摘要
本文提出了用单向逐步回归方法确定GMDH中最佳部分多项式结构和进行参数估计,并且给出了递推算法。这种算法避免了原算法中的大量重复运算,对多步预测模型更显出其优越性。实际应用例子证明了该算法的有效性。
In this paper a unidirectional stepwise regression algorithm is presented to determine the structure of the optimal partial polynomial and estimate parameters in GMDH. In order to avoid intensive iterative computation, a recursion algorithm is proposed. The algorithm is particularly suited to multistep predicted modelling. Application examples have shown its effectiveness.
出处
《自动化学报》
EI
CSCD
北大核心
1990年第4期310-316,共7页
Acta Automatica Sinica
关键词
非线性系统
系统辨识
GMDH
算法
Systems identification
GMDH
prediction
parameter estimation
nonlinear systems
