基于不同能量值特征优选的水声目标识别

采用Bartlet平均周期图、小波变换和经验模态分解方法分别提取信号的频段能量值特征、IMF能量值特征，重点对IMF能量值特征进行特征优选。通过设计BP神经网络分类器，对实测的四类舰船目标的辐射噪声信号进行测试，取得了较好的识别效果。

Underwater Acoustic Target Recognition Based on Feature Selection of Different Energy Feature

Bartlet average periodogram, wavelet transform and empirical mode decomposition method were used to extract the signal?s frequency band energy and intrinsic mode function energy feature, the feature selection of IMF energy is the key point. Then it designs the BP neural network classifier for testing the radiated noise signals of four kinds of skip targets, and gets a good recognition effect.