A novel optimization based method for separation of periodic signals

Abstract

In the present paper a new method is proposed for separating the individual periodic components of a mixed signal. The method is capable to extract not only a harmonic but an anharmonic signal component from the mixture. To achieve this, the component is extracted by an FIR narrowband filter, which can modulate the output harmonic signal by an appropriate time-shift function. The search for this function is based on the minimization of a functional, which is calculated as the sum of the unsigned differences of the separated signal in a certain time window. The theoretical basis of this optimization method is that the functional above has global minimum if the separation is complete. The introduced tests show that the proposed method is more robust than the matrix algebraic separation (MAS) system in the case of a slightly frequency-modulated test signal.