摘要
在集对分析中,差异度系数i是体现系统不确定性的重要变量,但如何取值一直在理论上没有解决,本文根据i的取值的不确定性与模糊性,提出了一种基于模糊集值统计的差异度系数取值方法。对于原始数据中的噪声,本文提出了一种基于模糊软阈值的小波去噪方法,并利用支持向量回归方法来进行同异反预测,在一定程度上克服了线性建模技术的不足。此外,为了克服同一度、差异度与对立度之间的归一化约束,本文还提出了一种熵变换的方法。
In Set-pairs analysis, the coefficient of discrepancy degree i is an important variable which Embodies the uncertainty. However, how to ascertain the value of i is still an unsolved problem. According to the uncertainty and fuzzy characteristics of the value of i, the paper provides a kind of method of ascertaining the value of i based on fuzzy set-valued statistics. The paper also gives a kind of wavelet noise-erased method based on Fuzzy soft threshold for the noises existed in the original data and goes on prediction of identity, discrepancy and oppositeness with support vector regression, which improves the shortcoming of linear modeling technology at some extent. Moreover, the paper puts forward a kind of entropy transform in order to overcome the constraint of fixed summation.
出处
《模糊系统与数学》
CSCD
北大核心
2009年第3期56-60,共5页
Fuzzy Systems and Mathematics
基金
广东省自然科学基金资助项目(5013318)
关键词
集对分析
模糊软阈值
模糊集值统计
小波分析
支持向量回归
Set-pairs Analysis
Fuzzy Soft Threshold
Fuzzy Sot-valued Statistics
Wavelet Analysis
Support Vector Regression
