신경회로망을 이용한 이산치 혼돈 시스템의 모델 예측제어

Model Predictive Control of Discrete-Time Chaotic Systems Using Neural Network

  • 김세민 (연세대 전기.컴퓨터공학과) ;
  • 최윤호 (경기대 전자공학과) ;
  • 박진배 (연세대 전기.컴퓨터공학과) ;
  • 주영훈 (군산대 제어계측공학과)
  • Kim, Se-Min (Dept. of Electrical & Computer Engineering, Yonsei Univ.) ;
  • Choi, Yoon-Ho (Dept. of Electronic Engineering, Kyonggi Univ.) ;
  • Park, Jin-Bae (Dept. of Electrical & Computer Engineering, Yonsei Univ.) ;
  • Joo, Young-Hoon (Dept. of Control & Instrumentation Engineering, Kunsan National Univ.)
  • 발행 : 1999.07.19

초록

In this paper, we present model predictive control scheme based on neural network to control discrete-time chaotic systems. We use a feedforward neural network as nonlinear prediction model. The training algorithm used is an adaptive backpropagation algorithm that tunes the connection weights. And control signal is obtained by using gradient descent (GD), some kind of LMS method. We identify that the system identification results through model prediction control have a great effect on control performance. Finally, simulation results show that the proposed control algorithm performs much better than the conventional controller.

키워드