摘要
本文提出了可构成多规则模糊神经网络的CMOS模拟单元电路,包括:类Gauss型隶属度函数电路,电压求小电路和重心算法去模糊电路.基于这些电路设计了一个两输入/一输出、25条规则的控制系统,并通过非线性函数逼近进行了验证.所有单元均采用SMIC 0.18-μm CMOS数模混合工艺制造,芯片测试结果表明:提出的单元电路结构简单,输出电压偏差小,便于扩展和调节;因而适于实现多规则,自适应调节的高速高精度控制系统.
This paper proposes several improved CMOS analog circuits for neuro-fuzzy network, including Gaussian-like membership function circuit, minimization circuit, and a centroid algorithm defitzzier circuit without using division. A two-input/oneoutput neuro-fuzzy network composed of these circuits is implemented and testified for non-linear function approximating. All the circuits have been fabricated in SMIC 0.18-μm CMOS technology. Experiment results show that all the proposed circuits provide characteristics of high operation capacity ,high speed, and simple structures. They are very suitable for rapid implementation of highspeed complex neuro-fuzzy networks.
出处
《电子学报》
EI
CAS
CSCD
北大核心
2007年第5期946-949,共4页
Acta Electronica Sinica