Frequency hopping modulation mode identification method

The invention relates to a frequency hopping modulation mode identification method, which comprises the following steps of: 1, receiving an FH signal, performing data preprocessing on the FH signal, and extracting an FH signal time-frequency energy diagram by using an SPWVD time-frequency transformation algorithm; removing noise of the time-frequency energy diagram by adopting self-adaptive Wiener filtering and morphological filtering algorithms; converting the frequency energy diagram into a time-frequency energy grey-scale map; step 2, feature extraction: extracting modulation identification feature parameters of three types of features including geometric invariant moment, pseudo Zernike moment and Rayleigh entropy of the time-frequency energy grey-scale map as a feature set of FH modulation identification; and step 3, feature selection and identification: selecting an optimal feature set and identifying a multi-type FH modulation mode through a reinforcement learning feature selection algorithm and a supp