Disorder indicator for nonstationary stochastic processes
Alexey Alexeevich Kislitsyn, Antonina Borisovna Kozlova, Marina Borisovna Korsakova, Yu. N. Orlov · Доклады Академии наук · 2019
The properties of a statistic called a self-consistent stationarity level of nonstationary time series are examined in this work. It is shown that a change in this statistic can be treated as a disorder in the nonstationarity properties of the series. The significance level of decision making is estimated, characteristic periods in a non-stationary stochastic process are detected, and an optimal sample size for constructing indicators in stochastic control problems is determined. A disorder indicator for electroencephalogram data from epilepsy patients is studied as a practical application.