Генерація нестаціонарних імітаційних процесів авторегресії

Abstract

The problem of obtaining non-stationary, correlated time series is considered. The use of an additive mixture of sinusoids with white noise to obtain them has limited practical value. The proposed method includes a stationary autoregressive model and a special case of a non-stationary autoregressive model - an integrated moving average. Formulas are given that make it possible to methodologically use the proposed model.