基于PLS和减聚类的模糊逻辑系统设计及应用  

Design of Fuzzy Logic Systems Based on PLS and Subtractive Clustering and Its Application

作　　者：贾润达[1] 毛志忠[1,2] 常玉清[1,2] 

机构地区：[1]东北大学信息科学与工程学院,沈阳110004 [2]东北大学流程工业综合自动化教育部重点实验室,沈阳110004

出　　处：《系统仿真学报》2010年第2期391-394,共4页Journal of System Simulation

基　　金：国家高技术发展计划(863)项目(2006AA060201)

摘　　要：结合PLS和减聚类算法设计了一种新型的模糊逻辑系统混合训练算法。通过将全部样本数据作为模糊规则选入系统,并引入PLS作为训练算法,从而有效的抑制样本数据中的严重噪声;为了使训练后的系统结构得以简化,利用减聚类算法合并系统中相似的规则。还通过仿真验证了该方法相比于传统OLS训练算法的优点;并将其应用于湿法冶金萃取组分含量软测量建模中,得到了满意的效果。A novel hybrid learning algorithm for fuzzy logic systems based on PLS and subtractive clustering algorithm was proposed. Through selecting all the sample data as fuzzy rules to establish systems, and introducing PLS as learning algorithm, serious noisy in the sample data was tolerated. For generating a more parsimonious system, subtractive clustering algorithm was used to merge similar rules. The merits of this learning algorithm which was compared with conventional OLS learning algorithm were validated through simulation, and it was applied to soft sensor modeling for the component content in extraction process of hydrometallurgy, and satisfactory results were achieved.

关 键 词：模糊逻辑系统 模糊基函数 偏最小二乘法 减聚类 软测量 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]