TASK ORIENTED IDENTIFICATION OF THE PROCESS MODEL

Abstract

In the paper there is considered a problem of the identification of the process model applying criterions fitted to the defined tasks of the designed control systems. After a linear, nonsingular transformation of model inputs there is derived an equivalent form of the process model. The least square estimators of the coefficients for this model are in form of stochastic independent processes, what simplified an analysis of the identification results. Applying introduced form of the identifications errors there is estimated an impact of these errors at the one-step prediction model and in cases of the models for closed-loop systems for the steadystate control and the minimal variance control. The derived estimators and simulated results are compared, in a numerical example of the third order, linear single input- single output process. The derived formulas can be used for determination of a proper value of a sampling time in the discrete-time control system.