基于中值滤波的灰色预测模型及其在大坝变形预测中的应用

The gray forecast model based on median filter and its application in dam deformation forecast

在大坝变形监测中,当用GM(1,1)模型对稳定变化的变形数据序列进行预测时,效果较好。但是,影响坝体变形的因素多种多样,且处于动态变化之中,观测数据中将不可避免地存在着一些随机扰动,这些扰动使大坝的变形曲线发生异常波动。此时仅用GM(1,1)模型进行预测,其精度和可靠性就会下降。为此,本文提出一种基于中值滤波的GM预测模型,即先用中值滤波算法对发生波动的原始变形监测数据进行滤波处理,而后再建立GM模型进行灰色预测。实例证明,基于中值滤波的GM预测模型可以有效地提高大坝变形的预测精度。

In dam deformation monitoring,it can get good effects when the GM(1,1)model is used in the forecast of steady changed deformation data.However,the influencing factors of dam deformation are very various and in dynamic processes.Thus some random errors will occur inevitable and make the deformation curve undulate abnormally every now and then.In this situation,the forecast precision and reliability of GM(1,1)model will decline.Therefore,this paper puts forward the GM forecast model based on median filter.This model uses median filter arithmetic to filter the original data of dam deformation monitoring firstly,then builds the GM model to forecast the dam deformation.It is proved through the example that the GM forecast model based on median filter could effectively improve the forecast precision.