一种基于线性模型预测的网络化预测模糊控制方法  

A networked predictive fuzzy control method based on linear model prediction

作　　者：佟世文[1] 方建军[1] 李红星[1] 李媛[1] 

机构地区：[1]北京联合大学自动化学院,北京100101

出　　处：《中南大学学报（自然科学版）》2015年第4期1281-1287,共7页Journal of Central South University:Science and Technology

基　　金：北京市属市管高校模式识别与智能机器人系统学术创新团队项目(PHR201106149);北京市自然科学基金资助项目(4142018);北京市优秀人才培养资助项目(2013D005022000003);北京联合大学新起点计划项目(ZK201216)~~

摘　　要：从方法论出发,通过对比、类比、条件转化、逆向法、整体和局部的关系、定性分析与定量分析相结合的方法,提出网络化预测模糊控制概念,将预测的思想和模糊控制相结合补偿网络中的时延,形成基于线性模型预测的网络化预测模糊控制方法。将变论域思想应用到网络化控制中,即论域随着误差的减小而收缩,而论域的收缩相当于控制规则的增加,从而可以用较少的控制规则实现较精确的控制。为解决控制器参数多、难于调节的问题,采用定性与定量分析相结合的方法,将参数按照物理意义分成组,从而将多参数的调节问题转化成研究每一组少数几个参数的调节问题,通过与PID及网络化预测控制(NPC)方法在有延时和无延时情况下的控制效果比较,证实本文提出的思想和方法能有效补偿网络化控制系统中的时延。By employing methodologies, such as contrast, analog, reverse, relations between global and local, qualitative and quantitative analysis, a networked predictive fuzzy control method was proposed based on the linear model predictor to compensate for the network delay. The idea of variable domain was applied to the networked control, i.e. the domain contracts with the decrease of error. The contraction of the domain was equivalent to the increase of the control rules. Therefore, a more precise control can be achieved with less control rules. To solve the parameters tuning problems of the controller, a combination of qualitative and quantitative analysis method was implemented. Multi-parameter adjustment was transformed into the tuning issue of a few parameters in each group after the parameters were divided into groups according to the physical meaning. The simulative results present good performance of the networked predictive fuzzy control method compared with PID and NPC controller.

关 键 词：网络化控制 模糊控制 模型预测 时延补偿 

分 类 号：TP273.4[自动化与计算机技术—检测技术与自动化装置]