基于熵评价模态分解的雷达信号双谱特征识别

针对在低信噪比下雷达信号调制识别准确率低、抗噪性差的问题, 提出一种基于熵评价模态分解和双谱特征提取的识别方法。利用双谱可以抑制高斯噪声的特点, 分析了在低信噪比下进行信号调制识别的可行性并引入了噪声项。由于噪声项的干扰, 双谱在0 dB以下时, 噪声抑制效果变差, 提出了基于信息熵评价的经验模态化分解对信号进行预处理, 提高信噪比。最后, 设计了卷积神经网络分类器, 实现对不同调制类型信号的识别。仿真实验结果表明, 本文方法相比传统方法具有良好的抗噪性, 能够在低信噪比下对不同类型信号进行有效识别。

Bispectrum feature recognition of radar signal based on entropy evaluation and modal decomposition

In order to solve the problems of low accuracy and poor anti-noise of radar signal modulation recognition under low signal to noise ratio (SNR), a recognition method based on entropy evaluation modal decomposition and bispectrum feature extraction is proposed. Based on the characteristics of bispectrum which can suppress Gaussian noise, the feasibility of signal modulation recognition in low SNR is analyzed and the noise term is introduced. Due to the interference of noise term, the noise suppression effect of bispectrum below 0 dB becomes worse. An empirical modal decomposition based on information entropy evaluation is proposed to preprocess the signal and improve the SNR. Finally, a convolutional neural network classifier is designed to recognize different modulation signals. Simulation experiment results show that this method has better anti-noise performance than the traditional method, and can effectively identify different types of signals in low SNR.