摘要
针对原始数据序列不满足ln(x)变换要求和变换后的数据序列不满足建模要求两种情况,提出了对原始数据序列进行ln(x+c)变换,先从理论上证明了此种变换可以使建模数据序列的光滑度提高,进而提高所建模型的预测精度,另外还给出了基于建模序列近似确定最佳常数c的方法。然后,选取满足建模要求的递增变形数据序列,通过实例模拟得到了实测值与预测值接近的结果,论证了此种变换的有效性及实用性。该方法弥补了原始数据序列和对数变换序列的不足,拓宽了灰色GM(1,1)模型的应用范围,使模型的拟合精度和预测精度均得到提高。
To solve the problems that the original data sequence does not satisfy function ln(x)transformation requirements and the data sequence after transformation does not satisfy the modeling requirements,in this paper,the method of function ln(x+c)(c>0)transformation against modeling data was put forward.Theoretically it has been proved that the smooth degree of data series for modeling after this transformation could be enhanced,and the GM(1,1)model precision based on the data transformation was also improved.A method based on modeling data sequence for approximatively?determining the best constant c was given.Additionally,this article selected deformation data sequences that were monotonically increasing and met modeling requirements.The results are obtained that the measured value is close to the predicted value.The practical application showed the effectiveness and practicability of the proposed approach.This method makes up the deficiency of the original data sequence and logarithmic transformation sequence,and broadens the application range of the grey GM(1,1)model,so that the fitting accuracy and the prediction accuracy of the model can be improved.
作者
邱利军
张波
张京奎
QIU Lijun;ZHANG Bo;ZHANG Jingkui(School of Civil Engineering,Hebei University of Architecture,Zhangjiakou 075000,China;Hebei Key Laboratory of Diagnosis,Reconstruction and Anti-disaster of Civil Engineering,Hebei University of Architecture,Zhangjiakou 075000,China;The 54th Research Institute of CETC,Shijiazhuang 050081,China)
出处
《人民长江》
北大核心
2020年第1期136-140,共5页
Yangtze River
基金
国家自然科学基金资助项目(91638203)
张家口市科技计划项目(1621013B)
张家口市科技计划项目(1821019B)河北建筑工程学院校科研基金项目(2016XJJYB04).
关键词
灰色GM(1
1)模型
函数变换
变形预测
数据序列
光滑度
grey GM(1,1)model
function transformation
deformation prediction
data sequence
smooth degree
作者简介
邱利军,男,讲师,硕士,注册测绘师,研究方向为变形监测数据处理。E-mail:qiuljun@126.com;通讯作者:张波,女,讲师,研究方向为工程结构抗震及变形研究。E-mail:qiuljun@126.com
