| The most widespread approach to optimal prediction of discrete-time systems relies on prediction error methods (PEMs), for which a bulk of theoretical results is available. In this thesis research, we consider the problem of identifying the transfer function of discrete linear systems using frequency domain approach. An ARX model structure is used to represent known systemmodels, and their approximate empirical transfer function estimates are generated using randominput sequence. The main objective is to recover the original system models using the empirical transfer function estimates as major input to our main MATLAB®coded program. The effectiveness of this method is illustrated by comparing, with no bias, the test problems’ results with those from MATLAB®system identification toolbox.Key Words: Empirical transfer function estimates(ETFE), ratdisk, ARX model. |