基于时间序列与神经网络的软岩隧道变形预测模型及其应用

Forecasting method of the deformation of soft rock roadways based on time series' analysis and BP neural networks and its application

针对软岩隧道具有变形量大、变形分析困难和稳定性差等特点以及传统的变形分析方法过于单一和精度低等问题,分别运用时序分析和神经网络方法对软岩隧道变形进行了预测,以单个方法的预测结果为基础,结合IOWHA算子,根据各方法的预测精度计算出它们在组合模型中的权重,建立了组合预测模型.通过工程监测获取原始数据,运用组合预测方法得到相应的预测结果,并将其与单个方法的预测结果进行了对比分析.研究结果表明:新的组合预测方法能够综合时序分析和神经网络方法的优势,预测结果精度明显提高,该方法的应用对具体软岩隧道的稳定性评价及隧道工程的施工与维护具有一定的指导意义.

Soft rock roadway has many characteristics, such as large deformation, diffi- culty in deformation analysis because of poor stabilities. Traditional deformation analy- sis methods are difficult overcoming their shortcomings, such as singleness, low accuracy and so on. The original data were obtained by project monitoring. The time series＇ analy- sis and BP neural networks＇ method were used to forecast the deformation of soft rock roadway, respectively. A combination forecasting model was established. Based on the prediction precision of forecasting results from the single method, the weight of each method was calculated according to IOWHA operator in the combination forecasting. The prediction results of the combination forecasting method and a single method were compared. The results indicate that the new combined forecasting method can have the advantages from the timing analysis and neural networks method, and the prediction precision of the combination forecasting are improved obviously. It is very meaningful for the stability evaluation ,construction and maintenance of soft rock roadway.