendz views and ThingsBoard dashboards.