Identifying linear systems via frequency response

In performing an analytical investigation of an automatic control system, the estimation of a probabilistic model describing the unknown control system is often necessary. Therefore, two digital computer-oriented techniques are developed to obtain the approximate parameters of the model from an experimental magnitude/frequency response data. The methods are developed on the basis of the decomposition of three Cauer forms of continued fraction expansion. The most significant quotients of the continued fraction expansion are systematically arranged for which they can be identified first; thus, a satisfactory accurate and reliable result can be obtained. The methods have been converted into digital computer programs. Several examples are included.Electrical and Computer Engineering, Department o