Add Covariance and Max-Covariance functions to Asset Analytics
Currently PI ACE calculations are used to verify if the data from redundant sensors is trustworthy. The PI ACE module implements Covariance and Max-Covariance calculations. The customer would like to move from PI ACE to Asset Analytics and is facing the issue that Asset Analytics is corrently not offering any extensibility nor Covariance / Max-Covariance functions. The Attachment describes the desired functions based on the PI ACE code.
The function description is attached with "Correlation Analysis Use Case_public.pdf".
Things we clarified / detailed more are:
- PI Data Archive doesn't support Array data types and there's not much sense in doing that. Hence inputs for both functions would be 2 numerical data streams.
- Function definition assumes the event count for both inputs to be equal which can be problematic with PI Points because of Exception and Compression. Hence functionality was detailed to make the first input the master, retrieve archived data (inside time boundaries) for the master, and use timestamps to get timed (interpolated) data for the second input.
- Both inputs (tag1 and tag2) must have 100% good data for the specified period.
- In case of a failed calculation the function is supposed to return “Calc. Failed”.
Covariance function is in PI Server 2017 R2, available now on https://techsupport.osisoft.com/