The invention provides a two-stage integrated modulation identification method based on relevant clusters. The two-stage integrated modulation identification method based on the relevant clusters aims at achieving correct identification of six modulation methods, namely binary phase shift keying (BPSK), binary frequency shift keying (2FSK), binary amplitude-shift keying (2ASK), quaternary phase shift keying (QPSK), 4FSK and 4ASK. The feature extraction method comprises the steps of carrying out seven-layer decomposition of six signals by a db3wavelet, reconstructing the signals of each layer, calculating difference of squares (S) of the signals of each layer, obtaining eight parameters from S1 to S8, and meanwhile introducing four instantaneous feature parameters, namely sigma ap, c2, sigma dp and M, so that the three signals of the 2PSK, the QPSK and the 2ASK are identified in a better mode. In order to overcome the defect of a hidden terminal, the two-stage integrated modulation identification method based on receiving signal relevance for clustering is adopted. According to the method, the clustering is carries out based on receiving signal energy, nodes inside the clusters extract various parameters for cooperation according to a received signal-to-noise ratio, and each node only extracts one feature parameter. According to the extraction method, primary feature integration is carried out on cluster heads, and the cluster heads are converged to the center for decision integration.