Abstract
The method of the estimation of the dynamic characteristics of a process by using the least square error criterion from the normal operating data and some estimation results of actual process by this method are reported in previous papers. Through a series of these practical approaches, it is recognized that the application of suitable techniques for data handling is no less important than the theoretical background of estimation procedures.
In this paper, several problems accompanied with data handling are pointed out, and some techniques to solve these problems are explained. The idea implied there is to introduce the filter with some specified characteristics in order to eliminate the unnecessary noise component in data. It is shown that using these techniques makes it possible to shorten the duration of recording, decrease the number of parameters to be estimated, and eliminate the drift component.
