Skip to content

Model Setup

"Model Setup" can be accessed in either of the following ways:

  • On the models overview page, click the "Create Model" button, or
  • From an existing model card, click the "Config Model" button

Creating or configuring a model includes the following steps:

Model Information

In this section, configure the general details of the model such as the name, description and select the mandatory variables. The initial anomaly alerting feature for batch production requires the following four variables:

  • Energy Consumer
  • Energy Resource
  • Batch variable
  • Reference variable

model-setup-information

For the variables to be available here, all these variables must be setup previously in the energy consumer Configuration.

When all variables are selected, click "Next" to proceed.

Baseline Data

In this section, select the baseline data that will be used to build and train the model. Use the date range picker to select the relevant data. The aggregated batch data for Energy per Reference is shown in the time chart, color coded for each product. Scroll horizontally through the time chart and zoom in to see the data more clearly. However, it is the full data set in the date range that is used as the baseline.

model-setup-baseline

Histograms are displayed below the time charts, that shows the distributions of Energy per Reference, one histogram for each product. The Energy Optimizer modelling framework includes automatic validation checks to ensure that the selected data is consistent and therefore suitable to be used as a baseline data set. If the histogram is bordered by a red box, then this means the data has failed the validation checks and therefore is not suitable for a baseline. When a longer time period is selected, one of the products passes the validation checks which is indicated by the green box.

model-setup-baseline-longer

It is possible to build the model even if only one product's data passes the validation checks. However, only the products with validated baseline data will be monitored. The products that did not have sufficient data will not be included in the model.

When the baseline data is finalised, click "Next" to proceed.

Performance Standard

In this section, specify the energy performance standard limits for the model. These standards correspond to the threshold values, above which an alert will be generated.

model-setup-standard

Similar to the previous section, histograms are displayed for each product. A vertical line indicates the threshold value for the Energy per Reference for that product. By default, this is set at the 95th percentile of the distribution. These values can be adjusted manually. When the performance standard limit is entered manually, the probability of exceeding the performance limit changes from its default of 5%. Also, an estimate of the number of anomalies that would be detected each day using this value is displayed. This is a useful indication of the number of alerts that would be received each day. Note that only product standard limits passing the validation checks can be edited.

Configure Actions

In this section, enter the email address to send notifications when there is an alert.

model-setup-notification

Click "Save". A warning message is displayed stating that only some of the products will be monitored with this model. Click "Confirm".

model-setup-save

The new model card will be displayed in the model overview page. The model will now be in the "Ready" state. To run the model, open the card menu and click "Start Model".

Note

When you start the model, one unit of your quota will be used.


Last update: January 15, 2025