基于分形插值的ARIMA大坝预警模型

ARIMA Dam Early Warning Model Based on Fractal Interpolation

针对ARIMA模型对含有缺失值的时间序列进行拟合预测会产生较大的误差,将分形插值与ARIMA模型相结合运用在大坝安全监测中.首先利用分形插值能通过分形物体的部分信息得出其整体形态的特点,对原始数据进行插值计算,然后建立ARIMA时间序列模型并进行预测.以小湾坝顶某监测点径向位移为例,建立基于分形插值的ARIMA模型,并与实测值比较.计算结果表明,插值后的ARIMA模型较原始模型的拟合和预测精度更高.

ARIMA models for time series containing missing values will have a greater error due to fitting pre- diction. This article combines the fractal interpolation and ARIMA models to monitor the dam safety. First, by using fractal interpolation sub-section shaped object information to derive its overall shape of the character- istics of the original data interpolation, and then the ARIMA time series model is created to make predictions. Taking the crest of Xiaowan Dam radial displacement monitoring point, for example, the ARIMA model is es- tablished based on fractal interpolation; and then the results are compared with the measured values; it is shown that the latter has higher forecasting accuracy than the original model.