THE DISORDER PROBLEM FOR STOCHASTIC JUMP PROCESSES
G. I. Salov
Institute of Computational Mathematics and Mathematical Geophysics, Siberian Branch, Russian Academy of Sciences, Novosibirsk, Russia, E-mail: sgi@ooi.sscc.ru
Pages: 200-210
Abstract
A Bayes formulation for the problem of rapid detecting appeared disorders in the probability characteristics of the observed stochastic jump process with values in a sufficiently general phase space is presented. It is supposed that the time of disorder appearance does not depend on the evolution of the observed process and coincides with the moment of one of the process jumps. Recurrent equations for a posteriori disorder probability are obtained. The lower bound for the optimal moment of alarming the happened disorder for minimizing the sum of false alarm probability (without disorder) and the mean time of alarm delay is found.
|
