计算机工程与应用 ›› 2014, Vol. 50 ›› Issue (12): 16-20.

• 博士论坛 • 上一篇    下一篇

使用新混合模糊聚类算法的模糊系统建模方法

周  頔1,孙  俊2,盛歆漪1   

  1. 1.江南大学 数字媒体学院 数字媒体技术系,江苏 无锡 214122
    2.江南大学 物联网工程学院,江苏 无锡 214122
  • 出版日期:2014-06-15 发布日期:2015-05-08

Fuzzy system modeling method based on novel hybrid clustering algorithm

ZHOU Di1, SUN Jun2, SHENG Xinyi1   

  1. 1.Department of Digital Media, School of Digital Media, Jiangnan University, Wuxi, Jiangsu 214122, China
    2.School of Internet of Things, Jiangnan University, Wuxi, Jiangsu 214122, China
  • Online:2014-06-15 Published:2015-05-08

摘要: 为了进一步提高模糊系统建立模型的精度,提出一种新的模糊系统算法ANFIS-HC-QPSO:采用一种混合型模糊聚类算法来对模糊系统的输入空间进行划分,每一个聚类通过高斯函数的拟合产生一个隶属度函数,即完成ANFIS系统的前件参数——隶属度函数参数的初始识别,通过具有量子行为的粒子群算法QPSO与最小二乘法优化前件参数,直至达到停机条件,最终得到ANFIS的前件及后件参数,从而得到满意的模糊系统模型。实验表明,ANFIS-HC-QPSO算法与传统算法相比,能在只需较少模糊规则的前提下就使模糊系统达到更高的精度。

关键词: 模糊系统, 自适应模糊推理系统(ANFIS), 混合聚类, 具有量子行为的粒子群算法(QPSO)

Abstract: This paper proposes a novel fuzzy system modeling algorithm ANFIS-HC-QPSO:a hybrid fuzzy clustering algorithm is used to divide the input space and every cluster generates a membership function by approximation to recognize the premise parameters of ANFIS roughly. The Quantum behaved Particle Swarm Optimization algorithm(QPSO) is applied with least square method to optimize the rough premise parameters until obtaining all the parameters of ANFIS. The experiments indicate that, compared with traditional methods, ANFIS-HC-QPSO can build the ANFIS model with less fuzzy rules, but much more precise.

Key words: fuzzy system, Adaptive Neuro Fuzzy Inference System(ANFIS), hybrid clustering, Quantum behaved Particle Swarm Optimization algorithm(QPSO)