CAJ | 학술논문

粒子群算法（PSO）和神经网络的有机结合是目前一个十分活跃的研究领域。为分类问题的研究提供了新的思路和方法。针对基本SOM算法聚类数目不确定、聚类效果不佳等问题，提出PSOM算法用于未知雷达信号分选，利用PSO的优化算法替代SOM的启发式训练，对基本SOM算法进行改进，最后通过仿真实验验证了该算法在未知雷达信号分选应用上的有效性。
입자군산법（PSO）화신경망락적유궤결합시목전일개십분활약적연구영역。위분류문제적연구제공료신적사로화방법。침대기본SOM산법취류수목불학정、취류효과불가등문제，제출PSOM산법용우미지뢰체신호분선，이용PSO적우화산법체대SOM적계발식훈련，대기본SOM산법진행개진，최후통과방진실험험증료해산법재미지뢰체신호분선응용상적유효성。
Currently the combination between PSO and NN is a very active area of research, which provides new ideas and methods for sorting. The PSOM algorithm is put forward to unknown radar signal sorting for the basal SOM has many problems, such as uncertain number of clusters, poor clustering results and so on. The heuristic training of SOM is replaced by PSO which advances the basal SOM. Finally, the PSOM algorithm is verified to be effective in the area of unknown radar signal sorting according to the emulated experiments.