一个具有完备性和鲁棒性的模糊规则提取算法

从实际检测数据中提取模糊规则进而建立有效的模糊模型对实现复杂系统的智能建模与控制具有重要意义. 在一些文献中对该问题进行了较深入的研究, 并提出了有效的从数值数据中提取模糊规则的算法(简称为WM 和iWM算法). 对WM和iWM算法的进一步分析研究表明, 该算法在完备性和鲁棒性方面还有进一步改进的可能. 本文采用数据挖掘技术提出一个改进的提取模糊规则的算法(简称DM 算法), 并在完备性和鲁棒性方面与WM 和iWM算法进行了比较研究. 模糊建模实例表明, 本文提出的DM算法具有更好的逼近能力和对不确定数据干扰的鲁棒性.

Extraction of Fuzzy Rules with Completeness and Robustness

Extraction of fuzzy rules from numerical data for fuzzy modeling and control is significant. Such a problem has received considerable attention in the past and some algorithms, termed as the WM algorithm and the iWM algorithm, have been proposed in the literature. Research on the WM algorithm and the iWM algorithm showed that some improvements on robustness and completeness of these algorithms could be done. This paper aims to develop an improved fuzzy rule extraction algorithm (termed as the DM algorithm) using data mining techniques, and the completeness and the robustness of rule-base for fuzzy modeling with noisy data are addressed. Some illustrative examples are given. Results demonstrate that our proposed rule extraction algorithm outperforms the WM algorithm and iWM algorithm in terms of both modeling accuracy and robustness with respect to noisy data.