Separation of Trend and Chaotic Components of Time Series and Estimation of Their Characteristics by Linear Splines
This paper considers the problem of separating the trend and the chaotic component of chaotic time series in the absence of information on the characteristics of the chaotic component. Such a problem arises in nuclear physics, biomedicine, and many other applied fields. The scheme has two stages. At the first stage, smoothing linear splines with different values of smoothing parameter are used to separate the “trend component.” At the second stage, the method of least squares is used to find the unknown variance σ2 of the noise component.
Unable to display preview. Download preview PDF.
- 2.A. V. Kryanev, G. V. Lukin, and D. K. Udumyan, Metric Analysis and Data Processing (Fizmatlit, Moscow, 2012) [in Russian].Google Scholar
- 4.O. S. Amosov and N. V. Muller, Sovrem. Naukoemk. Tekhnol., No. 3, 122–124 (2014).Google Scholar
- 5.Yu. P. Lukashin, Adaptive Methods for Time Series Short-Term Forecasting (Finansy Statistika, Moscow, 2003) [in Russian].Google Scholar
- 6.Neuroinformatics-2004, Proceedings of the Scientific Session of MEPhI, 6th All-Russia Conference, Ed. by O. A. Mishulina (Mosk. Inzh. Fiz. Inst., Moscow, 2004).Google Scholar