大坝变形度的不等维加权动态GM(1,1)预测模型

Multidimensional Weighted Dynamic GM(1,1) Model Applied in the Prediction of Dam Deformation Degree

针对灰色GM(1,1)模型预测结果易受模型中以前测得的陈旧数据的干扰,及等维动态GM(1,1)受缚于维数选择的情况,给出了不等维加权动态GM(1,1)模型的基本内容及建模过程,模型中计算出多种维数的GM(1,1)模型的预测值,并且通过萨函数加权法和BP神经网络计算出每种维数的权值,通过加权获得最终预测值。并且成功地将不等维加权动态GM(1,1)模型应用于大坝变形度的预测预报。实践证明,不等维加权动态GM(1,1)模型由于考虑了维数对模型结果的影响,而且及时地更新数据,提高了灰区间的白色度,预测效果比传统的GM(1,1)模型和等维动态GM(1,1)模型效果好。

The prediction result of GM（1,1） grey model is subject to be disturbed by outdated information previously measured in the system,while one-dimensional dynamic GM（1,1） model is restrained by the selection of the dimension.To overcome these problems,this paper studies the content and the modeling of Multidimensional Weighted Dynamic GM（1,1） model（MDWD-GM（1,1） model） in detail.Based on the prediction results of all the dimensions calculated by this model,the weight value of each dimension is obtained by Sa function weighting method and BP neural network,then the final predictive value is obtained by weighting.Moreover,the MDWD-GM（1,1） model has been applied to the dam monitoring system and the application manifests that it offers better prediction results than traditional GM（1,1） model and one-dimensional dynamic GM（1,1） model as it takes the effect of different dimensions into account and increases the white degree of the grey range by updating the data in time.