面向多普勒后STAP的基于先验知识训练样本选取方法

Prior knowledge-based training sample selection method for post-Doppler STAP

针对非均匀杂波环境中有效训练样本数严重不足的问题,提出一种面向多普勒后空时自适应处理(STAP)的基于先验知识的训练样本选取方法,给出了基于相邻距离-多普勒(RD)单元的地貌特征的最优权值系数选取准则.该方法联合相邻多个多普勒通道作为样本筛选的整体,通过相邻RD单元加权的方式,突出相邻多普勒通道在样本选取时参与决策的影响程度,然后基于加权归一化向量构建欧氏距离测度,实现对均匀训练样本的有效选取.仿真结果验证了本文所提方法的有效性.

Aiming at the serious shortage of effective training samples in non-uniform clutter environment,this paper proposes a training sample selection method based on prior knowledge for post-Doppler space-time adaptive processing(STAP),and also presents the selection criteria for the optimal weight coefficient of geomorphic features based on adjacent range-Doppler(RD)units.This method combines multiple adjacent Doppler channels as a whole for sample screening.The influence degree of adjacent Doppler channels participating in decision-making in sample selection is highlighted by weighting adjacent RD units.Then the Euclidean distance measure is constructed based on weighted normalized vector to realize the effective selection of uniform training samples.Simulation results verify the effectiveness of the proposed method.