基于RNPI的动态时滞网络主动队列管理算法

Robust nonlinear PI algorithm for active queue management in dynamic time-delay network

针对网络存在较大时滞和网络参数动态变化等问题,提出了一种鲁棒非线性PI主动队列管理算法。由流体流模型将动态网络推导为参数区间不确定一阶时滞系统,在确定使闭环系统稳定的非线性PI控制参数集合基础上,利用遗传算法寻找基于改进ITAE指标最优的PI控制参数。基于扩展到时滞系统的棱边定理,给出了RNPI算法的设计方法。仿真结果表明了该方法具有良好的控制性能,对网络区间不确定参数有较好的鲁棒性。

A robust nonlinear proportional integral （RNPI） algorithm for active queue management is pro posed to alleviate the time delay and parameters time varying problems in network. A first-order time delay model with parametric interval uncertainty is derived from fluid-flow model. The region of the nonlinear PI pa rameters making the closed-loop system stable is ascertained. The optimal parameters of the controller are ob rained using genetic algorithm with the optimal integrated time absolute error （ITAE）. The edge theorem ex tended to time delay systems is used to determine the design method of the RNPI controller. Simulation results show RNPI algorithm can achieve favorable performances and is robust against pa,ametric interval uncertainties.