Distinguishing sensor faults from system faults by utilizing minimum sensor redundancy

Abstract

Automated Fault Detection and Diagnosis (FDD) systems depend entirely on the reliability of sensor readings. This paper fills an important gap in the literature by pinpointing the distinction between sensor faults and system faults in the monitoring process. The proposed methodology determines the minimum degree of sensor redundancy necessary to achieve this. A priori knowledge of physical relationships between monitored variables is used to check the credibility of sensor observations. The generalization reveals that for serially connected systems if the number of sensors is greater than 1.5 times of the number of monitored variables, the task of distinguishing between sensor and system faults can be accomplished with certainty, as long as serial causality is valid between the monitored variables. This is verified using a system of interconnected multi reservoirs and control valves.