模糊系统建模方法及其在微生物发酵中的应用  

Novel hybrid ANFIS and its application in microbial fermentation engineering

作　　者：周頔[1] 孙俊[2] 

机构地区：[1]江南大学数字媒体学院,江苏无锡214122 [2]江南大学物联网工程学院,江苏无锡214122

出　　处：《计算机工程与设计》2014年第11期3974-3979,共6页Computer Engineering and Design

基　　金：江苏省自然科学基金项目(SBK201341929);江南大学自主科研基金项目(1232050205120960)

摘　　要：提出一种具有量子行为的模糊系统建模方法。避免事先指定聚类数目及中心,采用混合模糊聚类算法对模糊系统的输入空间进行划分,每个聚类通过高斯函数的拟合产生一个隶属度函数,完成ANFIS前件参数的初始识别;通过具有量子行为的粒子群算法与最小二乘法优化前件参数,得到ANFIS的前件及后件参数。将该方法应用于实际的抗坏血酸2-葡萄糖苷生产发酵模型的建立中,实验结果表明,该方法具有较高精度,符合实际生产需要。A novel fuzzy system modeling algorithm, namely the hybrid quantum-behaved ANFIS was proposed. A hybrid fuzzy clustering algorithm was used to divide the input space, and every cluster generated a membership function by approximation tb recognize the premise parameters of ANFIS roughly. Then the quantum behaved particle swarm optimization algorithm was ap- plied with the least square method to optimize the rough premise parameters, and all the parameters of ANFIS were obtained. The hybrid quantum-behaved ANFIS was applied to real microbial fermentation experiments. The results indicate that the pro- posed method is much more precise than the traditional ANFIS.

关 键 词：自适应模糊推理系统 混合聚类 具有量子行为的粒子群算法 模糊系统 微生物发酵 生产预测 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]