[1]李雪耀,张汝波,王武.基于支持向量回归机的HHT边界效应处理[J].智能系统学报,2007,2(3):39-44.
LI Xue-yao,ZHANG Ru-bo,WANG Wu.End effects processing in HHT based on support vector regression machines[J].CAAI Transactions on Intelligent Systems,2007,2(3):39-44.
点击复制
LI Xue-yao,ZHANG Ru-bo,WANG Wu.End effects processing in HHT based on support vector regression machines[J].CAAI Transactions on Intelligent Systems,2007,2(3):39-44.
基于支持向量回归机的HHT边界效应处理
《智能系统学报》[ISSN 1673-4785/CN 23-1538/TP] 卷:
2
期数:
2007年第3期
页码:
39-44
栏目:
学术论文—智能系统
出版日期:
2007-06-25
- Title:
- End effects processing in HHT based on support vector regression machines
- 文章编号:
- 1673-4785(2007)03-0039-06
- Keywords:
- end effects; HilbertHuang transform; support vector regression machines; particle swarm optimization
- 分类号:
- TP18
- 文献标志码:
- A
- 摘要:
- 针对希尔伯特-黄变换中的边界效应，提出了基于支持向量回归机的时间序列预测方法.在支持向量回归机的应用当中，参数的选取对它的泛化性能有很大影响.在讨论了参数对支持向量回归机的泛化性能的影响基础上，提出了通过微粒群优化算法来优化支持向量回归机参数的方法，使得支持向量回归机在应用中能够自适应的选择最优参数，从而获得了更好的泛化性能，提高了在端点处的延拓精度，很好地抑制了端点效应.试验表明，该优化算法能够很好解决支持向量回归机的参数选取问题.通过与神经网络的延拓方法和黄等人的 HHTDPS结果对比，基于支持向量回归机的时间序列预测方法可以更好地解决在希尔伯特-黄变换中存在的边界效应，得到的固有模态函数具有较小的失真.
- Abstract:
- In order to better restrain end effects in the HilbertHuang transform (HHT), a time sequence prediction technique is proposed based on support vector regression machines to improve time series prediction. In the application of support vector regression machines (SVRM), parameter selection has a great influence on generalization performance. So in this paper, the influence of parameters on the generalization of SVRM is discussed, and then a particle swarm optimization (PSO) algorithm is used to optimize parameters. Using this method, SVRM can select optimal parameters selfadaptively, so that higher generalization performance is obtained in applications, prediction accuracy is improved at both ends and the end effects are restrained effectively. In contrast to the neural network methods and HHTDPS proposed by Huang et al., the end effects can be restrained better and the Intrinsic Mode Functions have less distortion. Experiments show that this method can solve the problem of selecting parameters properly.
- 参考文献/References:
-
［1］ HUANG N E. The empirical mode decomposition and the Hilbert spectrum for n on linear and nonstationary time series analysis［J］. Proc R Soc Lond.A, 1998，45 4: 903-995.
［2］HUANG N E. A confidence limit for the empirical mode decomposition and Hil bert spectral analysis［J］. proc R Soc Lond, 2003，459: 2317-2345.
［3］VELTCHEVA A D, GUEDES S C. Identification of the components of wave spectr a by the Hilbert Huang transform method［J］. Applied Ocean Research, 2004，26:1- 1 2.
［4］RAY R R, ZHANG R C, LANCE V D, et al. On estimating site damping with soil n onlinearity from earthquake recordings［J］. International Journal of NonLin ea r Mechanics, 2004, 39(9):1501-1517.
［5］HUANG H, PAN J Q. Speech pitch determination based on HilbertHuang transf orm［J］. Signal Processing, 2006，86:792-803.
［6］赵进平. 异常事件对EMD方法的影响及其解决方法研究［J］.青岛海洋大学学报, 2001,31(6):805-814.
ZHAO Jinping. Study on the effects of abnormal events to empirical mode decompos ition method and the removal method for abnormal signal［J］. Journal of Ocean U niversity of Qingdao, 2001, 31 (6):805-814.
［7］盖 强, 马孝江, 张海勇, 等. 一种消除局域波法中边界效应的新方法［J］. 大连理工大学学报, 2002, 42(1):115-117.
GAI Qiang, MA Xiaojiang, ZHANG Haiyong, et al. New method for processing end e ffect in local wave method［J］. Journal of Dalian University of Technology, 200 2, 42 (1) : 115-117.
［8］张郁山, 梁建文, 胡聿贤. 应用自回归模型处理EMD方法中的边界问题［J］. 自然科学进展, 2003, 13(10):1054-1059.
