鲍燕妮, 沈丹祎, 石振明, 等. 2021. ARMA模型在锚碇基坑变形预测中的应用[J]. 工程地质学报, 29(5): 1621-1631. doi: 10.13544/j.cnki.jeg.2021-0527.
    引用本文: 鲍燕妮, 沈丹祎, 石振明, 等. 2021. ARMA模型在锚碇基坑变形预测中的应用[J]. 工程地质学报, 29(5): 1621-1631. doi: 10.13544/j.cnki.jeg.2021-0527.
    Bao Yanni, Shen Danyi, Shi Zhenming, et al. 2021. Application of ARMA model in deformation monitoring and forecasting of anchorage foundation pit[J]. Journal of Engineering Geology, 29(5):1621-1631. doi: 10.13544/j.cnki.jeg.2021-0527.
    Citation: Bao Yanni, Shen Danyi, Shi Zhenming, et al. 2021. Application of ARMA model in deformation monitoring and forecasting of anchorage foundation pit[J]. Journal of Engineering Geology, 29(5):1621-1631. doi: 10.13544/j.cnki.jeg.2021-0527.

    ARMA模型在锚碇基坑变形预测中的应用

    APPLICATION OF ARMA MODEL IN DEFORMATION MONITORING AND FORECASTING OF ANCHORAGE FOUNDATION PIT

    • 摘要: 本文在详细介绍时间序列预测原理、基本模型及预测步骤的基础上,选取新田长江大桥锚碇基坑工程施工过程中的基坑位移和危岩裂缝变形实测数据,运用时间序列法建立自回归滑动平均模型(ARMA),对施工过程中的基坑位移和危岩裂缝变形趋势进行预测,并将模型预测数据与实际监测数据进行对比。结果表明:ARMA模型可以通过差分算法很好地解决数据不稳定性问题,在预测周期较短时预测精度较高;随着预测周期的增加,ARMA模型的预测精度有所降低,其原因是由于随着预报步长的增加,预报所依赖的历史数据在减少;在建立ARMA模型时,定期加入新的数据,可以避免预测周期过长导致预测精度降低。本文的研究成果可以为今后类似工程的监测预报研究参考依据。

       

      Abstract: This paper introduces the time series prediction principle,the basic model and the prediction procedure in detail. It uses the measured data of foundation pit displacement and crack deformation during the construction of the foundation pit of Xintian Yangtze river bridge. It establishes the Autoregressive Integrated Moving Average Model(ARMA)using the time series method. On this basis,the deformation trend of the foundation pit and the crack of the dangerous rock in the construction process are predicted. The model prediction data are compared with the international monitoring data. The results show that the ARMA model can solve the problem of data instability by using difference algorithms,and has higher accuracy in the short-time prediction period. With the increase of the prediction period,the prediction accuracy of ARMA model decreases. The reason is that with the increase of the prediction period,the historical data on which the prediction relies are decreases. Therefore,when establishing the ARMA model,new data should be added regularly to avoid the reduction of prediction accuracy caused by too long prediction period. The research results of this paper can be a reference for the monitoring research of similar projects in the future.

       

    /

    返回文章
    返回