Methodology
How results are produced
This page explains evaluation assumptions and how pilots are verified in live environments.
Data Basis
Evaluation datasets
- Public simulation datasets (for example, CMAPSS variants) are used for controlled evaluation.
- Evaluation focuses on lead-time behavior and structural drift signals before failure labels.
- Results are reported as evaluation indicators and should be verified in your environment.
Assumptions
What the numbers mean
- Lead-time values are measured in dataset-native units (for example, cycles).
- Detection metrics depend on baseline windows, persistence gates, and drift thresholds.
- Dataset operating conditions may not match your exact environment or failure profile.
Pilot Verification
How pilot value is measured
- Scope to one target system with documented operator risk priorities.
- Track early-warning findings against known incidents/events in your environment.
- Report measured warning windows, confidence indicators, and actionability outcomes.