Multi-Level Hypotheses Testing for Fault Detection in Continuous Process

This paper presents the analytical development of a multi-level hypotheses testing algorithm and an extension of the sequential bi-level test procedure reported earlier [1]. The multi-level test procedure characterizes M distinct categories of faults in continuous processes like aircraft, spacecraft, and nuclear power plants, and is suitable for generating signals for early warning and failure. The test procedure has been implemented on a commercially available microcomputer for on-line detection of sensor faults in an operating nuclear reactor.