考虑蠕变模型参数空间变异性的高贴坡变形预测

打开文本图片集

中图分类号：TU42 文献标志码：A 文章编号：1000-4939（2025）04-0850-10

Abstract：Aiming at the problem that the spatial variability of soil property parameters is seldom considered in the existing slope deformation prediction，a stochastic prediction method for high sticking slope deformation is proposed.By analyzing the one-dimensional consolidation creep characteristics of loess，a semi-theoretical and semi-empirical creep model was constructed to describe its deformation characteristics.According to the consolidation creep variability of loess under repeated tests，the cross-correlation Gaussian random field model of creep model parameters is established by using K-L series expansion method，and combined with ABAQUS secondary development，MCS-SFEM was used for deformation prediction. When the deformation of a high stick slope is predicted， 89.06% of the measured deformation at the monitoring point is within the predicted deformation range，and the RMSE，MAPE and R2 of the predicted mean curve are close to that of the hyperbolic method，which is slightly larger than that of the BP neural network model.Moreover，it is beter than the hyperbolic method to consider the deformation nonlinearity caused by gravity changes during the construction period，and take the post-construction deformation nonlinearity into account by the deformation limit.The extension prediction is also more consistent with the main deformation trend，which proves the rationality and reliability of the forecast method.

Key words：high sticking slope；creep model; spatial variability；MCS-SFEM; deformation predictior

在西部大开发战略实施过程中，出现了许多黄土高填方边坡。（剩余14522字）