Application of wavelet and Fourier transform in time series analysis
-
摘要: 正确提取坐标时间序列中的特征信息是非线性变化分析的前提.根据傅里叶变换和小波变换各自的特点,提出将两种方法结合起来对时间序列在时域和频域上进行分析的算法.首先采用小波函数db4对坐标时间序列分解5层得到高频和低频部分,进而分析各次谐波的时域波形以及可能存在的突变信息和区间,再在快速傅里叶变换的基础上求得各次谐波的准确频率和幅值.研究结果表明,低频分析可以直观地得到“周年项”和“两年周期项”,而高频分析能够较准确提取“半周年项”、“一季项”等短周期.与单独采用傅里叶变换或小波变换相比,基于小波变换与傅里叶变换相结合的方法能够有效地提取坐标时间序列中的特征信息,具有较高的研究价值.Abstract: Exactly decomposing the feature information of time series is the precondition to nonlinear variation analysis. According to the characteristics of Fourier and wavelet transform, two methods are integrated and applied to analyze time series in time domain and frequency domain, and comprehensive algorithm of wavelet and Fourier transform is presented. Firstly, the wavelet function DB4 is used to decompose the coordinate time series into five layers to get the high frequency and low frequency parts. Then, the time domain waveforms of each harmonic and the possible sudden change information and intervals are obtained. Finally, the exact frequency and amplitude of each harmonic are obtained on the basis of fast Fourier transform. The results show that the low-frequency analysis can intuitively obtain the “annual term” and “two-year cycle term”, while the high-frequency analysis is consistent with the extraction of short-term cycles such as “semiannual term” and “one-season term”. So the method based on wavelet transform and Fourier transform has many advantages compared with Fourier transform and wavelet transform alone, it can effectively extract the feature information of station time series, and has a big research value.
-
Key words:
- time series /
- non-linear variation /
- wavelet transform /
- Fourier transform /
- multi resolution analysis
-
[1] 姜卫平,李昭,刘鸿飞,等. 中国区域IGS基准站坐标时间序列非线性变化的成因分析[J].地球物理学报, 2013, 56(7): 2228-2237. [2] 薛蕙, 罗红. 小波变换与傅里叶变换相结合的暂态谐波分析方法[J]. 中国农业大学学报, 2007, 12(6):89-92. [3] 宁津生, 汪海洪, 罗志才. 小波分析在大地测量中的应用及其进展[J]. 武汉大学学报(信息科学版), 2004, 29(8):659-663. [4] KELLER W. A wavelet approach for the construction of multigrid solvers for large linear systems[M]//Vistas for Geodesy in the New Millennium, Springer, 2002:265270.DOI: 10.1007/978-3-662-04709-544. [5] GIBERT D, HOLSCHNEIDER M, MOUL J L L. Wavelet analysis of the Chandler wobble[J]. Journal of Geophysical Research, 1998, 103(B11):27069-27089.DOI: 10.1029/98JB02527. [6] 黄声享, 刘经南, 柳响林. 小波分析在高层建筑动态监测中的应用[J]. 测绘学报, 2003, 32(2):153-157. [7] 郭英起, 史大起, 黄声享,等. 高精度GPS测量中小波分析的应用[J]. 测绘工程, 2009, 18(3):58-60,64. [8] 田亮. GPS测站坐标非线性变化规律分析与机制研究[D]. 郑州:解放军信息工程大学, 2011. [9] 吴大正. 信号与线性系统分析[M].3版. 高等教育出版社, 1998. [10] 潘泉. 小波滤波方法及应用[M]. 北京: 2005. [11] 张勤, 蒋廷臣, 王秀萍. 小波变换在变形监测中的应用研究[J]. 测绘工程, 2005, 14(1):8-10. [12] YUE W, MENG X H, LI S L. Wavelet analysis and its application in geophysics of China[J]. Progress in Geophysics, 2012, 27(2):750760.DOI:10.6038/j.issn.10042903.2012.02.043. [13] WILLIAM C DONALD P. Wavelet methods for time series analysis[J]. Technometrics, 2016, 43(4):491. [14] 文鸿雁. 基于小波理论的变形分析模型研究[D]. 武汉:武汉大学, 2004. [15] GHADERPOUR E, INCE E S, PAGIATAKIS S D. Least-squares cross-wavelet analysis and its applications in geophysical time series[J]. Journal of Geodesy, 2018,92(10):1223-1236.DOI:10.1007/S00190-018-1156-9. [16] 范朋飞. 高精度GPS站点坐标时间序列分析与应用[D]. 西安:长安大学, 2013. [17] 孙付平, 田亮, 门葆红,等. GPS测站周年运动与温度变化的相关性研究[J]. 测绘学报, 2012, 41(5):723-728. [18] ALTAMIMI Z, COLLILIEUX X. IGS contribution to the ITRF[J]. Journal of Geodesy, 2009, 83(3-4):375-383.DOI: 10.1007/S00190-008-0294-X. [19] 刘向丽, 王旭朋. 基于小波分析的股指期货高频预测研究[J]. 系统工程理论与实践, 2015, 35(6):1425-1432. [20] 尹咪咪. 心电信号分析处理及心肌梗塞疾病模型的建立[D].郑州:郑州大学,2016. [21] 马社祥, 刘贵忠, 曾召华. 基于小波分析的非平稳时间序列分析与预测[J]. 系统工程学报, 2000, 51(4):305-311.
点击查看大图
计量
- 文章访问数: 542
- HTML全文浏览量: 104
- PDF下载量: 113
- 被引次数: 0