基于Sugeno型模糊神经网络的双模控制器设计  

Design of a double model temperature controller trained by Sugeno fuzzy neural network

作　　者：张皓 高瑜翔[1,2] 唐军 黄天赐[1,2] 马腾 吴美霖 ZHANG Hao;GAO Yuxiang;TANG Jun;HUANG Tianci;MA Teng;WU Meilin(College of Communication Engineering,Chengdu University of Information Technology,Chengdu 610225,China;Meteorological Information and Signal Processing Key Laboratory of Sichuan Education Institutes,Chengdu 610225,China;College of Electronic Information and Artificial Intelligence,Yibin Vocational and Technical College,Yibin 644000,China)

机构地区：[1]成都信息工程大学通信工程学院,四川成都610225 [2]气象信息与信号处理四川省高校重点实验室,四川成都610225 [3]宜宾职业技术学院电子信息与人工智能学院,四川宜宾644000

出　　处：《中国测试》2021年第10期129-136,共8页China Measurement & Test

基　　金：四川省教育厅高校创新团队项目(15TD0022)。

摘　　要：针对如何实现一个控制器控制两种不同参数的被控对象,提出基于一阶Sugeno型模糊神经网络训练的双模控制器方法。采取一个网络训练两种被控对象模型,将训练好的网络作为控制器,使其在控制双模型时,具有良好的控制性能。Matlab仿真结果表明一阶Sugeno型模糊神经网络训练的双模控制器,在控制两种被控模型时,都具有最低的超调量和最短的调节时间,稳定性最强,综合性能指标最好,能实现双模自适应,满足控制要求。Aiming at how to realize a controller to control the controlled object with two different parameters,a double model controller method based on first-order Sugeno fuzzy neural network training is proposed.One network is used to train two controlled object models,and the trained network is used as a controller,so that it has good control performance when controlling dual models.Matlab simulation results show that the double model controller trained by the first-order Sugeno fuzzy neural network has the lowest overshoot and the shortest adjustment time when controlling the two controlled models,the strongest stability and the best comprehensive performance indicators,it can realize double model self-adaptation and meet the control requirements.

关 键 词：一阶Sugeno型 模糊神经网络 双模控制器 MATLAB 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]