基于最小二乘支持向量机的公路软基沉降预测

Predicting Settlement of Highway Soft Foundation Using Least Squares Support Vector Machine

通过影响因素分析,确定了软土层厚度、软土层压缩模量、地表硬层厚度、地表压缩模量、路堤高度、路堤顶宽、路基填筑时间和填筑竣工时沉降量等参数对公路软基沉降有影响。对公路软基的观测数据进行分析和取样,输入样本为各参数,输出样本为路堤中线下地表沉降值,利用最小二乘支持向量机的非线性映射和泛化能力,通过训练,建立了公路软基沉降预测模型。研究表明,所建立的模型对公路软基沉降进行预测具有较高的精度,同时具有很好的泛化性能。

Through analyzing factors, eight parameters may influence settlement of highway soft foundation, which are thickness of soft foundation, modulus of compressibility of soft foundation, thick- ness of surface hard formation, modulus of compressibility of surface, height of highway embankment, top width of highway embankment, time span of highway embankment reclamation and settling height after highway embankment reclamation. The Least Squares Support Vector Machine can be trained using a small set of parameters obtained from test data which may influence settlement of high-way soft foundation. The input layer includes parameters mentioned above and the out layer is settlement of highway soft foundation. The trained Least Squares Support Vector Machine can map these parameters and settlement of highway soft foundation. A numerical example is given. The results show that the accuracy of settlement of highway soft foundation predicted by the presented approach is validated by comparison with test data. The relationship between settlement of highway soft foundation and parameters can be found through parametric analysis, which has reinforcement mapping capabilities.