估计GM(1,1)模型参数的一种新方法

Journal of Systems Engineering and Electronics ›› 2009, Vol. 31 ›› Issue (2): 471-474.

估计GM(1,1)模型参数的一种新方法

刘威¹, 崔高锋²

1. 中国飞行试验研究院, 陕西, 西安, 710089;
2. 中国人民解放军96626部队, 福建, 泉州, 362100

收稿日期:2007-10-20 修回日期:2008-04-16 出版日期:2009-02-20 发布日期:2010-01-03
作者简介:刘威(1982- ),女,助理工程师,硕士,主要研究方向为应用数学,灰色系统理论与应用.E-mail:wei182@126.com

New method for the estimation of GM(1,1) parameters

LIU Wei¹, CUI Gao-feng²

1. Chinese Flight Test Establishment, Xi'an 710089, China;
2. No.96626 Unit. of PLA, Quanzhou 362100, China

Received:2007-10-20 Revised:2008-04-16 Online:2009-02-20 Published:2010-01-03

摘要/Abstract

摘要： 考虑到最小二乘法则的不足及背景值参数和边值的影响,提出基于最小一乘准则估计GM(1,1)模型参数,得到新的预测公式,引入粒子群算法直接求解最小一乘问题即可得到模型参数,简化了以往改进模型的二次求解过程.数值计算结果表明,基于粒子群算法及最小一乘准则估计灰色模型参数,对于平稳或非平稳序列,都具有较高的拟合与预测精度.

Abstract: Considering the disadvantage of the least square criteria and the influence of background value parameter and boundary improvement condition to model prediction accuracy,the least absolute criteria is offered to estimate the parameters of GM(1,1),and the new grey prediction formula is proposed.The parameters are obtained by particle swarm optimization in one step,predigesting the usual process based on the two steps method.The Analysis of some examples indicates that the new method improves the precision of fitting and forecasting of the model for stable or unstable data sequence.

中图分类号:

N941.5

刘威, 崔高锋. 估计GM(1,1)模型参数的一种新方法[J]. Journal of Systems Engineering and Electronics, 2009, 31(2): 471-474.

LIU Wei, CUI Gao-feng. New method for the estimation of GM(1,1) parameters[J]. Journal of Systems Engineering and Electronics, 2009, 31(2): 471-474.

[1]	王建华, 查怡婷, 王雪, 熊峰. 基于核和灰度的灰色马尔可夫预测模型及应用[J]. 系统工程与电子技术, 2020, 42(2): 398-404.
[2]	王正新, 党耀国, 刘思峰, 练郑伟. GM(1,1)幂模型求解方法及其解的性质[J]. Journal of Systems Engineering and Electronics, 2009, 31(10): 2380-2383.
[3]	姚天祥, 刘思峰, 党耀国. 初始值优化的离散灰色预测模型[J]. Journal of Systems Engineering and Electronics, 2009, 31(10): 2394-2398.
[4]	吴彩华, 朱小冬, 刘俊涛, 王毅刚. 一种新的软件可靠性增长模型 [J]. Journal of Systems Engineering and Electronics, 2009, 31(8): 2024-2028.