TrendMiner 2020.R1

21st of April 2020

The 2020.R1 release contains the following new functionality and improvements:

New Functionalities

  • Nested calculations: Extend your calculations beyond historian tags and create a formula of a formula, an aggregation of a formula or even an aggregation of an aggregation, and use them in searches, fingerprints, monitors etc...
  • Organize & share aggregations and formulas: Organize your saved formulas and aggregations together with your saved views, fingerprints etc. within (project) folders in your work organizer, and benefit from the existing sharing options to spread the word and give your colleagues insights into your searches and calculations.
  • Import of discrete and string tags: As well as analog tags, import discrete and string type tags under the Tag Builders import tool in TrendHub.
  • Durations for context items: Filter and sort your ContextHub views on (total) duration of context items and get insights into the durations of separate events in context item details.
  • Current state filter: Filter your context views using current event states, for example; target all context of type maintenance where the current state is 'in progress' or 'inspection'.
  • Extended OSIsoft PI time series connectivity: As well as OLEDB connectivity, TrendMiner now offers connectivity to time series data from OSIsoft PI Data Archives using the PI AF SDK.
  • New OSIsoft PI Vision plugin: Start your analytics journey by opening your PI Vision display up in TrendMiner.
  • TrendMiner system metrics available: Get more insights into the resource health of the TrendMiner system within your own existing Prometheus monitoring system.

Application Enhancements

  • Robust loading of TrendHub views: TrendHub views which contain deleted or inaccessible tags will still load healthy and accessible tags and show a clear message highlighting which tags could not be loaded.
  • Auto-sharing of depending work-items: When sharing an item it is no longer required to share depending items separately, e.g. when you share a dashboard with someone, the TrendHub views and ContextHub views used on the dashboard will automatically be accessible.
  • Create copy of saved item: The work organizer provides the action 'Make a copy' which can be used to copy TrendHub and ContextHub views, fingerprints, formulas, aggregations and dashboards.
  • Split Gantt chart based on context fields: Rows in the Gantt chart can be split by context fields that are added to context items.
  • Gantt chart increased performance: Visualise months of data and keep things smooth.
  • Ordering context field dropdowns.
  • Context item workflows can now be edited and states can be ordered.
  • Context SQL connector adds ability to include context fields in the sync; the option to add context to tags; and support for defining the timezone of the external data source.
  • OSIsoft PI Asset Framework support for substitution parameters: added support for setups using substitution parameters to make dynamic templates.
  • Support for setups where multiple historian connections are set up to be used with the same OSIsoft PI AF asset structure.
  • Support for multiple connectors: Multiple connectors can be connected to optimize connectivity to your data sources and remove the single point of failure. Connector and data source configuration is now part of ConfigHub.
  • Better insights into connectivity: For all connectors and data sources the last sync status and where relevant, the reason for the sync failure, are shown making it easier to troubleshoot connectivity issues. Re-syncs can be manually triggered per connector or data source.
  • Cross asset value based search: Attribute template names are displayed when adding conditions instead of attribute names, which can be different (only when using the TrendMiner in-application asset framework). Attributes without a template name are filtered out.
  • Improved index manager screen (admin): The indexing manager screen has been reworked, providing a better overview of the health of the indexed tags.  
  • Improved indexing retry policy: Tags of which the indexation process did not complete due to eg. historian connection problems, will automatically be re-added to the indexing queue when the tags are accessed (charting or monitoring). The same mechanism will append historical data when the index horizon is increased. It is no longer needed to reindex all tags in this scenario.
  • Context fields's default permissions are now setup to include delete permissions.
  • Fingerprint performance:  The creation of fingerprints has been optimised to accommodate use cases with a lot of layers and a lot of hidden tags (>20).

Bug fixes

  • Last updated time of work items is now shown in local time instead of UTC.
  • Fixed an issue in cross-asset searches, where not all attributes of an asset were listed if the asset contained more than 20 attributes.  
  • Fixed an issue where tags which were hidden in the base layer, but visible in sub layers, were included in a newly created fingerprint.
  • Fixed an issue which prevented ACLs to be assigned to some users in setups with more than 1000 users.
  • Fixed an issue where searches could not find results before 2010, even though the horizon was set before this date. 
  • Fixed a visualisation issue where results of value zero were not displayed on the search results when performing a calculation on top of the search results. 
  • Performing an influence factor analysis with shift detection will no longer automatically apply the shift on the active tags. 
  • Potential delays on monitoring results based on formula tags are mitigated by automatically updating the index of all underlying tags.

Known issues

  • Tag grouping, chart  type selection and fingerprint position are not yet remembered as part of the view state and may be lost when refreshing the view, clicking the back button of the browser or loading a saved view. 
  • Heavy load on the new PI AF time series connector might result in connectivity issues to the Historian server.
  • After an update, it can be possible that DashHub and ContextHub don't work. A hard refresh of your browser will fix this issue.
  • Aggregations on top of discrete and digital tags might result in incorrect visualisation for recent periods due to prolongation of the underlying tags. The index values are not affected.

