Abstract:
The paper discusses the identification problem for the case where the input signal is known accurately and the output signal is recorded by a system of $p$ sensors. The errors of each sensor are not correlated and have normal distribution. The variances of the sensors and their correlations are not known. Different estimates of the plant unknown variable vector are proposed that are based on the Gauss — Markov and Stein — James estimates and are used for studying the r.m.s. plant identification error.
Fulltext:
PDF file (2407 kB)