Create an Asset Anomaly Detection Guided Model

To create a model

1. On the Asset screen, select an asset.

2. Select Add Asset Anomaly Detection Guided Model from the Asset Actions menu.

   

   The Guided Predictive Analytics Model Details side sheet opens.

   

3. Enter a Model name.

4. Verify the Asset.

   To create a model for a different asset, back out of the side sheet and select a different asset.

5. Select the Tags to include in the model.

   Only analog tag types are supported.

   

6. If applicable, configure a filter to limit when to include some tags in the model calculations.

   Note You can only create one filter for each model and the filtered tags must be either analog or string type.

   

7. Set a Training Window to identify which historical data informs the model what normal behavior looks like.

   

   AVEVA Insight pulls training data every 60 minutes during the Training Window time range.

8. Select Save.

   Note: If the parent asset is associated with an asset type template, you'll have the option of saving this model to only the asset, or to the asset type for replication to other asset instances. For more information, see Asset Types.

Now, the algorithm:

• Gathers the training data based on your parameters

• Builds a model

• Begins predicting new values and calculating the anomaly score every 15 minutes

• Generates output tags that mimic the input tag name, but with a _(modelname)_predicted suffix

One output tag is generated per model. The tag calculates an overall anomaly score for that model, named {modelname}.AnomalyScore.

After a model is saved, you cannot disable it. You can only remove a model by deleting it, which also deletes the anomaly score and predicted tags.

AVEVA Insight creates a News entry when five anomaly scores greater than 10 are detected within a 90-minute time period.

News

In a News bar chart, click any bar to open a line chart of each tag.

Guided Analytics Models

On the Assets page, the Guided Analytics Models card displays all the models.

For an asset, you can view models that anybody created. Administrators can edit or delete any model. You can only edit or delete models that you created.

To edit a model

9. Open the model's action menu and then select Edit.

   

10. The Guided Predictive Analytics Model Details side sheet opens.

11. If applicable, change the Model Name.

    Note Changing the model name creates a new set of anomaly score tags but does not remove the existing old anomaly score tags. If required, delete the old anomaly score tags manually.

    

12. If applicable, delete tags by clicking the icon.

    Note Deleting tags from the model does not remove the existing anomaly score tags that were previously created. If required, delete the old anomaly score tags manually.

13. If applicable, change the filters or training window.

14. Click Save.