基于两级卷积神经网络的相控阵雷达行为识别

Phased Array Radar Behavior Recognition Based on Two-stage Convolutional Neural Network

相控阵雷达作为新体制多功能雷达广泛应用于战场。对相控阵雷达行为的分析与识别有利于电子对抗策略的优化。使用跟踪密度将雷达行为进行细化,并使用两级卷积神经网络(CNN)对相控阵雷达进行行为识别。将相控阵雷达一个波位上所发射的一组脉冲作为一个波形单元,并基于波形单元的幅度与波形序列,使用第1级CNN实现工作模式识别,进一步使用第2级多列CNN实现搜索加跟踪(TAS)模式下的跟踪密度估计。最终,结合工作模式与跟踪密度给出雷达行为识别结果。仿真实验验证了本算法的有效性。

As the multi-function radar with new system,the phased array radar has been widely used in the battlefield.The behavior analyzation and recognition for phased array radar are beneficial to the optimization of electronic countermeasure.This paper uses tracking density to refine the radar behavior,and employs two-stage convolutional neural networks(CNN)to recognize the behavior of phased array radar.A set of pluses transmitted from a beam position of phased array radar are taken as a waveform unit,based on the amplitude and waveform sequence of waveform unit,radar working mode recognition is realized via the first-level CNN,and the tracking density in the tracking and searching(TAS)mode is estimated via the second-level multi-column CNN.Finally,radar behavior is recognized by combining the working mode with tracking density.Simulation experiment validates the effectiveness of the proposed algorithm.