基于特征模式分解的水声目标特征提取方法

This work aims to address the classification and identification problem of underwater targets in complex underwater acoustic environments and proposes a novel feature-mode decomposition method for the weak feature extraction of underwater acoustic targets. In this method, correlative kurtosis is used as the optimal decomposition parameter of the optimization target to realize the optimal decomposition of the original underwater acoustic signals. Then, modes are fused in accordance with the similarity of subsignals to enhance feature expression, thus realizing the accurate recognition of underwater acoustic targets in complex environments. Sea experiments show that the accuracy of underwater acoustic target recognition by the proposed method reaches 90. 1%, which is 12. 5% higher than that of underwater acoustic target recognition by traditional methods.