测绘通报 ›› 2018, Vol. 0 ›› Issue (9): 64-68.doi: 10.13474/j.cnki.11-2246.2018.0281

• 学术研究 • 上一篇    下一篇

大坝变形时间序列的奇异谱分析

李世友, 王奉伟, 沈云中   

  1. 同济大学测绘与地理信息学院, 上海 200092
  • 收稿日期:2017-10-09 出版日期:2018-09-25 发布日期:2018-09-29
  • 作者简介:李世友(1993-),男,硕士生,主要研究方向为GNSS变形监测。E-mail:1633295@tongji.edu.cn
  • 基金资助:

    国家自然科学基金(41731069;41474017)

Singular Spectrum Analysis for Dam Deformation Time Series

LI Shiyou, WANG Fengwei, SHEN Yunzhong   

  1. College of Surveying and Geo-Information, Tongji University, Shanghai 200092, China
  • Received:2017-10-09 Online:2018-09-25 Published:2018-09-29
  • Contact: 沈云中。E-mail:yzshen@mail.tongji.edu.cn E-mail:yzshen@mail.tongji.edu.cn

摘要:

为了研究大坝变形规律并分析其影响因子,利用奇异谱分析(SSA)提取大坝变形时间序列的趋势和周期分量,并分析各分量与时效、温度和水位因子的关联性。通过分析某大坝实测数据,发现大坝存在徐变和周期性弹性形变。其中,徐变即趋势项,主要与时效因子有关;弹性形变对应的周期项主要与温度和水位的周期变化有关,且温度因子与弹性形变的相关性明显大于水位因子。利用提取的趋势和周期项对大坝变形时间序列进行拟合并预测,其拟合和预测误差分别为0.52和0.24 mm;若采用传统的多元线性回归模型进行拟合和预测,其误差则分别为0.81和0.57 mm。这表明奇异谱方法的拟合和预测精度均优于多元回归法,能够更准确地预测大坝变形。

关键词: 奇异谱分析, 大坝变形分析, 变形预测

Abstract:

In order to study the properties of dam deformation and corrsponding influence factors,we extract the trend and periodic components of deformation series by singular spectrum analysis(SSA).The correlation coefficient between the components and the aging,temperature and water level factors are calculated.Experimental results show that the dam has creep and periodic elastic deformation.Among them,the creep is mainly related to the time efficiency factor.The periodic elastic deformation is mainly related to the period of temperature and water level factors.Besides,the contribution of the temperature factor to the elastic deformation is greater than the water level factor.The deformation series of the dam are fitted and predicted by the trend and period of extraction.The fitting and prediction errors are 0.52 and 0.24 mm.The traditional multiple regression model is used to fit and predict,the errors are 0.81 and 0.57 mm,respectively.The results show that in this study the singular spectrum analysis can predict the dam deformation more accurately than traditional multiple linear regression method.

Key words: singular spectrum analysis, dam deformation analysis, deformation prediction

中图分类号: