A system of equations is obtained for finding Bayesian estimates of the parameters of many-dimensional linear Markov processes when only certain of the components are observed. The eequations enable the authors simultaneously to carry out filtration of the unobservable components in the case of unknown process parameters. Estimation of the unknown parameters of the correlation function of Gaussian processes with rational spectral densities reduces to this problem.
|Title of host publication||Eng Cybern|
|Number of pages||10|
|Publication status||Published - Nov 1971|
ASJC Scopus subject areas