Analyze

HydroSentinel

Context-aware water anomaly detection for managed facilities.

Choose a synthetic/simulated scenario, set operating context, then review the resulting evidence with a human decision-maker.

Choose a scenario

Compare normal demand, abnormal behavior, and legitimate high activity using the existing seeded scenarios.

Starting analysis service…
Why these scenarios?

They provide a controlled comparison of normal operation, suspicious flow/pressure behavior, legitimate high demand, and high demand with a flagged pattern. They are seeded synthetic/simulated scenarios, not live building telemetry.

Operating context

Operating context helps distinguish legitimate high-demand periods from suspicious behavior.

Why does operating context matter?

A higher flow rate during an event can be expected. The model evaluates flow and pressure alongside the selected operating context rather than treating high demand alone as a leak.

Starting the analysis service

This demo runs on a free backend that sleeps when inactive. Starting it again can take around a minute. Keep this page open. Analysis will become available automatically when the service is ready.

What happens when I click Analyze?

HydroSentinel evaluates the selected flow, pressure, and operating-context telemetry against the learned synthetic baseline, then returns a non-persistent result for human review.

Analysis status

Awaiting analysis

Choose a scenario to compare normal demand, suspected leak behavior, and legitimate high-activity demand.

Model score
—
Not a calibrated probability
Estimated loss rate
—
Operating context
—
Pattern label
—
What does model score mean?

It is the classifier score for the detected pattern in this synthetic/simulated scenario. It is not a calibrated probability or a confirmation that a physical leak exists.

System transparency
  • Data source: seeded synthetic/simulated scenarios.
  • Inputs: flow, pressure, and operating context.
  • Model score is not a calibrated probability.
  • Estimated loss is a model estimate.
  • Not physically deployed or validated on building infrastructure.
  • Human review is required.
What can and can’t the system conclude?

It can surface patterns that differ from its learned synthetic baseline and provide decision-support estimates. It cannot confirm a physical leak, replace site inspection, or guarantee outcomes.