基坑开挖诱发侧方隧道变形的PCA-MLR预测方法

Prediction of Displacement of Lateral Tunnel Induced by Foundation Pit Excavation Based on the PCA-MLR Model

研究基坑开挖对侧方隧道变形的影响是岩土工程领域的热点与难点问题。为预测基坑开挖诱发临近侧方隧道变形,首先通过主成分分析方法(PCA)对侧方隧道变形指标影响因素进行降维处理并获取其主成分指标,然后采用多元线性回归方法(MLR)建立隧道变形指标与主成分指标间的关系模型,最后基于工程实例验证提出的PCA-MLR预测模型的正确性。研究结果表明:PCA-MLR模型考虑了基坑开挖诱发侧方隧道变形的高维度影响因素耦合作用,且实现了低维度主成分指标与变形指标间的关系模型建立。采用PCA-MLR模型预测的隧道最大水平及竖向位移与实测值的相对误差均小于10%,验证了提出模型的正确性与适用性。

The influence of foundation pit excavation on the displacement of adjacent tunnel has been a hot and difficult issue in the geotechnical engineering. In order to predict the displacement of lateral tunnel induced by foundation pit excavation, the dimensionality of the influencing factors on the displacement of lateral tunnel is reduced and the principal components of that are obtained based on the Principal Component Analysis(PCA). Then, the relationship between the principal component with low-dimensional and the displacement of lateral tunnel is established using Multiple Linear Regression(MLR). Lastly, the accuracy of PCA-MLR prediction model is verified based on the actual projects. The results show that the coupling effect of the high-dimensional influencing factors on the displacement of lateral tunnel induced by foundation pit excavation was considered. Besides, the low-dimensional principal components were used to established the MLR model. The maximum horizontal and vertical displacement of lateral tunnel induced by foundation pit excavation were predicted using PCA-MLR model. The relative error between predicted and monitoring values were less than 10%, indicating the accuracy and applicability of the developed model.