ZHANG Yushan, LIANG Jianwen, HU Yuxian. The processing of end effects in EMD met hod by autoregressive model［J］. Progresses in Nature Science, 2003, 13(10):105 4-1059
［9］陈 忠, 郑时雄. EMD信号分析方法边缘效应的分析［J］. 数据采集与处理, 2003, 18(1):114-118.
CHEN Zhong, ZHENG Shixiong. Analysis on end effects of EMD method［J］. Journal of Data Acquisition and Processing, 2003, 18 (1):114-118.
［10］熊学军, 郭炳火, 胡筱敏, 等. EMD方法和Hilbert谱分析法的应用与探讨［J］.海洋科学进展, 2002,20(2):12-21.
XIONG Xuejun, GUO Binghuo, HU Xiaomin, et al. Application and discussion of emp irical mode decomposition method and Hilbert spectral analysis method［J］. Jour nal of Oceanography, 2002, 20 (2):12-21.
［11］ZENG K, HE M X. A simple boundary process technique for empirical mo de de composition［A］. Proceedings IGARSS ′04［C］. ［s.l.］,2004.
［12］ZHAO J P, HUANG D J. Mirror extending and circular spline function for em pi rical mode decomposition method［J］. Journal of Zhejiang University(Science),20 01, 2(3)：247-252.
［13］邓拥军, 王 伟, 钱成春, 等. EMD方法及Hilbert变换中边界问题的处理［J ］ . 科学通报, 2001,46(3):257-263.
DENG Yongjun, WANG Wei, QIAN Chengchun, et al. Boundary processing technique in EMD method and Hilbert transform［J］. Chinese Science Bulletin, 2001, 46 (3):25 7-263.
［14］许宝杰, 张建民, 徐小力, 等. 抑制EMD端点效应方法的研究［J］. 北京理工大学学报, 2006, 26(3): 196-200.
XU Baojie, ZHANG Jianmin, XU Xiaoli, et al. A study on the method of restraining the ending effect of empirical mode decomposition［J］. Transactions of Beijing Institute of Technology, 2006, 26(3): 196-200.
［15］KIZHNER S, BLANK K F. On certain theoretical developments u nd erlyi ng the HilbertHuang transform［A］. 2006 IEEE Aerospace Conference Proceeding［C ］. ［s.l.］, 2006.
【16］CHUANG C C, SU S F, JENG J T, et al. Robust support vector regression net wo rks for function approximation with outliers［J］. IEEE Trans Neural Netw, 200 2, 13(6): 1322-1330.
［17］VLADIMIR N V. The nature of statistical learning theory［M］. New York: S pringer, 1995.
［18］KEERTHI S S, LIN C J. Asymptotic behaviors of support vector machines wit h Gaussian kernel［J］. Neural Computation, 2003, 15 (7):1667-1689.
［19］SCHOLKOPF B, MIKA S, BURGES C, et a1. Input space versus feature space in k ernelbased methods ［J］. IEEE Transactions on Neural Networks, 1999, 10(5): 1 0 00-1017.
［20］RAA-UTSCH G, ONODA T, MFILLER K R. Soft margins for adaBoost ［J］.Mac hine Learning, 2001, 42(3): 287-320.
［21］KENNEDY I, EBERHAN, R C. Paticle swarm optimization［A］. Proceedings of IEEE lnlemational Conference on Neural Networks［C］. Piscalaway, NJ , 1995.
［22］SHI Y , EBERHART R C. Empirical study of particle swarm optimization［A］ . Proceeding of Congress on Evolutionary Computation［C］. Washington DC, USA, 199 9.
备注/Memo
收稿日期：2006-12-18.
基金项目：
国家自然科学基金资助项目（60475016）；
哈尔滨工程大学基础研究基金资助项目（HEUF04092）.
作者简介：
李雪耀，男，1944年生，教授，硕士生导师.主要研究方向为计算机听觉、模式识别和信号处理.曾获省部级科技进步奖3项，发表论文数十篇.
E-mail:lixueyao@hrbeu.edu.cn. TS)〗
张汝波，男，1963年生，博士，教授，博士生导师.主要研究方向为智能机器人软硬件体系结构、任务规划、路径规划、自主作业技术及强噪声背景下语音信号的检测与处理等. 发表论文100多篇，60多篇被SCI、EI、ISTP收录，出版专著5部.
王 武，男，1983年生，硕士研究生，主要研究方向为强噪声背景下的语音流检测.
更新日期/Last Update:
2009-05-07