A new scheme of data decomposition for non-stationary time series analysis

摘要

In this paper, a new scheme of data decomposition for non-stationary time series analysis is presented, which is based on so-called many-knot spline interpolation theory. Two examples are given to show the experiment analysis and results : from the beginning, the many-knot spline fitting curve is in low-frequency, then become in high-frequency more and more. Finally, the many-knot spline fitting curve will accord with the error completely, the error will become zero and the sum of the values of the many-knot spline fitting curves during decomposition process can reconstruct the original data, it is helpful for data composition, signal processing, information hiding and so on..