METHOD FOR AUTOMATIC IDENTIFICATION OF SPECTRAL COMPONENTS OF A SIGNAL

摘要

The invention relates to a method for automatic identification of the spectral components of a signal, comprising the following stages: application of a predetermined plurality N of distinct transformations to the signal in order to provide N signal spectra; automatic detection in each spectrum of all of the peaks which are not a priori associated with the noise spectrum present in the signal; association of a spectral map to each peak, said map containing a set of attributes including adjusted frequency, frequency interval, at least one adjustment error, false alarm probability; deduction of a spectral hypothesis such as pure frequency, double non-resolved frequency, multiple frequencies, narrow band; provision of a corresponding digital code; comparison in different spectra of the codes of the peaks belonging to a same frequency interval and indication of the most probable hypothesis.