An integrated approach for sensor failure detection and signal reconstruction

The objective of this study is the development of a signal valida tion system utilizing redundant arrays of sensors. The signal validation system must be able to detect, isolate, and identify sensor degradations as well as provide a reconstruction of the measurement.

This was accomplished by combining the Generalized Consistency Check and the modified Sequential Probability Ratio Test in a decision estimator module. The Generalized Consistency Check is a computationally efficient management system for redundant sensors, while the Sequential Probability Ratio Test provides the ability to make decisions based on the degradation history of a sensor. The methodology was also extended to the detection of noise degradations by separating the high and the low frequency components of signals.

To verify the signal validation system an experimental air flow loop was constructed. Signals containing bias and noise type degrada tions were recorded.

The signal validation system was able to detect and isolate bias and noise type failures under steady state and transient conditions. An estimate of the measured variable was successfully provided for all the failure situations considered.