基于多变量灰色模型的路基沉降预测

Subgrade Settlement Prediction Based on Multivariable Grey Model

路基沉降变形是一个复杂的动态变化过程,常用的单变量预测模型不能考虑各沉降监测点间的相关性,不足以反映路基整体的变形规律。多变量灰色模型是单变量GM(1,1)模型在多元变量条件下的拓展,可以实现对路基中相互影响的多个监测点变形预测模型的建模和预测。实例计算表明,与单变量GM(1,1)模型相比,多变量灰色模型具有更高的预测精度,显示了该方法进行路基工后沉降预测的有效性。

Subgrade settlement is a dynamic and complex diversification process, the conventional single variable forecasting models can＇t consider the correlation of settlement between the discrete monitoring points, so that it can＇t represent the integrated deformation regularity of subgrade. A multivariable grey model, which is an extension of the single variable model named GM（1,1）, is introduced to solve the problem. A example result demonstrates that, compared with the single variable GM （1,1） model, the multivariate gray model has higher prediction accuracy and effectiveness for post-construction settlement of subgrade.