基于稳健GM(1,1)模型的基坑变形监测数据处理方法

A Method of Processing Deformation Monitoring Data of the Foundation Pit based on the Robust Model GM(1,1)

由于基坑地质及观测环境的多样性和复杂性,基坑位移监测数据处理的理论与实际模型之间存在着模型误差。若模型误差显著,则所得到位移拟合模型的精度较差,将影响后继变形分析的准确性。针对这一问题进行了深入研究,提出了基于稳健估计的GM(1,1)模型,将最小一乘法应用于微分方程变量参数求解,建立了稳健灰色模型。实际应用表明,稳健灰色模型比常规灰色模型具有更好的抗干扰性能和受异常点影响较小的优点,对基坑变形预测有较好的参考实用价值。

Due to the diversity and complexity of geological conditions and observation environment of the foundation pit,errors between theoretical and practical models with respect to monitoring movement data processing of the pit occur unavoidably,and when the error is significant,the precision obtained from fitting model of the movement would be low,thus affecting accuracy of subsequent deformation analysis.In the paper,in the light of such problem,the robust estimation-based model GM(1,1) is put forward,in which the least absolute deviation is used to solve variable parameters of the differential equation,and then the gray robust model is established.Actual practice indicates that,compared with the conventional gray model,the robust gray model features better anti-interference property and less influence imposed by the singular points,and is of good practical value for predicting deformation of the foundation pit.