Threshold Monitoring
Threshold monitoring is a type of anomaly detection. It works by scanning interval readings for values that are outside some expected range.
Threshold monitoring runs once a day on data collected up to and including midnight (i.e. 00:00 of the current day). Netstream runs threshold monitoring after Validation and Estimation jobs are completed, i.e. validated and/or estimated readings are monitored.
When to choose a threshold monitor
Whenever you want to be automatically notified of anomalous readings, and provided you can exactly define what an anomalous reading is, then a threshold monitor can be used.
The basic steps are:
- Define a rule. This is a logical expression that defines the threshold, e.g. "Value <= 10.6".
- Optionally, specify the duration the rule must be breached in order to trigger, e.g. "at least 2 hours".
- Specify an interval data reading type for the rule, e.g. "thirtyMinute deltaData export energy PhaseABC (kWh)"
- Select the usage points to monitor
The same usage point can have multiple monitors and the same rule can be applied to multiple usage points.
Configuring a threshold monitor
Click Reports > Threshold Monitoring to view what threshold monitoring is currently in place.
To get started you will need to add a new rule.
Creating rules
From the Threshold Monitor screen click View Rules (or Add Rule).
A sample rules screen is shown above. The Threshold Rules screen has the following format:
| Column Name | Description |
|---|---|
| Reading Type | The reading type this rule applies to |
| Rule Name | A system-generated description of the rule |
| Minimum Duration | The time span a threshold rule must be breached in order to trigger |
| No. of Usage Points | The count of usage points monitored using this rule |
To change a rule, click the pencil at the end of a row, or click Add to create a new rule and fill in the form.
The above rule means trigger when the energy reading exceeds 2.5 kWh in a 15-minute interval for at least two hours (i.e. 8 consecutive intervals) outside the hours 0830 to 1800. The system-generated name for the rule will be:
(Value > 2.5) and (TimeOfDay not between 08:30:00 and 18:00:00)
To specify a rule, enter the following information
- Reading type. Choose a reading type from the drop-down list paying attention to the measurement period and unit of measure.
- A relational operator and a value (mandatory), e.g. "greater than 2.5"
- An optional time of day: either before or after a time of day, or within or outside a time interval. If the toggle switch is turned off then the rule applies at all times.
- An optional day of the week. Turn the toggle switch on to choose the days of the week the rule applies or does not apply.
- An optional duration. If the toggle switch is off then the monitor will trigger if one sample is encountered satisfying the rule. Otherwise, specify the duration (in hours) the rule must be satisfied in order to trigger.
Editing a rule that applies to multiple usage points will update all the usage points to use the modified rule.
Click OK to save the rule.
For a selected rule click cog > View usage points to select the usage points to which the rule applies.
By default all usage points are shown in the list. Select Current members to see just the usage points that this rule is applied to. Use the filter boxes to search for the usage points you need to monitor and tick those that are required. There's also a 'select/deselect all' tick box in the header row.
Click OK to save the usage points assigned to the rule.
You can only select usage points that implement the metrology requirement stipulated in the rule, otherwise the tick box is greyed out.
Check there is at least one usage points for a given threshold rule by reviewing the No. of Usage Points field.
Threshold monitor status
From the Rules screen click the View Monitoring button to return to the Reports > Threshold Monitoring screen.
The system displays one monitor per row in the following tabular format:
| Column Name | Description |
|---|---|
| Rule Name | A system-generated description of the rule |
| Usage Point ID/NMI | Unique usage point identifier. Either a National Metering Identifier (NMI) where allocated or an internally generated identifier. |
| Enabled | A flag to indicate if the monitor is active. Toggle the flag using the context menu: cog > Disable. |
| Status | Colour coded dots indicating threshold results in the past 28 days |
Hover the mouse of the coloured dots to display the date and threshold monitoring result for the given day. The colours mean:
| Colour | Meaning |
|---|---|
| Green | The threshold monitor did not trigger |
| Amber | The monitor did not run that day and so had no result to report. This could be due to missing readings, e.g. offline meter |
| Red | The threshold monitor triggered. If there are multiple triggers in the same day Netstream only reports the latest one. |
Trigger actions
When a threshold monitor triggers the default action is to notify the user. Any user account associated with a Retailer can receivce notifications. It is not possible to send notifications outside the set of user accounts set up in Netstream, e.g. to a third party.
Furthermore, users control their own communication preferences and may elect to receive notifications and their preferred means. Refer to the My Settings screen for further information. Users can choose to be notified by push, email or (where available) SMS.
Above: a sample push notification.
Above: extract of a sample email notification.