大坝沉降变形的灰色预测分析研究

Study on modeling dam settlement by grey system method

用实测沉降资料预测大坝沉降过程是目前国内外大坝沉降变形监测分析中常采用的一种方法,包括双曲线拟合法、指数曲线拟合法和基于灰色理论的GM(1,1)模型预测方法。由于灰色理论在预测时不需要较长的观测数列,且能通过动态预测模型,最大程度的反映新信息在预测中的作用,因此在沉降变形预测方面具有很大的优势。本文针对目前大坝沉降变形GM(1,1)预测分析中存在的一些问题,根据大坝沉降变形的特点,分析了沉降观测数据列长度、观测时间的不等步长性、数据累加方式等对建模预测精度的影响,探讨了用灰色等维预测模型动态预测非稳态沉降变形过程的方法,提出了灰色等维动态预测模型的维数尺度标准。

Prediction of settlement by using observation data is a world-wide popular method for dam settlement monitoring analysis, which includes the hyperbolic curve method, exponential curve method and the grey system theory GM（1,1）method. The grey system method GM（1,1）method has great advantages in predicting settlement process because it does not need a long observation series, and it can utilize dynamic forecasting model to reflect the effect of new information with a maximum degree. Aiming at some problems of GM（1,1）, the authors analyzed the influence of observation series quantity, observation time-space and data summation method on settlement process and prediction accuracy. This paper investigated a method of forecasting non-steady-state settlement procedure with an equal-dimensional dynamic forecasting model and recommended the dimension scale standards for the equal-dimensional dynamic forecasting model.