An output error model and algorithm for electromagnetic system identification (Q1101051)

For a finite-order, linear, time-invariant system the problems of fitting complex exponentials to data and of modeling the input-output properties are two problems of automatic control. In this paper, the authors dicuss difference equation models whose parameters, the poles and residues of a complex exponential model, can be calculated. The discussion of a possible choice for a model is presented, too. After that, the authors give an algorithm, a nonlinear least squares one, which is used to estimate the parameters. The actual implementation of the algorithm is described. Examples with synthetic transient electromagnetic data are such that a random number generator is added to the data. After some interesting conclusions, the authors finably caution the potential user of complex modeling in electromagnetic scattering, concerning the role of the entire function and the need for a model that correctly represents the behavior of the system.