摘要
用美国国家标准和技术研究院 (NIST)公布的各种类型数据 ,检验缩张算法拟合非线性方程的可行性 .结果表明 :改进缩张算法的反馈系统 ,可方便和稳健地实现非线性模型的最优参数估计 .经检验还发现NIST公示结果尚有个别问题未达全局最优 ;由缩张算法获得的参数估值与NIST给出的估值基本一致或比之更优 ;用缩张算法实现全局最优拟合的难度和NIST的划分不尽相同 ;该算法对多自变数非线性方程同样有效 .
The feasibility and precision test of an algorithm to optimal fitting of nonlinear equation was carried out by contractionexpansion algorithm, using NIST datasets. The results showed that with improved feedback system, the algorithm could reach optimal fitting for all testing problems. It also revealed one NIST result was not optimum. The parameter estimates were a little different from certified values, usually with higher precision for most of them. And the difficulty level to optimal solution was also not similar with NIST classification. It also could be used to cope with multivariate nonlinear regression.
出处
《扬州大学学报（自然科学版）》
CAS
CSCD
2001年第3期16-19,共4页
Journal of Yangzhou University：Natural Science Edition
关键词
缩张算法
非线性方程
全局最优
参数估计
拟合
contraction\|expansion algorithm
nonlinear regression
global optimum
parameter estimation
