Section 68 of 102
67. Sensor Evidence and Data Quality
Stable section ID: S05-CON-010-SECTION-68 · 32 content blocks
Sensor information shall be treated as engineering evidence with defined limitations rather than as inherently correct data.
Each significant sensor record should include:
- Sensor identity.
- Sensor type.
- Measurement units.
- Physical location.
- Measurement target.
- Calibration status.
- Accuracy or tolerance.
- Sampling method.
- Timestamp.
- Communication status.
- Power status.
- Environmental limitations.
- Data-processing history.
- Current health indication.
Data quality evaluation may consider:
- Completeness.
- Freshness.
- Accuracy.
- Precision.
- Consistency.
- Plausibility.
- Resolution.
- Calibration validity.
- Communication reliability.
- Redundancy.
- Agreement with related measurements.
- Suitability for the intended decision.
A missing or failed sensor shall not be interpreted as a normal physical condition. The absence of evidence shall remain distinguishable from evidence of absence.
Safety-critical determinations should use independent confirmation when required by the applicable risk and Engineering Profile. Confirmation may come from redundant sensors, deterministic physical relationships, inspection, testing, or qualified human judgment.
Processed, filtered, fused, or AI-interpreted sensor information shall remain traceable to the original measurements. Original evidence shall be preserved according to applicable retention requirements.