中文摘要：

传统沉降预测方法对于量值小、相对波动大的短期沉降不适用,为此提出适用于结构沉降反复剧烈波动的自修正短期动态预测方法.该方法首先对已获取的第1～n时刻的结构沉降数据进行多次累加,得到一条单调光滑数据累加曲线,然后对该曲线进行最高（n-1）次多项式拟合,外延得到第（n＋1）时刻的沉降预测累加值,再通过回归递推,即得到第（n＋1）时刻的沉降预测值.同时,对该方法引入“新陈代谢”思想,以更好地反映沉降变化趋势的目前特征.最后,将某盾构施工现场结构沉降预测值与实测值进行比较.结果表明,该方法能取得较好的预测结果.

英文摘要：

Traditional forecasting methods are proved to be difficult to forecast small and short-term fluctuant settlements in structures.This paper proposed a forecasting method for repeated fluctuant settlements based on multi-accumulation and polynomial fitting.Firstly,it cumulates the observed repeated structural settlements from No.1 to No.n to get a smooth curving data line.Secondly,a maximum n-1 power polynomial fitted function is used for fitting the data line to achieve the forecasting accumulative value on the time of No.n ＋ 1.Then the forecasting value on the time No.n ＋ 1 can be derived by multi-regressive recursion.Another characteristic of this method is self-correcting forecasting which helps to reflect the change tendency more accurately.At last,a comparison between forecasting and real settlements in a structure near an under-passing subway tunnel construction is carried out.The forecasting value was compared with test value for one shield settlement and the comparing results show that the proposed method can forecast the small and fluctuant settlements with reasonable accuracy.