一种基于LS-SVM与PID复合的逆控制系统

作者：黄银蓉，张绍德,季民 日期：2010-03-26/span> 浏览：4373 查看PDF文档

一种基于LS-SVM与PID复合的逆控制系统

黄银蓉，张绍德,季民
（安徽工业大学 电气信息学院，安徽 马鞍山 243002）

摘要：针对逆系统中非线性逆模型辨识困难的问题，研究了基于最小二乘支持向量机（LS-SVM）的逆模型辨识及控制，并用微粒子群算法（PSO）优化LS-SVM的参数和核函数参数。提出了一种由LS-SVM的逆模型与PID结合的复合控制系统,由LS-SVM辨识非线性系统的逆模型作为前馈控制器,形成直接逆控制。同时,由PID控制器构成反馈控制,克服直接逆控制鲁棒性不强的缺陷。仿真研究结果表明LS-SVM的逆模型辨识能力强，该复合控制系统具有比基于最近邻聚类的RBF神经网络逆控制系统更优的动态跟踪性能，更好的抗干扰能力和鲁棒性。
关键词：逆模型辨识；最小二乘支持向量机；微粒子群算法；逆控制
中图分类号：TP18文献标识码：A文章编号：1001-4551(2010)02-0075-04

A inverse control system based on least squares
support vector machine and PID
HUANG Yin-rong, ZHANG Shao-de, JI Min
(School of Electrical Engineering&Information, Anhui University of Technology, Maanshan 243002, China)
Abstract: Aiming at the problems of the inverse model identification in inverse system method, the realization of inverse system identification and control using least squares support vector machine(LS-SVM) were studied. A methodology, based on particle swarm optimization algorithm(PSO), for parameters selection of least squares support vector machine was proposed. A compound control strategy combing LS-SVM inverse system with PID controller was proposed. LS-SVM was used to identify the inverse model of nonlinear system, and this inverse model was used as feed-forward controller to design direct inverse control. Moreover, PID controller was used to realize feed-back control, which could overcome the limits of direct inverse control in performance and robustness. Simulation results demonstrate that LS-SVM has a good approximate capability for inverse model, and the proposed compound control system has better dynamic track performance, resistance to disturbance of system and robustness than RBF neural network.
Key words:  inverse model identification; least squares support vector machine(LS-SVM); particle swarm optimization algorithm(PSO); inverse control
参考文献(References):
［1］李春文,冯元现.多变量非线性控制的逆系统方法［M］.北京:清华大学出版社,1991.
［2］刘士荣,俞金寿.神经模糊逆模PID复合控制在CSTR中的应用［J］.控制理论与应用,2001,18(5):769-773.
［3］VAPNIK V N. The Nature of Statistical Learning Theory［M］. 2nd ed. New York: Springer Verlag,1995.
［4］DREZET P M L, HARRISON R F. Support vector machines for system identification［C］//Proc. of UKACC Int. Conf. on Control’98. UK: university of Wales Press,1998:668-692.
［5］SUYKENS J A K, GESTEL T V, BRABANTER J , et al. Least Squares Support Vector Machines ［M］. World Scientific,2002.
［6］王宏志,陈帅,侍洪波.基于最小二乘支持向量机和PSO算法的电厂烟气含氧量软测量［J］.热力发电,2008,37(3):35-38.
［7］王启付,王战江,王书亭.一种动态改变惯性权重的粒子群优化算法［J］.中国机械工程,2005,16(11):945-948.
［8］WANG Hui, PI Dao-ying, SUN You-xian. Online SVM regression algorithmbased adaptive inverse control ［J］. Neurocomputing,2007,70(4-6):952-959.