According to the nonlinear relationship between subgrade settlement and observation time,the accuracy of the traditional least squares parameter estimation is not high,so a power polynomial subgrade settlement prediction model with strong approximation ability is established, of which the model coefficients are estimated by quantile regression.Engineering example shows that the prediction based on the quantile regression estimates of power polynomial model has high accuracy,which is better than the least squares estimation of power polynomial prediction model and multi variable grey forecast model.This model can provide a new method for settlement prediction.%针对路基沉降与观测时间存在非线性关系,且传统最小二乘参数估计精度不高的问题,建立具有较强逼近能力的幂多项式路基沉降预测模型,并用分位数回归估算模型系数.工程实例表明,基于分位数回归估计的幂多项式预测模型具有较高的精确度,优于最小二乘估计的幂多项式预测模型和多变量灰色预测模型,为沉降预测提供一种新方法.
展开▼
