基于粒子群算法改进SVM的滑模控制研究  

Sliding Mode Control Based on SVM Optimized by PSOl

作　　者：齐立磊[1] 陆剑[2] 

机构地区：[1]南阳理工学院软件学院,河南南阳473004 [2]河南工业职业技术学院,河南南阳473009

出　　处：《计算机测量与控制》2014年第10期3230-3232,共3页Computer Measurement &Control

基　　金：河南省科学技术厅"基于物联网的城市智能停车服务系统研究与应用"(132102210566)

摘　　要：为了提高离散时间系统的控制品质以及削弱系统抖振,提出了基于粒子群算法(particle swarm optimization,PSO)改进支持向量机(support vector machine,SVM)的滑模控制方法并进行了仿真研究;通过SVM识辨参量模型与PSO寻优处理,获得趋于理想滑模运动的趋近律参数,确保寻优处理时间短、精度高;利用PSO和SVM在线调整滑模趋近律参数,可以克服常规滑模控制中需要事先设定趋近律参数限制的弊端,加快跟踪速度,削弱系统抖振,完善控制质量;仿真实验表明,该方法可以克服因PSO寻优过程中的寻优时间过长等不足,又可解决SVM精度不高或计算量大的缺点;该方法用于离散时间系统是可行、有效的,工程实用性强。For discrete--time system, Based on an improved SVM of particle swarm algorithm, a new support vector machine sliding mode control theory is proposed. Through SVM parameters identified model and PSO optimization process, reaching law parameter tends to ideal sliding movement is obtained, ensure that optimization with short processing time and high accuracy. Adjust sliding reaching law param- eter online by adopting particle swarm optimization and support vector machine, which can overcome the shortage of reaching law parameter required pre--set in the conventional sliding mode control, accelerate the tracking speed, weaken the system chattering, improve quality con- trol. The results show that the method can overcome the shortcomings and deficiencies of optimization parameter optimization＇ s difficult to find or too long, etc caused by improper handling in the PSO optimization process, and also can solve shortage of the SVM with low accuracy and large computation, which is feasible and effective, with good usability.

关 键 词：滑模控制 粒子群算法 支持向量机 跟踪 抖振 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]