巷道支架位移ARIMA模型预报时的还原误差问题

Restoring Error Problem in the Prediction of Roadway Support Deformation with the ARIMA Model

在现场采集大量数据的基础上,应用时间序列分析线性模型的理论,对巷道支架位移动态数据建立了ARIMA模型。发现用该模型预报时,误差不仅与预报步数有关,还与差分次数有关。并阐明了还原误差问题,找出了差分次数影响误差的递推方法。

The ARIMA model built by time series analysis has been used to analyze the problem: how to set up a statistical relationship between the tunnel displacements and time, on the basis of a lot of in-situ observation data, and predict the future values of displacements of the runnel. In this paper, It has been found that predicfing error is related with not only predicting steps but also the differential times. The restoring error problem is expounded, and the recursive method for finding the errors affect ed by differential times has been found.