摘要： 提出一种变参数在线核聚类算法(OKCAP)，将其应用于未知雷达辐射源信号分选中。OKCAP基于支持向量机的思想，采用核映射技术将数据映射到高维线性空间中进行处理，利用随机梯度下降法更新类的边界函数，且梯度下降步长和惩罚项参数可根据雷达信号动态调整，从而实现雷达辐射源信号的在线分选。仿真结果证明，该方法具有较快的聚类分选速度和较高的分选准确率。

关键词: 信号分选, 在线聚类, 核方法, 支持向量机, 随机梯度下降

Abstract: This paper presents an Online Kernel Clustering with Parameters Adaptation(OKCPA), and applies it to the unknown radar emitter signal sorting. OKCPA algorithm which is based on Support Vector Machine(SVM), uses the kernel trick to map the dates into the high-dimensional linear space, and uses stochastic gradient descent to update the boundary function, the step size and penalization parameter is updated with the dates coming, which accelerates the speed of clustering sorting. Simulation results show that this method has higher clustering sorting speed and higher sorting accuracy.

Key words: signal sorting, online clustering, kernel method, Support Vector Machine(SVM), stochastic gradient descent

中图分类号: 

• TN92