摘要
本文采用基于模糊聚类的模糊神经网络模型对系统进行辨识,首先利用模糊聚类技术来确定系统的模糊空间和模糊规则数,然后利用模糊神经网络来调整模型的前件参数和后件参数。用此设计方法对函数逼近问题进行仿真,结果表明利用聚类技术可以获得较好的初始值,学习速度快、建模精度高。
This paper presented a new fuzzy neural network model based on fuzzy clustering to identify system, fuzzy space and a number of fuzzy rules is established by fuzzy clustering, then parameters of the conditional and consequent part of rules will be optimized by fuzzy neural network. Finally we take example of nonlinear approximation, simulation results show that good initial values can be obtained by clustering and calculation can be speeded up to achieve satisfactory results.
出处
《甘肃冶金》
2006年第4期6-8,共3页
Gansu Metallurgy
关键词
模糊聚类
模糊神经网络
系统辨识
fuzzy clustering
fuzzy neural networks
system identification
