Equivalent to TagTot using multiplication, for example, to calculate a geometric mean
TagTot will provide the sum of event values for a given time range. Customers wishing to calculate the geometric mean could use an equivalent function that multiplies each value together.
Additionally, a specific geometric mean function may be useful as well. (this would include taking the result to the power of (1/n) where n is the number of events in the calculation.
Potential caveats include both floating point precision issues, as well as overflowing a double data type. The calculation may also be quite complicated if the input tag has very frequent events.
A possible workaround at this time would be to import the data into Excel using DataLink and use the built in GEOMEAN function. Other options include custom applications or ACE calculations.
ADMIN RESPONSE
Aug 20, 2022
The current release of AF has the ability to calculate event based and time based averages. At this time, we have no plans to add a new function for geometric mean.
Another workaround:
Suppose that your data tag is A. Create an event-triggered expression analysis that outputs ln(A) to tag B. Then use TagMean(B, start, end) * EventCount(B, start, end) to sum the raw, uninterpolated values of B in between the "start" and "end" times. Store this in variable C. In variable D, calculate e^C to get the product of the raw, uninterpolated values of A between the start and the end times.
"ln" is called "Log" and "e^" is called "Exp" in PI System Explorer. We take advantage of these properties of logarithms:
ln(a) + ln(b) + … = ln(a * b * …)
e^ln(a * b * …) = a * b * …
There is no function that sums the raw, uninterpolated values of a tag, but there is a function (TagMean) that averages the raw, uninterpolated values of a tag. Multiplying by TagMean by EventCount gets us our sum, assuming that none of your events are system digital states.
