光滑曲线去噪算法在分段线性拟合时间序列中的应用研究

Research of Denoising Algorithm Smooth Curve in the Application of Piecewise Linear Fitting Time Series

时间序列在经济社会等多个领域发挥着重要的作用。然而,时间序列通常含有较多不规则波动,这些不规则波动易对时间序列数据挖掘造成影响。因此,对时间序列进行降噪处理则是一个亟待解决的问题。该文介绍了一种基于光滑曲线去噪算法在分段线性时间序列中的应用方法。通过对时间序列进行光滑去噪处理,从而得到去噪后的光滑曲线数据,再通过时间序列分段线性的方法找出该序列数据的关键点,进行时间序列的线性分段拟合。实验表明:与直接分段拟合相比,先通过光滑去噪后再进行分段线性拟合得到的结果更好。

Time series plays an important role in the economic, society, and other fields. However, time series usually contains many irregular fluctuations which are easy to cause an negative effect on time series data mining. Therefore, noise rcduction processing is a problem to be solved. This paper introduced a denoising algorithm based on smooth curve in the application of piecewise linear time series. Smooth curve data is created after removing the time series noise. Then by using the method of time series piecewise linear, data points arc found out to fit the original time series. The experiments show that： compared with direct subsection fitting methods, the experiments results are much better by doing smooth denoising firstly and then pieccwise linear fitting.