摘要
灰色模型的预测精度很大程度上依赖于背景值的构造方法,以往的背景值构造方法主观地认为新旧数据对背景值的贡献和为1,无法真实反应新旧信息对背景值贡献的大小,而基于任意加权改进的RGM(1,1)模型又忽略了各个背景值构建时权值不同的情况。针对以往模型的缺陷,该文提出一种基于改进的任意权值背景值优化方法的GAGM(1,1,p,q)模型。结合遗传算法、运用MATLAB编程语言实现了改进灰色模型的预测程序。将改进的模型应用于边坡表面变形预测,取得了较好的效果。将预测结果与传统GM(1,1)模型及任意权值改进的RGM(1,1)模型的预测结果作对比,结果表明,文中提出的改进模型具有更高的拟合和预测精度,可应用于工程实践。
The prediction accuracy of gray model is largely dependent on the construction method of background values. The past methods of background values construction subjectively consider that the new and old data makes contribution to background values for one, it can not accurately reflect the contri-bution size that the new and old data making to the background values. And the RGM ( 1,1 ) model ig-nored the different weights of various background values. Taking the defects of previous model into ac-count in the paper, the GAGM ( 1, 1, p, q) model , an improved background values optimization meth-od by using random weight is proposed. Combined with genetic algorithms, the prediction program of im- proved grey model is realized by using MATLAB programming language. The proposed method is applied to predict surface deformation of the slope, and obtained good results. The predicted results are compared with the results by traditional GM ( 1, 1 ) model and any weighted modified RGM ( 1, 1 ) model, it shows that the improved model has higher fitting precision and prediction accuracy. The modified GAGM ( 1, 1, p, q) model can be applied to engineering practice.
出处
《勘察科学技术》
2013年第5期30-35,共6页
Site Investigation Science and Technology
关键词
灰色模型
GAGM(1
1
p
q)模型
背景值优化
遗传算法
边坡变形预测
gray model
GAGM ( 1,1 ,p, q) model
background values optimization
genetic algorithms
slope deformation prediction