改进灰色模型在地基沉降变形的预测应用研究

Modified grey model for prediction of foundation settlement deformation

沉降变形监测与预测是施工的重要环节,然而因为成本及监测点破坏等客观情况,沉降数据往往出现不完整或时间间隔不固定的问题,导致预测精度的下降。本文针对普通灰色预测模型的局限性进行了改进,并结合某码头项目深厚软黏土层的沉降监测数据,对其进行了验证,结果表明:1)改进与普通灰色预测模型的相关系数R^(2)均大于0.9,平均误差率都小于15%。2)相较于普通灰色预测模型,改进灰色预测模型预测曲线的中后段更贴近沉降监测数据,模型整体精度较高,证明了新模型的优越性。3)S17预测精度最差,可能由于沉降数据过少,且初始时间间隔较大。建议沉降监测应在成本范围内尽量频繁和完整。

Settlement deformation monitoring and prediction is the important aspects of building.However,because of finite cost and damage of monitoring points,settlement data often have incomplete or irregular time intervals,leading to worse predictions.In this paper,the limitation of the grey model is improved and validated with the settlement data of the deep soft clay layer of a cruise terminal project.The results show that:1)the R^(2) of both the improved and the common grey models are greater than 0.9,and the average error is less than 15%.2)the middle and latter part of the prediction curve of the improved grey model is closer to the settlement monitoring data than that of the common grey model.The accuracy of the improved model is higher,which proves the superiority of the new model.3)S17 has the worst accuracy,which may attribute to little settlement data and larger initial time interval.It is recommended that settlement monitoring should be as frequent and complete as possible within the cost range.