摘要
本文讨论了指数自回归模型的辨识问题,证明了该模型的最小二乘问题的非凸性,并给出其保证凸性的条件,最后运用混合算法,辫识了该模型,并用数值算例加以说明。
Identification of exponential autoregressive model(EAR model) was discussed, It wasproved that the least square problem is not convex and a condition was given when the varince of resid-ual of the least square estimate is convex. A method of identification on EAR model with hrbridmethod was given and explained with a numerical example。
出处
《东南大学学报(自然科学版)》
EI
CAS
CSCD
1994年第4期96-100,共5页
Journal of Southeast University:Natural Science Edition
基金
国家自然科学基金
国家教委博士点基金
关键词
非线性规划
指数
自回归模型
time series analysis,nonlinear programming/exponential autoregressive model,AIC cri- terion,globl minimization,local minimization, hybrid method