Research on feature extraction method of ship radiated noise with K-nearest neighbor mutual information variational mode decomposition, neural network estimation time entropy and self-organizing map neural network

Complicated and changeable ocean environment causes the particularity of ocean background noise. Therefore, feature extraction of ship radiated noise is an urgent problem in signal processing. To solve this problem, a feature extraction method of ship radiated noise with K-nearest neighbor mutual information variational mode decomposition (KNN-MIVMD), neural network estimation time entropy (NNetEn) and self-organizing map (SOM) neural network is proposed. To overcome the problem of choosing K value in variational mode decomposition, KNN-MIVMD is proposed. To begin with, decompose ship radiated noise into intrinsic mode function (IMF) by KNN-MIVMD. Then, separately calculate the difference of NNetEn between each IMF and ship radiated noise. Next, select IMF with the smallest difference in NNetEn, and calculate its NNetEn as the feature vector. Lastly, classify and recognize the feature vector by SOM. Combining KNN-MIVMD and NNetEn to compare EMD, EEMD, VMD and MDE, MPE, FDE respectively, the results show that the proposed method has excellent decomposition effect and higher classification result.