基于多维特征的密集转发式干扰识别方法

针对密集转发式干扰与真实目标回波的高相干性，传统基于单维特征的干扰识别方法识别率低的问题，提出了基于多维特征的干扰识别方法。从分数阶傅里叶变换域提取调频参数，从自相关幅度谱和匹配滤波幅度谱分别提取矩峰度系数，从频域提取频谱半带宽方差比系数，组成特征向量，作为决策树和BP神经网络的输入，对几类密集转发式干扰进行分类识别。仿真结果表明，所提特征参数对噪声不敏感，具有较好的稳定性和分离性，在干信比大于6dB时，能够有效辨识几类密集转发式干扰。

A Recognition Method of Dense Repeater Jamming Based on Multiple Features

The high coherence between dense repeater jamming and real target echo makes the classical jamming recognition method based on single dimensional feature have low recognition rate. For this problem, a recognition method based on multiple features is proposed. The chirp-rate parameter is extracted from fractional Fourier transform domain, the kurtosis coefficient is respectively extracted from auto-correlation amplitude spectrum and matched filter amplitude spectrum, and the spectrum half-band-width variance ratio coefficient is extracted from frequency domain, which constitutes the feature vector as the input of decision tree and BP neural network to recognize several kinds of dense repeater jamming. Simulation result shows that these features have better robustness, stability and separability. When the jamming-to-signal ratio is greater than 6dB, the proposed method can effectively recognize several kinds of dense repeater jamming.