隧道围岩变形的非线性自回归时间序列预测方法研究

发布时间：2019-04-09 15:08

【摘要】：针对传统时间序列预测模型的单一线性和忽略施工过程影响的静态局限性,提出非线性自回归(包括NARNN与NARXNN)时间序列预测模型.该模型通过引入动态施工影响因子作为附加的外部输入,同时结合模型本身的反馈结构和延迟单元,在结构和动态特性上更加符合实际系统,可以非线性动态地考虑隧道施工全过程.运用该模型对史家山2号隧道施工过程中的围岩水平收敛和地表变形进行预测.结果表明:1)非线性自回归预测模型比传统的ARMA预测模型的预测精度高、适应性好;2)通过多次预测并对结果取平均值,可以保证非线性自回归预测模型预测结果的预测精度和稳健性;3)通过优化动态施工影响因子的取值方法,可以进一步提高NARXNN时间序列预测模型的预测精度.
[Abstract]:The nonlinear auto-regression (including the NARNN and the NARXNN) time series prediction model is proposed for the static limitation of the single linear and neglected construction process of the traditional time series prediction model. In this model, the dynamic construction influence factor is introduced as an additional external input, and the feedback structure and the delay unit of the model itself are combined, so that the whole process of the tunnel construction can be considered in a non-linear dynamic manner in a more practical system in terms of structure and dynamic characteristics. The model is used to predict the horizontal convergence and surface deformation of the surrounding rock during the construction of the No.2 tunnel of Shijiashan. The results show that:1) The non-linear auto-regression prediction model is more accurate than the traditional ARMA prediction model, and the adaptability is good;2) The prediction accuracy and the robustness of the prediction result of the nonlinear auto-regression prediction model can be guaranteed by the multiple prediction and the average value of the results. And 3) the prediction accuracy of the NARXNN time series prediction model can be further improved by optimizing the value method of the dynamic construction influence factor.
【作者单位】：北京交通大学城市地下工程教育部重点实验室;中国中铁工程设计咨询集团有限公司;
【基金】：中央高校基本科研业务费专项资金(2016YJS116) 国家自然科学基金(U1234210)~~
【分类号】：U451.2

【相似文献】