抄録
Realizing the importance of identification, many identification methods have been developed, among them the statistical methods are known to be especially preferable not only for the identification of a time invariant plant, but also for the on-line identification of a time varying process in adaptive or optimum control system.
In this paper, traditional statistical methods are classified into two categories, i. e., the methods to express an unknown plant (black box) using known elements (white box) and the methods to determine unknown parameter contained in elements whose forms are known (grey box). Some considerations on the mathematical properties of these methods are presented.
After considering the shortcomings of these methos, a new identification method which can determine the order of a plant and give the optimum pole location is proposed. In this method, the orthonormal function with varying parameters is used, and this method is derived from the idea of the correlation coefficient.
Some illustrating examples are also presented.
