Degradation Detection in a Redundant Sensor Architecture.

Area of Science:

• Automation and Control Systems
• Reliability Engineering
• Sensor Technology

Background:

• Safety-critical automation relies on redundant systems for dependable operation.
• The one-out-of-two (1oo2) sensor architecture uses redundant sensors to ensure reliability and traceability of readings.

Purpose of the Study:

• To propose a novel fault prognosis algorithm for 1oo2 sensor systems based on discrepancy signal analysis.
• To analyze discrepancy changes caused by degradation processes in 1oo2 sensor configurations.

Main Methods:

• Analysis of discrepancy signal changes in 1oo2 sensor configurations using publicly available databases.
• Application and evaluation of two trend detection methods for discrepancy data.
• Training and comparison of several models to describe discrepancy dynamics.
• Prediction of future discrepancy behavior to identify critical failure points.

Main Results:

• Discrepancy between redundant sensors exhibits changes (increase or decrease) over time due to degradation.
• Trend detection methods effectively identify increases or decreases in discrepancy data.
• Specific models were identified as best fitting for describing discrepancy dynamics.
• The developed models successfully predict future discrepancy behavior and potential failure times.

Conclusions:

• The proposed fault prognosis algorithm can predict sensor failure dates based on discrepancy signal analysis.
• Predictive maintenance scheduling can be implemented to prevent system entry into a safe state or shutdown.
• This approach enhances the reliability and operational continuity of safety-critical automated systems.