AN OPTIMIZATION OF THE IDENTIFICATION PROCESS

The paper presents a definition of the notion and a method for determining the measurability of the model of an identified object. The measurability was defined as the measure of the distance between the assumed reference model and the model to be determined with consideration of the properties of the identification system. The method enables the estimation and optimization of the influence of hardware and computing part of identification systems on the measurability of the model. The presented method of the measurability analysis can be applied to arbitrarily complex objects and identification systems provided that it is possible to construct sufficiently accurate models of them. As an example, the optimization of the system used to identification of the industrial aluminium electrolyser is discussed in detail. Results confirm the effectiveness of the presented method, as it enabled the reduction of the total error of the determined model by many times.