基于粗集理论的雷达目标类型识别

Type identify of radar target based on the theory of rough sets

将粗糙集决策算法用于常规雷达目标多特征融合分类,并进行雷达目标类型识别,可提高目标的识别率,增强系统的稳健性。通过建立目标特征知识系统与分类决策表,提出了决策规则的应用和系数的确定,并提出了目标融合分类的步骤与方法。当雷达从不同姿态角照射目标时,从回波序列提取的目标径向投影特征属性取值将分布在一个与目标结构尺寸有关的区间之内,加之雷达测量误差使目标属性测量值呈平稳正态分布,所以,对决策表的条件属性取值进行区间划分将有助于增加条件属性矢量的容差能力,提高分类决策的稳健性。以一个包含6类30种目标的知识系统为例的分类实验达到了不低于90%的识别率。

The fusion calculate way of rough sets fusion classification is applied in normal radar target identify.By building up the target characteristic knowledge system and the classification decision tables,decision rule and the method of the coefficient assurance are put forward,the step and methods for radar target blend classification are presented.The classification experiment with a knowledge system that includes 6 types and 30 kinds of targets has come to a resulte of no lower than the identify rate of 90%.