Identifiability of linear parametric stochastic systems. II: Identifiability conditions (Q1088636)

The identifiability of parametric stationary stochastic systems, described by linear difference equations, is analyzed. With the aid of identifiability equations various structures of univariable and multivariable as well as open-loop and closed-loop systems are studied. The assertions concerning unconditional identifiability of system elements as well as linear dependence of variables are formulated. The methods of elimination of nonidentifiability are also investigated. The author asserts that the identifiability equations constitute a convenient tool for investigation of the identifiability of stochastic systems. This is due to their generality and simplicity of analysis and also to the fact that the results of utilization of identification methods with a quadratic criterion are the same. The described second part of the paper is closely related to the first part [see the author, Avtom. Telemekh. 1985, No.5, 64-78 (1985; Zbl 0601.93058)] and is not legible, separately. The author uses the results of the first part very frequently without citing them.