On pole-zero modeling of linear systeman improvement to ITIF algorithm

Resumen

Here is presented an algorithm which allows the determination of transfer function of a linear system. This algorithm is based upon a non simultananeous estimation on their poles and zeros. It works directely on the system output and it can be applied in situation where impluse train excitation is supposed to exit. The poles are estimated after linear prediction is applied to a segment of the output in which this signal is "predictable" from itself. One that estimation has been obtained, the zeros are estimated by applying the ITIF (interactive Inverse Filtering) algorithm with a suitable modification. The method proposed in this paper have been experimentally compared with the ITIF algorithm. By the experiments it has been shown that this method can estimate both the center frecuencies and bandwidths of poles and zeros more accurately that the ITIF.