一种结合小波分析的EMD周期分析方法——以南宁市年降水量为例  

An EMD periodic analysis method combining with wavelet analysis

作　　者：杜懿[1] 麻荣永[1] 赵立亚 

机构地区：[1]广西大学土木建筑工程学院,南宁530004

出　　处：《广西水利水电》2018年第1期8-12,共5页Guangxi Water Resources & Hydropower Engineering

基　　金：国家自然科学基金项目(51369005);广西防灾减灾与工程安全重点实验室系统性研究项目(2013ZDX04)

摘　　要：针对传统EMD方法中Hilbert变换未必能正确计算出本征模态函数瞬时频率这一问题,采用小波分析替代Hilbert变换来优化EMD方法,为水文时间序列的周期分析提供了一种新途径。并对南宁市1961~2015年降水量时间序列进行了周期分析。结果表明,南宁市年降水量序列经改进的EMD方法分解可得到1个趋势项和4个IMF分量。趋势项表明南宁市历年年降水量呈现出明显的下降趋势;对IMF分量进行小波分析得到了年降水量序列分别具有4~6a、10~13a、25~29a的准周期特征。The Hilbert transform of traditional EMD method may be unable to correctly calculate the instantaneous frequency of intrinsic mode function. In view of this problem, the Hilbert transform is substituted by wavelet analysis to improve the EMD method and provide a new route for periodic analysis of hydrological time series. The annual precipitation time series of Nanning from 1961 to 2015 was analyzed with the improved EMD method, showing one trend and four IMF components. The annual precipitation of Nanning City has a significant dropping trend. The results of wavelet analysis of IMF components demonstrate the annual precipitation series have quasi-periodic characeristics of 4~6a, 10~13a and 26~29a.

关 键 词：周期分析 经验模态分解 本征模态函数 小波变换 年降水量 

分 类 号：P333[天文地球—水文科学]