基于高次时频谱特征的舰载相控阵雷达信号降噪识别

Signal denoising recognition of shipborne phased array radar based on high-order time-frequency spectrum features

针对舰载相控阵雷达信号在复杂电磁环境中,噪声信号频谱能量超过目标信号,导致雷达信号难以识别这一问题,提出基于高次时频谱特征的舰载相控阵雷达信号降噪识别方法。通过CWD时频变换方法,将原始雷达信号映射到时频域,提取信号二维时频分布信息后,对CWD时频分布信息进行幂处理,计算时频信息中各元素幂次信息,提取信号高次时频谱特征;由广义S变换方法,调节信号高次时频谱特征滤波所用高斯窗函数与时频分辨率,完成雷达信号时频降噪。将降噪后的信号输入循环神经网络中,捕捉降噪信号中时序依赖关系,学习不同信号类型之间区分特征,完成雷达信号降噪识别。经测试,此方法在强电磁干扰下对信号的识别结果准确,可以有效弱化复杂电磁环境干扰影响。

Aiming at the problem that the noise signal spectrum energy of shipborne phased array radar signals exceeds the target signal in complex electromagnetic environments,making it difficult to recognize radar signals,a denoising recog-nition method for shipborne phased array radar signals based on high-order time-frequency spectrum features is proposed.By using the CWD time-frequency transformation method,the original radar signal is mapped to the time-frequency domain.After extracting the two-dimensional time-frequency distribution information of the signal,the CWD time-frequency distri-bution information is power processed to calculate the power information of each element in the time-frequency information and extract the high-order time-frequency spectrum characteristics of the signal;By using the generalized S-transform meth-od,the Gaussian window function and time-frequency resolution used for filtering high-order time-frequency spectrum fea-tures of the signal are adjusted to achieve time-frequency denoising of radar signals.Input the denoised signal into a recur-rent neural network,capture the temporal dependencies in the denoised signal,learn the distinguishing features between dif-ferent signal types,and complete radar signal denoising recognition.After testing,this method has shown accurate recogni-tion results for signals under strong electromagnetic interference and can effectively weaken the impact of complex electro-magnetic environment interference.