数据驱动的雷达目标近场交会回波建模方法

Data Driven Echo Modeling for Radar Target Near-field Engagement

摘要:

基于大量测试或仿真得到的近场弹目交会回波数据,提出了数据驱动的近场弹目交会回波建模方法。分析近场交会回波包络的峰值特征,建立以峰值幅度、位置和个数为特征参数的点散射源模型。采用基于支持向量机的统计学习方法,对大量近场交会回波包络的峰值特征参数进行训练学习,分别建立回波峰值幅度、位置和个数的支持向量回归模型。对上万条仿真点目标近场交会回波进行支持向量学习,结果表明,高斯核函数的支持向量回归模型能够实现高精度的目标回波峰值特征参数建模。

Abstract:

A data driven echo modeling approach for the near-field engagement isproposed based on a large number of measured or simulated echo data, The echo envelope otnear-field engagement can be characterized by peak features, whose is represent by a pointscattering model with its peak amplitude, position and order parameters. The statisticaearning method based on support vector machine (SVM) is applied to train peak featuresfrom a large amount of echo envelopes, and then the support vector regression (SVR)models for peak amplitude, position and order are obtained, represtively. The experiment often thousand simuluated echoes from point target near-field engagement indicates that theSVR with the Gaussian kernel can model peak features of target echoes accurately.