Abstract: A new method of applying blind signal processing is presented to identify noise source quantitatively of same frequency band. Blind signal processing is a method to recover source signals by only utilizing observed signals on the assumption that the source signals are mutually independent. Blind signal processing can only be used to identify noise source qualitatively because of the inherent indeterminacy of amplitude and order. Without considering the order indeterminacy, a new method is presented to eliminate the effects of amplitude indeterminacy in noise source identification by using the adaptive noise characteristic of cancelling out the estimating transfer function adaptively. The feasibility of this method is confirmed theoretically, and then testified by tank experiment. The method proposed can also be used to solve the problem of reference inputs coupling in adaptive noise cancellation, and improve the analysis accuracy.

Key words: Adaptive noise cancellation, Blind signal processing, Decoupling, Source identification