Adaptive Structural and Parametric Identification of Dynamic Objects for Real Control Systems

The accurate structural and parametric identification of dynamics objects is a required step in controller design. In most cases, the only source of original data for identification is the experiment. Very often, the structural scheme of the object of control is known in advance, and the task of identification is to find out the values of the transfer function coefficients. This paper shows that when dealing with objects of fourth and higher order in real-world conditions with analogue noise and ADC discretization, the effect of order reduction appears. That fourth-order object appears to be a third-order object. According to the study's findings, the discrete modal controller based on the third-order model has acceptable transient performance with the fourth-order object. That means that before parametric identification, structural identification based upon the dependency of the covariance matrix trace against the sampling period must be carried out to allow for the determination of an order of the model suitable for controller realization