新型神经模糊系统及其在乙烯裂解炉建模中的应用  

The New Neural-fuzzy System and Its Application in Ethylene Cracking Furnace Modeling

作　　者：段向军[1] 朱跃[1] 陈乾伟[1] 

机构地区：[1]南京信息职业技术学院机电工程系,江苏南京210046

出　　处：《石油化工自动化》2007年第1期39-41,共3页Automation in Petro-chemical Industry

摘　　要：为满足实际应用的需要,提出一种带Laplace(概率密度函数)型隶属函数的模糊逻辑系统(FLS),它具有多层前向网络的结构。但这个新型FLS不满足Stone-Weierstrass定理条件。笔者采用微分中值定理和Weierstrass定理证明它的通用逼近性。用所提出的神经模糊系统(NFS)对乙烯生产装置进行建模,预测乙烯产率,得到较高的拟合精度。实际应用表明了它的可行性和有效性。In order to satisfy actual requirement,a new fuzzy logic system（FLS） with Laplace （probability density function） membership function is proposed. It has the structure of multi-layer forward network. But this new FLS can not satisfy the conditions of Stone-Weierstrass theorem. Its universal approximation has been proved by using differential intermediate value theorem and Weierstrass theorem. The NFS has successful application in ethylene plant modeling problem. This model is used to forecast the ethylene yield ratio and gets high fitting precision. The actual applications show its feasibility and availability.

关 键 词：神经网络 模糊逻辑系统 神经模糊 建模 乙烯生产装置 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]