基于神经网络的多变量非线性系统的广义预测控制  

The Generalized Predictive Control of Multivariable Nonlinear System Based on Neutral Network

作　　者：郑秀萍[1] 

机构地区：[1]太原科技大学电子信息工程学院,山西太原030024

出　　处：《科技情报开发与经济》2008年第11期142-144,共3页Sci-Tech Information Development & Economy

摘　　要：为了使广义预测控制的思想成功应用于多变量非线性系统,用神经网络对其进行开环解耦得到单变量非线性系统后,采用一种复合多层前馈神经网络结构作为单变量非线性系统预测模型,利用递推最小二乘法和Davidon最小二乘法作为在线学习算法,建立了一种适合多变量非线性系统的自校正广义预测控制器。In order to apply successfully generalized predictive control into multivariable nonlinear system, after using neutral network to decouple it and obtain the single variable nonlinear system, and taking a feed forward neutral network as the predictive model of single variable nonlinear system, this paper establishes a self-tuning generalized predictive controller for multivariable nonlinear system by using recursive least square method and Davidon least square method as online learning method.

关 键 词：神经网络 多变量非线性系统 解耦 广义预测 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]