Точность процедур фильтрации, экстраполяции и интерполяции случайных процессов

Abstract

The accuracy of filtering estimation, extrapolation and interpolation of random processes has been analyzed. It is shown that the accuracy of extrapolation decreases with increasing of prediction interval. It is argued that the extrapolation is meaningful only within the interval, which does not exceed the correlation interval. It is shown that due to the interpolation of the sequence of the recursive point estimates, the accuracy of estimated realization recovery of the random process exceeds the estimates obtained by Kalman-Bucy.