针对需要按照一定的评估标准快速地从大量冗杂的脉冲数据中筛选出高可分特征,减少雷达信号分选算法复杂度的同时可适应复杂信号环境的问题,文中提出一种对脉冲描述字特征进行重要性度量的预分选改进方法,通过求取特征信息熵及互信息值的方法估计特征对分类的影响程度,选择出重要性高的特征进行预分选。实验验证表明:根据重要性高的特征进行预分选可有效提高分选准确率。该算法解决了传统分选方法单一采用固定特征进行分选识别导致错选、误选的问题,提高了分选准确率,更具有实际工程应用意义。

For the large amount of pulse data, it is necessary to quickly select highly separable features, retain more effective features for sorting, reduce sorting complexity and adaptive environment. A method for measuring the importance of pulse descriptor features is proposed. By extracting the feature information entropy and mutual information value between features to estimate the degree of influence of classification, low entropy feature is considered that have higher importance. Sorting the importance degree of features, and calculating the mutual information value between high importance features, and the high importance feature is adaptively selected for pre-classification by threshold judgment. This method solves the problem that traditional methods can not adapt to the environment, and single fixed feature is used for sorting and recognition, which leads to miss-election. Experiments show that after measuring the importance of radar pulse descriptor features, the sorting accuracy can be effectively improve based on high-importance features. The algorithm solves the problem that the traditional method adopts fixed features for sorting and identification, resulting in mis-selection. It improves the accuracy of sorting and has practical engineering application significance.

TN957. 51

国家自然科学基金资助项目(61571146);中央高校基本科研业务费专项资助项目(HEUCFP201808);装备预研基金资助项目(No. 61404150103030)