北京航空航天大学学报 ›› 2006, Vol. 32 ›› Issue (09): 1072-1076.

• 论文 • 上一篇    下一篇

基于PNN与FNN模型神经网络控制器设计与分析

王锴, 王占林, 付永领, 祁晓野   

  1. 北京航空航天大学 自动化科学与电气工程学院, 北京 100083
  • 收稿日期:2005-10-20 出版日期:2006-09-30 发布日期:2010-09-19
  • 作者简介:王 锴(1973-),男,江苏南京人,博士生, wangkai@asee.buaa.edu.cn.

Design and analysis of based on PNN and FNN controller

Wang Kai, Wang Zhanlin, Fu Yongling, Qi Xiaoye   

  1. School of Automation Science and Electrical Engineering, Beijing University of Aeronautics and Astronautics, Beijing 100083, China
  • Received:2005-10-20 Online:2006-09-30 Published:2010-09-19

摘要: 模糊神经网络和预测神经网络分别是基于经验和学习的新型神经网络控制系统,通过在卧式电液仿真转台中框控制器上分别采用这2种控制方法来研究它们的控制特性和应用范围.其中,模糊神经网络结合了模糊控制的经验和神经网络的学习能力,但控制精度取决于人为经验;所研究的预测神经网络采用了基于非线性自回归滑动平均模型建立预测模型,实现在线学习和在线控制,但初始阶段控制精度不高.仿真研究证明,根据具体的控制对象采用适当的控制方法或是将2种方法合理地结合起来将会达到较高的控制精度.

Abstract: Fuzzy neural net(FNN) and predictive neural net(PNN) are new neural net controllers, two neural net controllers based on practical methods from actual control system and self-study. FNN and PNN controllers avoid many shortcomings of usual artificial neural net. Two neural nets for electronic-hydraulic simulating rotary-table′s middle gimbal were used to research their control characteristics and application ranges.FNN controller unites fuzzy control experiences and neural net self-study capability, but control precision only depends on summarize personal experiences; PNN controller uses nonlinear auto regressive moving average (NARMA) model for predictive model, makes real time study and control for all process,but control precisions is lower in start phase. Simulation results of FNN and PNN controllers show that differnet methods for different control objects or unit two methods for different control objects have achieved high precisions.

中图分类号: 


版权所有 © 《北京航空航天大学学报》编辑部
通讯地址:北京市海淀区学院路37号 北京航空航天大学学报编辑部 邮编:100191 E-mail:jbuaa@buaa.edu.cn
本系统由北京玛格泰克科技发展有限公司设计开发