摘要
针对非线性被控对象,该文提出了基于支持向量机的逆系统控制方法。对于最小相位的非线性离散系统,该方法根据系统的输入输出数据,使用支持向量机回归的方法来辨识构造原系统的α-阶逆系统。将辨识构造出的逆系统与原系统相联结,就能形成α-阶纯延时伪线性系统。这样,使用线性系统的成熟控制方法(如极点配置等等),就能有效地对非线性系统进行控制。仿真实验显示,即使对于非仿射并且非线性很强的系统,在没有系统模型的先验知识的情况下,利用该方法都能准确地建立逆系统的模型,从而获得良好的控制效果。
The most important and difficult step in inverse control methods is the modeling of the inverse nonlinear dynamic system. Support vector machines can more easily handle this problem because of their outstanding pability for regression analyses and system modeling. This article describes an inverse control method based on a support vector machine (SVM) utilized as a regression tool to approximate the inverse dynamic system of the minimum-phase nonlinear discrete process. A pure time delay system is obtained when the inverse system approximated by the SVM is used to control the process, so that traditional linear control algorithms (such as pole placement) can be used. Simulations demonstrate that without prior information about the process, the method accurately models the inverse dynamic system of the process (even complex non-affine processes) to provide excellent control results.
出处
《清华大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
2005年第1期100-102,106,共4页
Journal of Tsinghua University(Science and Technology)
基金
国家"十五"科技攻关项目(2001BA609A-12)
关键词
智能控制
逆系统控制
支持向量机
回归
非线性系统辨识
最小相位
intelligent control
inverse control
support vector machine (SVM)
regression
nonlinear system identification
minimum phase