考虑地下水位影响的软弱路基沉降预估模型

Prediction Model for Settlement of Soft Subgrade Considering Groundwater Level Influence

摘要:

地下水位是影响地基沉降的重要因素，然而，目前常用的曲线拟合预估方法仅考虑时间与沉降的联系，忽略了地下水位变化的影响，导致沉降预估准确性受限。以江苏省某高速公路的软弱路基为研究对象，通过埋设传感器监测地下水位与沉降情况，深入探讨了地下水位变化对软弱路基沉降规律的影响。基于BP神经网络，构建了包含时间、地下水位及沉降量的关系模型。与实测数据及传统曲线拟合模型对比发现，BP神经网络模型的沉降预估值与地下水位呈正相关，且预估误差平均仅为0.17%，相比传统方法精度提高了3.58%。研究结果表明：BP神经网络沉降预估模型能有效考虑地下水位的影响，显著提升软弱路基沉降预估的准确性。

Abstract:

The groundwater level is a key factor influencing the subgrade settlement. However， the conventional curve-fitting prediction method only considers the relationship between time and settlement， and neglects the impact of groundwater level fluctuations， which limits the accuracy of settlement predictions. Taking the soft subgrade expressway in Jiangsu Province as the study object， the groundwater level and settlement are monitored by embedding the sensors to deeply discuss the effect of groundwater level variations on the settlement rule of soft subgrade. Based on a BP neural network， the relationship model containing the time， groundwater level and settlement is built. Comparing the measured data with the traditional curve-fitting model， it is found that the settlement prediction value of BP neural network model is positively correlated with the groundwater level， and the average prediction error is only 0.17%. Compared with the traditional method， the precision is improved by 3.58%. The study result shows that the BP neural network settlement prediction model can effectively consider the influence of groundwater levels， and the accuracy of settlement prediction of soft subgrade is greatly improved.