摘要
本文简要介绍了Minimax法和MSAR法对小样本数据的研究,随之重点介绍了1种实用的新算法IMSAR法及其应用。该法是通过变权松弛最小二乘法,确定适当的权因子来实现的。它克服了MSAR法的运算量随样本数据量增加而显著增加的突出缺点,仿真结果表明:IMSAR法具有较强的抗干扰能力,可适应于噪声分布未知的情况。最后,本文将IMSAR法应用于转炉炼钢动态模型及药物房室模型的建立,收到了令人满意的效果。
The paper briefly introduces the study of parameter estimation by small-sample observations. It first deals with the topic using the methods of Minimax and MSAR. Then it presents an improved algorithm i.e. the MMSAR'. This algorithm overcomes the disadvantage of Minimax and MSAR in which ths number of necessary operations involved increases greatly as the amount of sample data increases. It can appropriately be used under the condition when the noise distribution is unknown. Finally, the papar dis-closes that application of the MSAR method in the establishment of mathematical models for a LD converter plant and pharmacokinetic model of a medicine all showed very satisfactory results,
出处
《北京理工大学学报》
EI
CAS
CSCD
1989年第1期92-98,共7页
Transactions of Beijing Institute of Technology
关键词
小样本
参数估计
时变参数
建模
small sample, estimation, time varying, modelling.
