【摘要】 针对当前雷达信号分选中多参数域混叠严重、簇边界点划分模糊的问题,提出了一种基于共享邻域的雷达信号预分选算法。算法在密度峰值聚类算法的基础上引入共享邻域理论,通过评估样本间邻域共享程度来量化样本间的相似度。仿真实验结果表明,与其他聚类预分选算法相比,算法不仅提高了对簇边界样本点的处理能力,还避免了对预设参数的依赖,在调整互信息指数(AMI)、调整兰德指数(ARI)、福尔克斯-马洛斯指数(FMI)评价指标下均取得较好的聚类分选效果,使海战场复杂电磁环境下整体信号分选准确率有显著提升。更多还原

【Abstract】 In order to solve the problems of serious aliasing in multi-parameter domains and fuzzy division of cluster boundary points in the current radar signal selection,a radar signal pre-sorting algorithm based on shared neighborhood is proposed.Based on the density peak clustering algorithm,the algorithm introduces the shared neighborhood theory to quantify the similarity between samples by evaluating the degree of neighborhood sharing among samples.The simulative experiment results show that compared with other clustering pre-sorting algorithms,the algorithm not only improves the processing ability of cluster boundary sample points,but also avoids the dependence on preset parameters,and better clustering pre-sorting results are achieved under the evaluation indexes of adjusted mutual information(AMI),adjusted rand index(ARI) and Fomkes-Mallows index(FMI),so the accuracy of signal sorting in the complex electromagnetic environment of the naval battlefield is significantly improved.更多还原