摘要
针对实际测量数据中噪声对模糊建模规则的影响,研究了基于矩阵正交变换的模糊规则选取方法。分别利用正交最小二乘和奇异值分解,分析模糊规则对辨识结果的贡献度大小。舍去贡献度较小的模糊规则,得到模糊规则中的有效信息,减少了模糊规则数。分别用Mackey-Glass混沌数据验证了它们的有效性和实用性。
In according with the affect of noise in actual measured data for fuzzy modeling, fuzzy rules selection methods based on matrix orthogonal transformation are investigated in this article. This paper make use of orthogonal least squares and singular value decomposition, which analyze the contribution for identification result. The algorithm will discard the small contribution fuzzy rules, obtain valid information in fuzzy rules and reduce fuzzy rules. Mackey-Glass chaos datas are utilized to demonstrate the performance and application of these algorithms.
出处
《仪器仪表用户》
2016年第2期8-10,7,共4页
Instrumentation
关键词
正交变换
模糊辨识
正交最小二乘
奇异值分解
orthogonal transformation
fuzzy identification
orthogonal least squares
singular value decomposition