Important

  • Connector and data source configuration is now part of ConfigHub.
  • This release requires plant integration version 3.5.1189 or later.
  • TrendMiner can not be updated when there are duplicate states for string tags in the database. Contact support in case your update is stopped.
  • Starting this version the officially supported version for PI Event Frames is AF SDK version 2017.R2 or higher. To work with the version lower than 2017.R2 requires support from TrendMiner and will lack functionality and performance that comes with newer versions.

Synopsis functionality

Nested calculations

With the introduction of nested calculations it will become possible to; use formula tags as a variable in other formulas, create aggregations on top of other aggregations and create aggregations on top of a formula up to 2 levels deep and with a limit of 10 variables per formula. This way a formula can be created with a dependency of up to 100 tags. 

The ability to nest calculations also allows you to give more logical names and structures to your tags, improving the readability and reusability of your calculated tags.

Use nested calculations to remove noise from a signal which was created by summing up values after converting them to the same unit:
Reactor1-flow-sum = BA:Flow.1 * (1000/60) + BA:Flow.2 + BA:Flow.3 * 1000
Reactor1-flow-sum-no.noise = average(Reactor1-flow-sum)

As a next step to optimize work organization in TrendMiner, aggregations and formulas have been added to the work organizer in this release. You can easily filter out other items with the work organizer type filter to get a clear overview of all your aggregations and/or formulas. Organize aggregations and formulas

R1_1.png

The benefit of having aggregations and formulas in the work organizer is not limited to the option to organize them in specific project folders but also allows you to share the definitions with colleagues.

R1_2.png

Use this to give your colleagues better insight in how to use your calculated tags. This will increase their trust in your work and avoid duplicated calculated tags.


Import of discrete and string tags

Next to analog tags also discrete and string tags can be imported. Other tag import improvements are:

  • Option to delete imported tags from the overview panel.
  • Support for importing units of measure.
  • More than 100 tags can be imported per user (although the limit of 100 tags per file remains).
  • Timezone support.
  • Improved feedback: when a tag import fails, a specific error message indicates the reason of failure and the location of the failure in the .csv file.
  • Import future data: values with a timestamp in the future will become available over time.

R1_3.png

R1_4.png

 

Durations for context items

Like time series, all contextual data in TrendMiner is related to time. Events of context items can therefore represent durations. Durations can be used to easily display and sort on cycle times, searching for production steps or unit procedures that are quicker or longer than expected.

R1_6.png

As displayed on the image of the detail screen you get a Total duration field and durations of the separate events in the context item.

In the Context views you can display the duration column where you can sort and filter on duration however you like, with the usual operators <,>, =,...

R1_7.png

Current state filter

Filter your context views using current event states, for example; target all context of type maintenance where the current state is 'in progress' or 'inspection'. This Filter option is available in the filter menu under the name 'Current state'.

R1_8.png

Pi Vision plugin

The TrendMiner PI Vision plugin enables you to send your tags and attributes directly from your PI Vision display to your TrendHub app, ready for analysis. 

The advantage of this option is that there is now no need to look for your tags or attributes in the asset browser setup in TrendMiner. By clicking the TrendMiner symbol, you send them directly to TrendHub where they will be readily displayed for analysis in the same time frame of your Pi Vision display.

R1_9.png

Improved index manager screen (admin)

The indexing manager screen has been reworked, providing a better overview of the health of the indexed tags.

R1_10.png

The following application enhancements were introduced: 

  • Sorting on index start date and last updated date.
  • Filtering on index status.
  • Easier identification of indexing problems, by introducing multiple states:
R1_11.png Tag completely up to date. Backwards indexing was successful and recent data was successfully appended.
R1_12.png Backwards indexation process in progress.
Will switch to OK after completion.
R1_13.png Backwards indexing was performed but an error (e.g. historian connection loss) occurred during this process.The Index start date indicates where the backwards process stopped.
The indexation will automatically be retried.

Recent data was successfully appended
R1_14.png Recent data was not successfully appended, delaying potential monitoring events.

This state gets priority on incomplete state.
R1_15.png  TrendMiner is unable to update the index of the tag (tag or dependent tag no longer exists).If the root cause is resolved, the indexation process will automatically resume.
R1_16.png In the event that the responsible service for indexing goes down, tags which were in progress will fail. These tags should be reindexed afterwards to ensure their correctness. 

 

TrendMiner system metrics (system-admin)

Customers with an existing monitoring system in place can now integrate TrendMiner within it, as TrendMiner can now expose Prometheus metrics for visualisation on a Grafana Dashboard, as illustrated below.

System_metrics1.png

Notes:

  • Please contact TrendMiner support to enable this feature.
  • Sample dashboard is available on the download-site.