Generalized adaptive notch filters - does gradient smoothing technique help?

Generalized adaptive notch filters (GANF) are used for identification/tracking of quasi-periodically varying dynamic systems and can be considered extension, to the system case, of classical adaptive notch filters. We analyze the enhanced GANF algorithm, proposed in the literature, which incorporates gradient smoothing. We show, both analytically and by means of computer simulation, that gradient smoothing does not improve tracking performance of generalized adaptive notch filters.