Abstract: Aiming at the problem that the accuracy rate of the traditional sorting algorithms based on full pulse parameters is reduced in the complex electromagnetic environment of battlefields, this paper proposes a signal sorting method, which extracts the symmetrical Holder coefficients of the higher-order spectrums of the signals as the in-pulse features. This method first uses the symmetric Holder coefficient method to extract the intra-pulse features of the higher-order spectrums of the signals, and then combines the extracted intra-pulse feature parameters and the stable inter-pulse parameters into a new feature vector, and finally uses the K-means algorithm to sort the signals. As an intra-pulse feature, the symmetric Holder coefficients of high-order spectrum have a larger space for optimization than the resemblance coefficients of 1st-order spectrum. Adding this feature to the signal feature vector can make the new feature vector more separable. Simulation experiment results show that adding this feature and using the new feature vector can effectively improve the sorting accuracy of radar signals of different modulation types.