长江科学院院报 ›› 2020, Vol. 37 ›› Issue (3): 57-63.DOI: 10.11988/ckyyb.20181146

• 工程安全与灾害防治 • 上一篇    下一篇

基于EEMD-PCA-ARIMA模型的大坝变形预测

郑旭东1,2, 陈天伟1,2, 王雷1,2, 段青达1,2, 甘若1,2   

  1. 1.桂林理工大学 测绘地理信息学院,广西 桂林 541004;
    2.桂林理工大学 广西空间信息与测绘重点实验室,广西 桂林 541004
  • 收稿日期:2018-10-26 出版日期:2020-03-01 发布日期:2020-05-09
  • 通讯作者: 陈天伟(1965-),男,广西北流人,教授,硕士,主要从事测绘数字化与空间数据建模。E-mail:ctw@glut.edu.cn
  • 作者简介:郑旭东(1994-),男,河南郑州人,硕士研究生,主要从事变形监测与数据处理。E-mail:zhengxudongcs@126.com
  • 基金资助:
    广西自然科学基金项目(2017GXNSFAA198308);广西空间信息与测绘重点实验室主任基金项目(15-140-07- 09)

Dam Deformation Prediction Using EEMD-PCA-ARIMA Model

ZHENG Xu-dong1,2, CHEN Tian-wei1,2, WANG Lei1,2, DUAN Qing-da1,2, GAN Ruo1,2   

  1. 1.College of Geomatic Engineering and Geoinformatics, Guilin University of Technology, Guilin 541004, China;
    2.Guangxi Key Laboratory of Spatial Information and Geomatics, Guilin University of Technology, Guilin 541004, China
  • Received:2018-10-26 Online:2020-03-01 Published:2020-05-09

摘要: 针对大坝观测数据中存在的噪声容易掩盖实际变形曲线走势的问题,提出一种基于集合经验模态分解(EEMD)、主成分分析(PCA)和自回归移动平均模型(ARIMA)的大坝变形预测方法。通过对观测数据进行EEMD和PCA,从而构建映射矩阵,然后利用映射矩阵对原始数据构建的样本矩阵进行转换,实现消噪效果,进而对处理后的观测数据进行ARIMA建模预测,据此构建EEMD-PCA-ARIMA模型。依据所提出的模型对实际大坝坝顶水平位移观测数据进行预测分析,并与实测数据和经直接去掉高频分量消噪后的ARIMA预测模型、ARIMA预测模型、BP神经网络模型预测模型进行对比分析。结果表明:此方法能够更好地获取大坝的实际变形曲线,对于大坝变形预测而言是一种有效的方法。

关键词: 大坝变形, 变形预测, 集合经验模态分解, 主成分分析, 自回归移动平均模型

Abstract: The trend of actual dam deformation curve is easily covered up by noises in dam observation data. In this paper, a prediction method for dam deformation integrating ensemble empirical mode decomposition (EEMD), principal component analysis (PCA) and autoregressive integrated moving average (ARIMA) model is put forward. The mapping matrix is constructed by EEMD and PCA of the observed data, and then the sample matrix constructed by the original data is transformed by the mapping matrix to accomplish denoising; subsequently, the ARIMA model of the processed observation data is built and employed to predict and analyze the observed data of horizontal displacement of dam crest. The prediction result is compared with the measured data, ARIMA predicted data directly removed of high-frequency components, ARIMA predicted data, and BP neural network predicted data. Results suggest that the proposed method is effective for dam deformation prediction as it could better acquire the actual deformation curve.

Key words: dam deformation, deformation prediction, EEMD, PCA, ARIMA

中图分类号: