• State Not Answered
  • Replies 3 replies
  • Subscribers 3 subscribers
  • Views 488 views
  • Users 0 members are here
Options
Related

PI Explorer - Daily Average Analysis Tag

Mark Fitzowich

I'm wondering what the best way to calculate the daily average for an analysis tag in PI Explorer is. I want the tag to get a calculated average for each day for a vibration value:

 

Example:

July 22 - 0.04ips

July 21 - 0.02ips

July 20 - 0.03ips

 

There are a few issues I am running into that I want the analysis to handle:

  • Handles outliers.
  • Returns only one value per day.

 

This is my current logic:

First, calculates an hourly average value, while ensuring that outliers are removed.


image.png
Secondly, calculates a daily average value - based on hourly average values, while ensuring that outliers are removed.


image_1.png
 

I have the daily average value set to trigger once a day, yet when I trend it, there are many values each day.


image_2.png
This is not ideal because when I trend it in PI Vision, all the values change as the timeframe changes, it isn't a set in stone value for a single day.

What logic do I have to add in or remove to achieve the desired result?

Parents
  • 0

    Hi ​,

     

    Thank you, this solution worked, and now I am getting a daily average value. Could you review the logic that I have set up? Is there an alternative, more robust way of calculating daily average? The reason I have the 'NegativeFilter' and 'OutlierFilter' variables set up at 0 and 1 is because the High Alarm and High High Shutdown for this tag are 0.3ips and 0.5ips, respectively. Therefore, anything outside the range of 0 and 1 is likely to be an anomaly or noise. Is there a better way of handling anomalies/noise in PI AF?

Reply
  • 0

    Hi ​,

     

    Thank you, this solution worked, and now I am getting a daily average value. Could you review the logic that I have set up? Is there an alternative, more robust way of calculating daily average? The reason I have the 'NegativeFilter' and 'OutlierFilter' variables set up at 0 and 1 is because the High Alarm and High High Shutdown for this tag are 0.3ips and 0.5ips, respectively. Therefore, anything outside the range of 0 and 1 is likely to be an anomaly or noise. Is there a better way of handling anomalies/noise in PI AF?

Children
No Data