While our maintenance team was servicing a sensor that I'm monitoring with the Tag Historian, the sensor recorded some unusually high and low values. These extreme spikes are causing inaccuracies when I review the historian for high and low readings during that timeframe.
Is there an effective method to identify and exclude these data points from the SQL database? It would be preferable to have no data during this period rather than bad data.
The proper method would be to disable the tag before servicing and re-enable afterward. You may have to poke at the raw data after the fact.
Tag history is not intended to be removed. The only way to remove this data is to do so via a query or your DBMS. Either will require a decent understanding of the historical tables and their relationships.
You could possibly mark the points as bad quality, and then "Ignore Bad Quality" in your chart.
There isn't any built-in way to filter out erroneous data points that aren't marked as bad quality data, the historian system does not know to treat them as such, and so they are always treated as good data.
In the future, for planned maintenance consider disabling history until the maintenance has been completed.
Otherwise, if this type of thing will happen often in your environment and you still need to record valid data then there are ways to still use the historical data tables while filtering the data points that are actually allowed to be collected.
Thanks. Sounds like I'll need to get with our Maintenance and have them communicate work items that will effect sensors or PLCs so I can disable tags if need be.
