基于ADP方法求解未知非线性零和微分对策问题-Journalof.PDFVIP

第33卷第12期 东 北 大 学 学 报 ( 自 然 科 学 版 ) Vol. 33, No. 12 2012 年 12 月 Journal of Northeastern U niversity( Natural cience) D ec. 2 0 1 2 ADP 张 欣, 会国涛, 罗艳红 ( , 110819) : , (ADP) ( RNN) RNN , RNN RNN , ADP : ; ; ; ; : TP 183 : A : 1005-3026(2012) 12-1673-05 ADP Approach to Solve Unknown Nonlinear Zero-Sum Game ZHAN G Xin, H UI Guo-tao, L UO Yan-hong ( chool of Information cience Engineering, Northeastern University, henyang 110819, China. Corresponding author : ZHANG Xin, E-mail: jackie zx @ yahoo. com. cn) - Abstract: An approximate dynamic programming ( ADP) approach was proposed for a class of unknow n nonlinear zero-sum game. A model based on a recurrent neural network ( RNN) was used to approximate the unknown system dynamics. A novel adjustable term related to the modeling error w as added to the RNN model, w hich guaranteed the modeling error convergent to zero. T hen, an ADP approach w as given to solve the optimal performance index and the optimal control pair under the saddle point of the zero-sum game existence or not. imulation results demonstrated the effectiveness of the proposed control scheme. Key words: unknow n nonlinear system; zero-sum game; recurrent neural network; ADP; saddle point , ; , , [6- 7] , , , , , ; ,