摘要
为了提升差分自回归移动平均模型ARIMA拟合的精确程度,把解决非线性无约束问题的谱共轭方向思想运用到模型参数优化估计中.给出一种改进的谱共轭梯度法,即结合不同谱共轭梯度法的优势之处,提出新的参数标量和搜索方向迭代公式.理论上证明该算法的充分下降性和全局收敛性,数值实验结果验证其是一种更为快速有效的方法,实例分析进一步证实本文算法的可操作性.
In order to improve the fitting accuracy of difference auto regressive moving average model (ARIMA) and apply the idea of spectral conjugate direction, which is often used to solve nonlinear uncon strained problems, to optimum estimation of model parameters, an improved spectral conjugate gradient method is given, where the advantages of different spectral conjugate gradient methods are employed to bring forward new parameter scalar and iterative formula of searching direction. The sufficient descent property and global convergence of this algorithm are proved theoretically. It is verified by the result of nu meric experiment that it will be faster and more effective method and it is further verified by the analysis of result of and the practical example that the algorithm presented in this paper will be practicable.
作者
单锐
王国芳
黄威
刘文
王美霞
SHAN Rui;WANG Guo-fang;HUANG Wei;LIU Wen;WANG Mei-xia(College of Science,Yanshan University,Qinhuangdao 066004,China)
出处
《兰州理工大学学报》
CAS
北大核心
2018年第4期152-156,共5页
Journal of Lanzhou University of Technology
基金
国家自然科学基金(51405424
51675461
11673040)
关键词
ARIMA模型
谱共轭梯度法
全局收敛
参数估计
ARIMA model
spectral conjugate gradient method
global convergence
parameter estimation