Box-Cox变换及其在顶管施工引起地面沉降分析中的应用

Application of Box-Cox transformation in analyzing the ground subsidence induced by pipe jacking

由于实际顶管施工中引起的地面沉降问题的复杂性,实际监测数据无法严格满足函数线性、误差独立正态分布等条件,使得利用多元线性回归方法建立地面沉降分析模型的理论依据不够严密,导致结果精度较低。本文基于Box-Cox变换理论,首先提出对实际沉降监测数据进行正态变换预处理,然后再进行传统多元线性回归分析。最后,实例结果表明,计算和预测精度都有明显提高,且更符合实际情况。

The ground subsidence caused by pipe jacking is very complicated ; therefore, the actual monitoring data can not meet the conditions of strict linear functions, independent normal errors and so on. Also, the theoretical basis of ground subsidence model based on multiple linear regression is not enough strict, and the accuracy of the result is poor. This paper gives a normal transformation pretreatment for the actual subsidence monitoring data based on Box-Cox transformation theory. Then, the traditional multiple linear regression analysis on the processed data is taken. In this way, the results show that both the calculating and the forecasting accuracy have been well improved, and can meet the real situation.