Glossary · Verification and validation
Sensor drift
German: Sensordrift
In measurement and control, sensor drift is a gradual change in sensor output relative to the measured quantity. It can shift switching thresholds and measured values without triggering any fault message.
- Verification
- Data and interfaces
In one sentence
Sensor drift is a slow change in sensor output against the true value, which can shift safety thresholds unnoticed.
Example
A pressure sensor that monitors the clamping pressure of a CNC fixture slowly reads higher than the true value, so the controller allows machining with a clamping force below the safe limit.
How it applies
- Functional safety: Drift is a dangerous failure mode when it moves a value toward the unsafe side. Redundant sensors with comparison or plausibility checks can detect it.
- Data and interfaces: Downstream systems receive drifted values as valid data. Quality flags, calibration dates and sensor data integrity measures help consumers judge whether values are trustworthy.
- Technical documentation: State recalibration intervals and the symptoms of drift in the maintenance manual.
- AI and retrieval: Models trained on sensor data can learn drifted values as normal. Monitoring input distributions helps separate sensor drift from real process changes.
Sensor drift vs. calibration drift
Sensor drift describes the change of an operating sensor's output. Calibration drift describes the loss of calibration state of measuring equipment in general. In practice, sensor drift is often discovered as calibration drift at the next check.