The invention discloses a
feature extraction and identification method for a
radio frequency signal of an unmanned aerial vehicle, and belongs to the technical field of
radio signal monitoring and
signal identification. The method comprises the following steps: constructing a
cyclic autocorrelation function for an original
radio frequency signal, and extracting a characteristic
cyclic frequency; enhancing the original
radio frequency signal based on the characteristic
cycle frequency, and obtaining a preliminary effective signal in combination with a cycle autocorrelation function; extracting an
instantaneous phase of the initial effective signal, and obtaining a
direct path effective signal according to a phase second-order variable quantity; phase and amplitude of the
analytic signal of the effective signal of the
direct path are extracted, modulation intensity of the two signals is obtained, and a modulation
coupling function is constructed; extracting modulation
instability from the modulation
coupling function, extracting cyclic rigidity through cyclic mapping, and calculating cyclic
instability; and finally, weighting the modulation
instability and the cyclic instability to obtain an interference confidence coefficient, and when the interference confidence coefficient is smaller than a preset confidence coefficient threshold value, determining that the signal is an unmanned aerial vehicle signal. The identification precision of the
radio frequency signal of the unmanned aerial vehicle is improved.