海洋声速剖面的自动聚类研究  

Study on automatic clustering of sound speed profile in the ocean

作　　者：谢骏[1] 胡均川[1] 笪良龙[1] 李玉阳[1] 卢晓亭[1] 

机构地区：[1]中国人民解放军海军潜艇学院战术水声环境数据中心,山东青岛266071

出　　处：《海洋学报》2009年第2期34-39,共6页

基　　金：国防预研基金;新世纪优秀人才支持计划NCET

摘　　要：以海区30′网格方区多年月平均统计的声速剖面作为原始数据集,提取声速剖面的表层、主跃层和深海等温层分层结构特征,把我国近海及其邻近海域预分为Ⅰ,Ⅱ和Ⅲ类区。对Ⅱ,Ⅲ类区声速剖面,应用有序样本聚类算法分别进行表层分离。根据各类区的表层声速剖面数据,通过归一化处理和Akima差值采样得到梯度剖面,建立起按月归一化后的声速剖面分层梯度样本集,并应用系统聚类法和SOFM神经网络方法分别进行聚类分析,再根据分类结果并结合各类型海区的声学特点,得到各类型海区声速剖面的典型类型。通过对大量历史数据的分析结果表明,该方法为自动分类海洋声速剖面提供了一条有效路径,弥补了长期以来海洋声速剖面主要依靠人工分类的不足。Oceanic region is classified as Ⅰ,Ⅱ and Ⅲ types based on the structures of sound speed profiles of surface layer, main thermocline and deep isothermal layer. The sound speed profiles of surface layer, Ⅱ and Ⅲ oceanic region types are separated with sequential clustering analysis arithmetic. Sound speed gradient profile sample database is estahlished through normalization process and Akima sampling method for the sound speed profiles which are derived from 30′×30′latitude-longitude historical statistic data of mixed layer of the sound speed profile for every month. Hierarchical clustering and SOFM neural network clustering analysis arithmetic is performed to classify the sound speed profile based on the sample database. Representative types of the sound speed profile are summarized depended on clustering result and acoustic character istics of different oceanic region types. The classification results based on a great deal of historical statistic sound speed profile show that the above-mentioned method is a efficient technology road map to automatic classification of sound speed profile in the ocean and makes up for manual classification all along.

关 键 词：声速剖面 声速剖面类型 聚类分析 神经网络 特征提取 

分 类 号：P717[天文地球—海洋科学] P733.2