Discover*e

The Core Technology

Discover*e is Pattern Discovery Technologies' analytics engine, built on technology developed with the University of Waterloo. It uses pattern recognition algorithms to isolate complex interdependencies in industrial process data and surface meaningful relationships. The engine combines industry-standard methods (clustering, regression, decision trees) with the company's own patented techniques, including high-order pattern discovery and temporal pattern discovery across multiple time series. It is designed to be embedded easily into end-user applications tailored to a specific environment 

How this applies in CompressionInsight

CompressionInsight is an automated tag-tuning and tag-behaviour identification tool for the Aveva/OSIsoft PI System. It is built around an information measure the company calls "Data Fidelity." The workflow has four steps:

1. The software takes the raw, uncompressed snapshot feed going into the PI server and compares it with the archived values as they would be recreated under the current settings. 
2. Using information-theory principles, it calculates how much of the raw signal's information survives in the archive and expresses that as a percentage.
3. It then recommends revised exception and compression parameters (ExcDev and CompDev) that balance fidelity against storage. This flags both over-compressed tags, which lose too many points, and under-compressed tags, which keep too many. 
4. Thousands of tags can be analysed at once, and changes can be written back to PI automatically or shared as reports for offline updating through PI DataLink.

 Why it matters. The product replaces static rules of thumb, default settings and copy-paste configuration with an unbiased approach driven by each signal's actual behaviour. The result is that downstream analysis can rely on archived data that faithfully represents the original signal.