摘要
根据补偿模糊神经网络的建模特点,提出了基于聚类和文化算法的补偿模糊神经网络建模方法。该网络的学习分为两步:结构辨识和参数辨识。在结构辨识中,采用改进的聚类算法确定模糊规则数及初始参数,构造一个初始模糊模型;在参数辨识中,采用基于多层信念空间的文化算法对具有5层结构的补偿模糊神经网络参数进一步优化,使其具有更高的精度。通过对TE过程的故障诊断建模,结果表明该网络在建模精度和收敛速度上均优于常规补偿模糊神经网络和常规模糊神经网络。
According to the characteristic of compensatory fuzzy neural networks（CFNN）, this paper proposes a compensatory fuzzy neural network based on clustering and cultural algorithms. The identification of the proposed network is composed of two steps：structure identification and parameter identification. In the process of structure identification, the improved clustering method is used to obtain the number of inference rules of fuzzy model and the initial parameters in order to construct the initial fuzzy model; During parameter identification, cultural algorithm based on multilayer belief spaces is used to optimize the parameters of the compensatory fuzzy neural network with five layers so that the model obtains a higher accuracy. Finally, the result of the fault diagnosis modeling of TEP shows that the proposed network is superior to the conventional FNN and CFNN in modeling precise and convergence rate.
出处
《华东理工大学学报（自然科学版）》
CAS
CSCD
北大核心
2009年第2期302-307,共6页
Journal of East China University of Science and Technology
基金
国家863计划课题(2007AA04Z171)
