多传感器雷达侦察信号分选新方法研究

New Method on Radar Reconnaissance Signals Separation for Multi-Sensor

由于许多盲源分离方法局限于平稳、非高斯且相互独立的源信号,而实际的雷达信号并不满足这些条件。文章针对现有新体制雷达所发射信号的非平稳性,采用平滑伪Wigner-Ville分布并结合联合对角化的方法,对所侦察到的非平稳雷达信号进行盲分离。从仿真结果可知,分离信号与源信号相关系数的绝对值均在0.99以上,并且误差均在允许的范围之内,这充分说明本文所用的方法在一定条件下完全能够对非平稳雷达侦察信号进行盲分离,为后续侦察信号的分选、识别提供了依据。

For most of the blind source separation methods are confined to stationary,nongaussian and mutually independent source signals, and the real radar signals always don＇t satisfy this condition. View to the nonstationarity of the signals that emitted by modern new radar system, we chose the method of Smoothed Pseudo Wigner-Ville time-frequency distribution and integrate joint-diagonalization to blind separate the radar signals that was reconnaissanced. From the result of simulation, both the correlation coefficient between estimate signals and sources were larger than 0. 99, and the global error was confined to per- mission, this sufficiently implied that the method of time-frequency distribution can separate nonstationary signals under some special conditions completely, which provide evidence for radar reconnaissance signal＇s selection and recognization.