【摘要】 针对雷达目标回波的非平稳特性,推导了S变换及其实现算法,利用电磁场时域有限差分算法仿真了三种军用飞机的宽带散射信号,采用S变换对飞机目标的雷达回波进行时频分析,提取时频分布图的奇异值特征作为目标特征矢量,利用径向基函数神经网络对特征矢量进行训练和学习,最后对三种飞机做了分类识别,取得了很好的识别效果。更多还原

【Abstract】 A new approach to the time-frequency signal analysis was proposed, using the S transform method, the time-frequency representation that could produce a distribution equal or close to a sum of the Wigner distributions of individual signal components. The new time-frequency signal decomposition method was evaluated on the simulated data using finite difference time domain (FDTD). A new approach of feature extraction was proposed based on singular values of the S transform time-frequency matrix. The RBFNN was employed as the classifier. Results demonstrate that it provides an effective way for radar automatic target recognition. The proposed method could also be used in other signal processing applications.更多还原