CAJ | 학술논문

针对现有网格聚类算法需要人为确定网格划分、边界处理精度低的问题，提出了基于固定网格划分和动态网格划分的雷达信号预分选算法。基于固定的网格划分算法对输入信号脉冲顺序不敏感，根据网格数据压缩率自适应确定网格划分和密度阈值。基于移动技术的动态网格聚类算法能够识别任意形状和大小的聚类。仿真实验表明，这两种方法能有效适用于雷达信号分选，且能很好地识别出孤立点和噪声。
침대현유망격취류산법수요인위학정망격화분、변계처리정도저적문제，제출료기우고정망격화분화동태망격화분적뢰체신호예분선산법。기우고정적망격화분산법대수입신호맥충순서불민감，근거망격수거압축솔자괄응학정망격화분화밀도역치。기우이동기술적동태망격취류산법능구식별임의형상화대소적취류。방진실험표명，저량충방법능유효괄용우뢰체신호분선，차능흔호지식별출고립점화조성。
The conventional gird clustering algorithm has some problems such as needing parameters input and low clustering precision. So a signal sorting algorithm based on fixed grid partition and a signal sorting algorithm based on dynamic grid partition are proposed. The former has a low sensitivity to the order of radar signal input. It partitions the grid and gets density threshold automatically by grid data compression ratio. Dynamic grid clustering and shifting technique can discover arbitrary shapes and size of clusters. Both the algorithms are verified to be effective and better robustness to noise by simulations.