Page 97 - 《应用声学》2021年第3期
P. 97

第 40 卷 第 3 期                                                                       Vol. 40, No. 3
             2021 年 5 月                          Journal of Applied Acoustics                      May, 2021

             ⋄ 研究报告 ⋄



              基于改进变分模态分解的北极海域声速剖面分类                                                                       ∗





                                  吕玉娇      1,2,3  尹 力    1,2†  刘崇磊      1,2  黄海宁      1,2


                                              (1  中国科学院声学研究所      北京   100190)
                                        (2  中国科学院先进水下信息技术重点实验室           北京  100190)
                                                (3  中国科学院大学     北京   100049)

                摘要：应用支持向量机对北极声速剖面进行分类，特征量提取是关键。该文采用一种基于经验模态分解的改
                进变分模态分解算法，以准确提取声速剖面特征量。算法首先对声速剖面信号进行经验模态分解，依据最大类
                间方差原则划分各分量边际谱主频带，以相似度作为最小分解层数判断标准，获得最小分解层数，进行变分模
                态分解。对北极区海水声速实测数据 (信号) 处理表明，该方法可有效提取信号经验模态分解各分量的希尔伯
                特边际谱特征，进行支持向量机分类，实现对北极海域声速剖面的分类识别，解决以往人工分类耗时久的问题。
                关键词：北极；声速剖面；最大类间方差；变分模态分解；支持向量机
                中图法分类号: TB52+6            文献标识码: A          文章编号: 1000-310X(2021)03-0415-07
                DOI: 10.11684/j.issn.1000-310X.2021.03.013

               Sound speed profile classification of Arctic sea area based on improved VMD



                              LYU Yujiao 1,2,3  YIN Li 1,2  LIU Chonglei 1,2  HUANG Haining 1,2

                               (1  Institute of Acoustic, Chinese Academic of Sciences, Beijing 100190, China)
                (2  Key Laboratory of Science and Technology on Advanced Underwater Acoustic Signal Processing, Chinese Academy of
                                                 Sciences, Beijing 100190, China)
                                  (3  University of Chinese Academic of Sciences, Beijing 100049, China)

                 Abstract: Using support vector machine to classify the North Pole sound velocity profile, feature extraction
                 is the key. In this paper, an improved variational mode decomposition algorithm (IEVMD) based on empirical
                 mode decomposition is used to extract the sound velocity profile features accurately. The algorithm firstly
                 conducts empirical mode decomposition for the sound velocity profile signal, divides the main frequency band
                 of the marginal spectrum of each component according to the principle of maximum intercategory variance,
                 takes similarity as the judgment standard of the minimum decomposition layer number, obtains the minimum
                 decomposition layer number, and performs the variational mode decomposition. The processing of measured
                 acoustic velocity data (signals) shows that this method can effectively extract the Hilbert marginal spectrum
                 characteristics of the empirical mode decomposition of signals, carry out support vector machine classification,
                 realize the classification and identification of acoustic velocity profiles in the Arctic sea area, and solve the
                 problem of time-consuming artificial classification in the past.
                 Keywords: Arctic; Sound speed profile; Otsu; Variational mode decomposition; Support vector machine


             2020-07-25 收稿; 2020-10-23 定稿
             国家重点研发计划项目 (2018YFC1405904)
             ∗
             作者简介: 吕玉娇 (1995– ), 女, 山东烟台人, 硕士研究生, 研究方向: 信号与信息处理。
             † 通信作者 E-mail: yl@mail.ioa.ac.cn
   92   93   94   95   96   97   98   99   100   101   102