基于时频图像和神经网络的LPI雷达信号调制识别

Modulation Identification of LPI Radar Signals Based on Time Frequency Image and Neural Network

针对LPI雷达信号调制识别问题,提出一种基于时频分析、图像处理和神经网络的LPI雷达信号识别新方法。该方法先对LPI雷达信号进行时频分析,获得时频分布图像,然后利用图像处理的方法对时频图像作预处理,最后再用RBF神经网络对处理后的图像进行识别分类。仿真实验表明,该方法在信噪比高于3dB时,平均正确识别率达到了92%。

To correctly classify low probability of intercept（LPI） radar signals,a novel method was presented based on time frequency analysis,image processing and neural network.First,LPI radar signals were analyzed in time frequency domain and their time frequency images were obtained.Then,the time frequency images were pre-processed by image processing method.Finally,the images were used to train a RBF neural network for automatic identification and classification of the LPI radar signals.The experiments show that the correct identification rate of the method comes to 92% when the signal-to-noise ration（SNR） is above 3dB.