Frequency domain identification of processes with output drift

The identification of linear system dynamics from measurements corrupted by low frequency aperiodic noise is discussed. It is shown that by considering the low frequency harmonic content of the disturbing noise, accurate elimination of the disturbance can be achieved for typical linear/nonlinear trend characteristics. Methods are discussed for dealing with the transient response of systems to periodic input signals when the identification time is an important issue.<
>