摘要
为解决Kmeans算法随机指定初始点聚类和海洋Argo浮标数据异常问题,提出一种改进Kmeans算法的海洋数据异常检测方法。提出一种改进Kmeans算法DMKmeans(density mathematics Kmeans),选取给定邻域范围内最近邻数据点最多的点为初始中心点,迭代聚类,直到准则函数收敛,聚类结束;基于DMKmeans算法对数据集聚类,使用数学模型为准则进行海洋监测数据异常检测。通过海洋监测数据异常检测仿真实验,将DMKmeans算法与传统Kmeans算法及MinMaxKmeans算法做对比分析,其结果表明,提出算法能有效提高聚类准确率和异常检测率。
The initial point clustering is assigned randomly and the marine Argo buoy data anomaly problem exists in the Kmeans algorithm.To solve the problems,an improved Kmeans algorithm for ocean data anomaly detection method was proposed.An improved Kmeans algorithm DMKmeans(density mathematics Kmeans)was proposed.The nearest neighbor point in the given neighborhood was chosen as the initial center point,and iterative clustering was implemented.Until the criterion function converged,the clustering ended.Based on the DMKmeans algorithm,the data set was clustered,and the mathematical model was used as the criterion for the ocean monitoring data anomaly detection.The DMKmeans algorithm was compared with the traditional Kmeans algorithm and MinMaxKmeans algorithm.The results show that the proposed algorithm can effectively improve the clustering accuracy and anomaly detection rate.
作者
蒋华
季丰
王慧娇
王鑫
罗一迪
JIANG Hua;JI Feng;WANG Hui-jiao;WANG Xin;LUO Yi-di(School of Computer and Information Security,Guilin University of Electronic Technology,Guilin 541000,China)
出处
《计算机工程与设计》
北大核心
2018年第10期3132-3136,共5页
Computer Engineering and Design
基金
2016广西高校中青年教师基础能力提升基金项目(ky2016YB150)
桂林电子科技大学研究生教育创新计划基金项目(2017YJCX48)
