Abstract: Traditional recognition approaches to radar signal radio frequency are based on statistics，but the result is not satisfied.Clustering is a classical analytic method of pattern recognition which can be applied to radio frequency recognition as a new thought，and the theory and calculating of it are simple.The radio frequency mode can be analyzed through the results of clustering with radar signal radio frequency sequences.According to the simulation results，this recognition based on clustering is effective.At last，the simple clustering is replaced with dynamic clustering，whose results are more accurate.

Key words: dynamic clustering, radio frequency mode, absolute value distance, pattern recognition, threshold

摘要： 对雷达信号的载频模式进行识别的传统方法一般都是基于统计法的，但是往往识别效果并不理想。聚类分析方法是一种经典的模式识别方法，它原理简单，计算方便，可为雷达载频模式识别提供一种新的分析思路。通过对雷达信号的载频序列进行聚类分析，并根据聚类结果来对雷达信号载频模式进行识别，仿真实验验证了算法的有效性。最后在此基础上，使用动态聚类分析方法替换原来的聚类分析法，取得了更好的识别效果。

关键词: 动态聚类法, 载频模式, 绝对值距离, 模式识别, 门限