Title :
A new method for studying some statistical failure detection algorithms
Author :
De Cambry, Olivier
Author_Institution :
Groupe ESIEE, Noisy le Grand, France
fDate :
1/1/1993 12:00:00 AM
Abstract :
A class of dynamical algorithms for detecting a change of parameter in a statistical model is described. Using large deviation results, the author approximates the expected time between false alarms. The expected time delay for failure detection is approximated. These approximations provide a criterion to measure the delay for detection, which is easily computable within the whole class
Keywords :
statistical analysis; expected time delay; false alarms; parameter change detection; statistical failure detection algorithms; Change detection algorithms; Delay effects; Detection algorithms; Detectors; Fluctuations; Heuristic algorithms; Random variables; Statistical analysis; Stochastic processes; Testing;
Journal_Title :
Automatic Control, IEEE Transactions on
