Calamity Detection Using Nonparametric Statistics

Abstract

The problem of detecting a sudden change or calamity in the normal operation of a control system is investigated for the case of a deterministic system with measurements corrupted by unknown stationary noise. The basic technique concerns the use of nonparametric two-sample tests. As a consequence, the system under observation does not need to be known in detail; all that is required is a set of observations corresponding to the normal operation of the system.