基于时间序列模型的边坡位移预测研究

Research on Slope Displacement Prediction Based on Time Series Model Fujian Academy of Building Research

本文采用差分自回归移动平均模型(ARIMA),应用于泉州市某边坡2019年12月18日-2021年2月7日时间序列坡顶水平位移监测数据。通过差分方式将原始数据转化为平稳时间序列,利用ACF函数、PACF函数与BIC准则进行模型定阶,最终建立了边坡坡顶水平位移时间序列预测模型,对今后滑坡状态进行预测。结果表明,利用ARIMA模型,对边坡位移进行预测,与监测单位的实际监测结果进行对比,二者结果基本一致,证明了利用时序模型预测边坡位移的有效性。

In this paper,the differential autoregressive moving average model(ARIMA)is used to apply the horizontal displacement monitoring data of the top of a slope in Quanzhou from December 18,2019 to February 7,2021.The original data is transformed into a stable time series by difference method,and the model order is determined by ACF function,PACF function and BIC criterion.Finally,a time series prediction model of the slope top horizontal displacement is established to predict the future landslide state.The results show that,the slope displacement predicted by the ARIMA model is basically same as the actual monitoring results of the monitoring unit,which proves the effectiveness of using the time series model to predict the slope displacement.