Product Updates: Windowed Analytics for Trend Baselines

A single bad reading rarely means anything on its own. A pump vibrating at 4.2 mm/s for one scan interval could be a sensor hiccup; a pump holding at 4.2 mm/s for the last six hours is a bearing that needs attention. The difference is context, and building that context by hand, in a spreadsheet or a one-off script, is exactly the kind of work condition monitoring is supposed to remove.

Windowed Transform Functions

ExoSense’s data pipeline now includes windowed transform functions: Count, Minimum, Maximum, Sum, Average, and Standard Deviation, each computed over a rolling time interval you set - 15 minutes, an hour, a full shift, whatever matches the equipment. Point one at a vibration signal and get a rolling standard deviation. Point it at a flow meter and get an hourly average plus a count of readings, useful for catching a sensor that’s gone quiet.

Instead of alerting on a single vibration spike (which might just be a truck passing outside), a rule can now trigger off a rolling standard deviation that stays elevated for 30 minutes straight - a much better proxy for a bearing or imbalance problem actually developing.

Because these run as ordinary Transform functions, they feed straight into Threshold Rules, dashboard panels, or downstream Insight functions the same way any other signal does - no separate analytics layer to stand up.

Full release notes: 2025 archive, v2.46.22. Transforms documentation: Transforms - Exosite Documentation.

Want to see rolling windows applied to your own equipment data? Schedule a Demo.