Research On Blind Separation Methods And Applications Of Complex Mixed Acoustic Signals

In complex environment, the signals measured by acoustic signal sensors were mixed with several signals, and the useful acoustic signals are often mixed in noisy signals and can not be identified clearly, there's a lot of information in acoustic signals. The separation of mixed acoustic signals in complex environment is important for theory and application.For complex mixed acoustic signals, blind source separation methods and applications are researched. The main contents of this paper are summarized as follows:(1) The research significance, history and research status of the blind signal separation are described in this paper. The math modal and theory of blind source separation are introduced in detail and some typical blind source separation algorithms are classified in this paper.(2) In view of the deeply study in the bigradient blind source separation algorithm based on kurtosis, the bigradient blind source separate algorism based on negentropy is proposed. The algorism uses stronger robust negative entropy as criterion function. This algorithm is not sensitive to wild value like kurtosis. Simulation result shows that the algorithm has a good precision and speed in separation.(3) The blind separation of correlated sources using typical algorithm is uncertain according of the theory and simulation analysis. The research situation of blind separation for correlated sources is summarized. A wavelet packet filter sub-band blind separation is applied. When the source signals are correlated, this method improves the performance of blind signal separation and verified by simulation.(4) The acoustic signals in cabin are various and complex, including speech signals, non-speech signals, and engine noisy signals. In this paper, a method is proposed that the blind source separation can be used in the separation of mixed acoustic signals in cabin. Based on the measured acoustic signals, the algorithms are obtained through program. The fittest algorism in acoustic signals processing is selected through comparison. The algorithm applied in this paper is feasible through the identification and separation of source signals in cabin was realized by this method.