基于灰预测的PID主动队列管理算法  

Active queue management based on PID control with grey prediction

作　　者：夏冬雪[1] 阳树洪[2] 杨斌[1] 

机构地区：[1]西南交通大学信息科学与技术学院,四川成都610031 [2]广西工学院计算机工程系,广西柳州545006

出　　处：《计算机工程与设计》2008年第1期68-71,共4页Computer Engineering and Design

摘　　要：提出一种基于灰色预测的智能PID(GI-PID)主动队列管理(AQM)算法,该算法采用GM(1,1)模型在线预测路由器队列长度,补偿滞后以解决网络状况反馈不及时的问题;同时根据队列误差的变化趋势,应用专家经验动态改变PID控制器的参数,使参数实时地随着网络环境变化而调整,实现智能控制。仿真试验表明,GI-PID算法相比传统PID算法大幅度地抑制了队列长度的振荡,路由器队列收敛于期望值,同时具有较小的分组丢弃概率。A new AQM algorithm based on Intelligent PID control with grey prediction （GI-PID） is proposed. Online GM （1,1） model based on consistent dimension innovation is successfully introduced into feedback dam＇s advanced prediction to compensate AQM＇s feedback delay. On the other hand, according to the queue error fluctuation tendency, with specialists＇ experience to change the parameters of PID control machine, the own parameters of it adapt themselves to the Internet accordingly then intelligent control is realized. Con- trasted with traditional PID algorithm, GI-PID algorithrn can restrain the queue length fluctuation in a large degree and converge to queue size-setting value, and can get lesser packets loss rate.

关 键 词：主动队列管理 拥塞控制 灰色预测 智能控制 专家系统 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]