摘要
基于时间序列分析模型,运用高阶差分、灰色模型、频谱分析和奇异谱分析方法对非平稳时间序列提取趋势项,然后对平稳残差序列进行时序建模。通过对大坝位移监测序列建模分析进行预测,发现奇异谱分析和频谱分析时序分析模型相对于传统的高阶差分和灰色模型时序分析模型的预测精度有所提高。
On the basis of the time series analysis model, higher difference, gray model, spectrum analysis and singular spectrum analysis method are used to extract non-stationary time series trend, and model smooth resid- uals for time series. The analysis results of dam displacement monitoring series indicate that prediction accuracy is improved by using singular spectrum analysis and spectrum analysis of time series analysis model compared to the traditional model of higher difference timing analysis model and gray prediction model.
出处
《大地测量与地球动力学》
CSCD
北大核心
2013年第6期150-154,共5页
Journal of Geodesy and Geodynamics
关键词
趋势项提取
高阶差分
灰色模型
频谱分析
奇异谱分析
trend extraction
higher difference
gray model
spectrum analysis
singular spectrum analysis
