Radar radiation source signal recognition method based on deep learning network

The invention belongs to the technical field of recognition of radar radiation source signals in electronic countermeasures, and discloses a radar radiation source signal recognition method based on adeep learning network. Firstly, received radar radiation source signals are subjected to frequency reduction preprocessing; a cross-ambiguity function based on linear canonical transform and a cross-ambiguity function based on a linear canonical domain of the received signals are calculated respectively, and the maximum cross section value of the cross-ambiguity function is extracted respectivelyas a feature sample set; and finally, through a sparse filtering-based capsule network, classification and recognition are carried out. Through extracting features of the linear canonical transform-based cross-ambiguity function of the signals, the uncertainty of the feature parameters in a data range in a complex electromagnetic wave environment can be overcome, and the recognition credibility is improved; and hierarchic