基于多维时间序列模型的大坝位移变形监测  

Dam Displacement and Deformation Monitoring based on Multidimensional Time Series Model

作　　者：陆爱萍 焦元冰 LU Aiping;JIAO Yuanbing(Zhejiang Institute of Surveying and Mapping Science and Technolog,Hangzhou,Zhejiang 310030,China)

机构地区：[1]浙江省测绘科学技术研究院,浙江杭州310030

出　　处：《测绘标准化》2022年第4期52-56,共5页Standardization of Surveying and Mapping

摘　　要：为了提高现有单因子时间序列模型精度,更好地发挥时间序列预测模型的作用,在单维时间序列模型的基础上,根据变形点多维因子之间的关联性,结合大坝位移变形时间序列非线性和非平稳性特征,提出将多维时间序列模型应用到大坝变形预测中。通过模型识别、参数估计实现模型构建,利用大坝实测数据对单一时间序列模型与多维时间序列模型进行对比分析。结果表明,单一时间序列模型预测结果总体精度为0.30 mm,多维时间序列模型的预测精度更高,总体精度为0.20 mm;多维时间序列一步预测的最大残差为0.23 mm,多步预测最大残差为0.53 mm。相比于单维时间序列模型,多维时间序列模型能更全面地对大坝位移时间序列的变化趋势进行描述,预测精度更高。由于预测结果的残差会随着预测期数的增加而变大,因此应使用新数据重新建模,可有效提高预测精度。In order to improve the accuracy of the existing single factor time series model and give better play to the role of the time series prediction model, this paper proposes to apply a multidimensional time series model to the dam deformation prediction on the basis of the single dimension time series model, according to the correlation between the multidimensional factors of the deformation points, combined with the nonlinear and non-stationary characteristics of the dam displacement deformation time series. Then the paper constructs the model through model identification and parameter estimation, and compares the single time series model with the multidimensional time series model using the dam measured data. The results show that the overall accuracy of the single time series model is 0.30 mm, and that of the multidimensional time series model is higher, with an overall accuracy of 0.20 mm;The maximum residual of one-step prediction of multidimensional time series is 0.23 mm, and the maximum residual of multi-step prediction is 0.53 mm. Compared with the single dimension time series model, the multidimensional time series model can describe the change trend of the dam displacement time series more comprehensively and has higher prediction accuracy. Since the residual error of the prediction results will increase with the increase of the number of prediction periods, new data should be used to re model, which can effectively improve the prediction accuracy.

关 键 词：大坝变形监测 时间序列 多维因子 参数估计 

分 类 号：TV690.1[水利工程—水利水电工程]