PSO算法在多模型自校正动态矩阵控制中的应用  

Application of PSO to Multimodel Self-Tuning Dynamic Matrix Control

作　　者：岳恒[1] 张海军[1] 柴天佑[1] 

机构地区：[1]东北大学流程工业综合自动化教育部重点实验室,辽宁沈阳110004

出　　处：《东北大学学报（自然科学版）》2008年第2期153-156,165,共5页Journal of Northeastern University(Natural Science)

基　　金：教育部新世纪优秀人才支持计划项目(NCET-05-0294);国家高技术研究发展计划项目(2006AA04Z179)

摘　　要：为解决多模型控制中固定模型获取问题,将粒子群优化(PSO)算法应用于多模型自校正动态矩阵控制.对一类含跳变参数的单输入单输出离散时间系统,当模型参数突然跳变时,通过PSO算法在线优化自适应模型参数,并根据模型相似度实现固定模型的动态管理,以有效控制模型数量和减轻系统负担.模型切换策略用于选择当前与实际被控对象最接近的控制器.仿真结果表明,该方法能够较大地改善系统的瞬态响应,优于常规的自校正动态矩阵控制算法;并说明了其有效性和可行性.To obtain the fixed models in muhimodel control problem, the PSO （particle swarm optimization） algorithm is applied to multimodel self-tuning dynamic matrix control. For a class of SISO discrete time systems with jumping model parameters, the PSO algorithm is used to optimize the self-tuning model parameters online and manage dynamically the fixed models in accordance to the similarity between models, so that the number of models may be handled and system burden can be reduced effectively. Model switching tactics is used to select the controller best to describe the actual controlled system. Simulation results show that the method proposed can improve the transient response greatly in comparison with conventional control algorithms and demonstrate its efficiency and feasibility.

关 键 词：多模型 粒子群 自校正 动态矩阵 预测控制 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]