ON LINE CHANGE DETECTION WITH NUISANCE PARAMETERS

Dealing with nuisance parameters is an important issue in on-line monitoring safety-critical complex systems and detecting events/changes that affect their functioning. A linear stochastic-dynamical system with deterministic nuisance parameters and additive changes is considered in the paper. The problem of quickest change detection with nuisance parameters is solved. The peculiarity of the deterministic nuisance parameter rejection in the considered case is the fact that the rejection operator is a function of the size of sliding observation window which is used to decide between two hypotheses (no fault and there is a fault within a given time window). In the case of sequential change detection this window varies that leads to a new and difficult problem of change detection with an unknown dynamic profile of changes.