Calibration and Estimation in Multiply Redundant Measurement Systems

For multiply redundant measurements, a calibration and estimation technique has been developed in the framework of a fault detection and isolation (FDI) methodology where the decisions are made on the basis of consistencies among all measurements. The consistent measurements are calibrated on-line such that their residuals are minimized. An estimate of the measured variable is obtained as a weighted average of the calibrated measurements where the individual weights are updated on-line on the basis of the respective a posteriori probabilities of failure instead of being a priori fixed. The calibration and estimation algorithm has been computer-coded for real time applications and verified by on-line demonstrations in an operating nuclear reactor.