基于混沌—神经网络模型最优控制及应用  

Optimal control and application based on chaos-neural network model

作　　者：窦春霞[1] 

机构地区：[1]燕山大学电气工程学院,河北秦皇岛066004

出　　处：《系统工程学报》2004年第3期229-233,共5页Journal of Systems Engineering

基　　金：国家自然科学基金资助项目(60102002);河北省基金资助项目(6011224);霍英东基金资助项目(81057).

摘　　要：由于非线性混沌时间序列内部确定的规律性,其重构相空间具有高精度短期预测性.为此,为了实现非线性、大时滞系统的自适应控制,文章根据具有混沌特性非线性、大时滞系统的时间序列重构相空间,计算相空间饱和嵌入维数、最大Lyapunov指数和系统的可预报尺度,并以此为指导,建立神经网络预测模型对系统作高精度的短期预测;在此基础上,通过反馈校正,将校正误差和控制增量引入性能函数寻优得最优控制决策,实现了对非线性、大时滞系统高精度的自适应预测控制.将该控制决策应用在锅炉过热汽温控制中,仿真表明该控制的有效性、快速性和鲁棒性.Because of chaotic time series internal certain regularity, their reconstructing chaotic attractors phase space can be used for high precision short-term forecast. Therefore, in order to realize adaptive control of a nonlinear big-lagged system, the phase space is reconstructed, its saturated embedded dimension, the maximal Lyapunov exponent and the forecast measure are calculated by the nonlinear big-lagged system time series in this paper. After that, a neural-network model is constructed, which can make high precision short-term forecast for the system. On the basis of this, an optimal controller is designed by a feedback rectification term and the control input error is introduced into a performance function, and then a high precision adaptive forecast control is realized to the nonlinear big-lagged system. The controller is applied to the overheat steam temperature system of drum boiler. The validity, the high-speed and the robustness of the control system are demonstrated by simulation results.

关 键 词：神经网络 最优控制 混沌时间序列 LYAPUNOV指数 鲁棒性 

分 类 号：O211.61[理学—概率论与数理统计] TP273[理学—数学]