摘要
为了更加高质量地利用海况数据,避免由复杂因素导致的对海况误判问题,基于KMeans、核技巧、自组织映射(Self-organizing Mapping, SOM)神经网络构建了自组织映射混合核KMeans(SOM-Gaussian and Polynomial Kernel-KMeans, SGPK-KMeans)算法.克服了KMeans对复杂数据聚类效果不佳、核KMeans需要指定聚类数目和对初始聚类中心敏感的问题.通过海况数据聚类实验,将SGPK-KMeans算法的聚类效果与经典KMeans、单核KMeans和SOM神经网络算法进行对比分析.结果表明SGPK-KMeans对于海况数据聚类具有更加稳定的效果且能更加准确的识别出数据中的异常值.
In order to make better use of sea state data and avoid misjudgment of sea state caused by complex factors,Self-Organizing Mapping mixed kernel KMeans(SOM-Gaussian and Polynomial Kernel-KMeans,SGPK-KMeans)algorithm has been construsted on the basis of KMeans,kernel skills and Self-organizing Mapping(SOM)neural network for complex sea state data.It overcomes the following problems,for example,KMeans has poor clustering effect on complex data,kernel KMeans needs to specify the number of clusters and is sensitive to the initial clustering center.With the sea state data clustering experiment,the clustering effect of SGPK-KMeans algorithm is compared with that of classical KMeans,single-core KMeans and SOM neural network algorithms.The findings show that SGPK-KMeans has a more stable effect on sea state data clustering and can identify outliers in the data more accurately.
作者
陈晓曼
苏欢
CHEN Xiao-man;SU Huan(School of Science,Harbin Institute of Technology(Weihai),Weihai 264209,China)
出处
《陕西科技大学学报》
北大核心
2023年第3期208-214,共7页
Journal of Shaanxi University of Science & Technology
基金
山东省自然科学基金面上项目(ZR202102220411)。
