Anomaly Detection

Anomalies is a feature aimed to detect and inform customers about suspected consumption deviations in their home or business that could be caused by an appliance failure. Anomaly status can be shown to the end user in an app or provided as an event to send out to end users as a notification, or both.

How does it work?

The anomalies algorithms are event-based and run when data for a new day or month has been processed. If an anomaly is detected, a notification is scheduled to be sent out during the following day.

Anomaly notifications are available through the Eliq Webhook Solution (see more details below).

Daily Anomaly Notification

The Daily Anomaly Notification alerts end-users when the energy usage for a day is unusually high or low. This can help the end-user to identify events that uses a lot of energy or become aware if a device is faulty/broken using significantly more or less energy than normal. This is useful in many cases, one example being vacation home owners getting notified if their home suddenly uses less energy which could be caused by a heating system or refrigerator that stops working.

How it works

Eliq learns each location’s normal pattern, including seasonal changes, and flags days that fall well outside it. The threshold is set per location, so homes with very steady use and homes with large swings both get few false alerts.

High Monthly Forecast Notification

The High Monthly Forecast Notification alerts end-users when the forecasted energy usage for a month is unusually high. This notification can warn of potential high energy usage that could result in a bill shock allowing the end user to act before it is too late. It also serves the purpose of informing the end-user of an unusually high consumption and potential bill shock ahead of the bill arriving.

How it works

At the start of a month, a forecast using all available data is done to predict the total energy usage for the coming month. This forecast is updated and evaluated daily as new energy data is available. If the forecast based on new data is unusually high when compared with previous forecasts for the location, a High Monthly Forecast Anomaly event is created which can alert the end user.

API Reference Documentation

The Insights API Endpoint Get Anomalies can be used request the list of the Anomalies that are enabled for a location including the current status for each anomaly. This can be used to show an Anomaly Status for the end user in an app.

Webhook Solution

The Eliq Webhook Solution can be used to receive all events that Eliq can provide for different features and to generate notifications to end users in apps. Read more about how to Get Started using the Eliq Webhook Solution.

Supported fuel types and Units

Anomalies are available for the following fuels and units.

• Electricity – Energy
• Gas – Energy
• District Heating – Energy

Supported Location Profile Types

Anomalies can be generated for both Residential and Business Locations. For more information about Location Profile Types, please visit the Location Profile article

Requirements

Required data – Daily Anomalies

• Minimum 30 days of daily energy data.
• Works best with 365 days of historic data or more.

Required data – High Monthly Forecast

• Minimum 30 days of daily energy data.
• Minimum 6 months of monthly energy data.

Use Cases

Daily Anomaly

Daily anomaly alerts can inform of unusually high or low consumption values compared to consumption patterns for previous days. Below are few examples when anomalies can be applicable:

• A faulty heating or hot water heating device that is suddenly consuming unusually high amounts of electricity.
• A faulty heating or how water heating device that has suddenly turned off.
• Empty holiday homes where no-one is living that see a sudden increase or decrease of electricity usage due to a faulty appliance.

High monthly forecast

High monthly forecast alerts aim to help inform an end user of a unusually high forecasted consumption and to act before the end of the month or to inform the user that the current months energy consumption is expected to be a lot higher and to reduce bill shock. Some examples are:

• A faulty appliance that has been running unusually high over the course of several days of a month but not enough for daily anomaly detection.
• Alert about natural increase in energy consumption e.g. due to more people living in the home.

‍

‍

‍