基于三次样条滚动优化多步预测算法的网络学习控制  

Networked Learning Control Based on Multi-step Prediction Algorithm of Cubic Spline Rolling Optimization

作　　者：易军[1] 费敏锐[1] 

机构地区：[1]上海大学机电工程与自动化学院,上海200072

出　　处：《信息与控制》2006年第1期1-5,共5页Information and Control

基　　金：国家自然科学基金资助项目(60274031);高等学校博士学科点专项科研基金资助项目(20040280017);上海市科委重点基础研究资助项目(04JC14038);新世纪优秀人才支持计划资助项目

摘　　要：对于网络延时给基于网络学习的控制带来的不利影响,提出了基于三次样条滚动优化多步预测模型和神经网络在线调节的复合控制,并对令牌网上的复杂、时变、非线性及有外部干扰的被控对象进行了跟踪控制仿真.仿真结果表明:采用本文的复合控制策略,能够较好地克服网络延时对复杂对象控制所产生的不利影响,具有很好的快速性和跟踪精度.For the adverse effect caused by network delay on networked-learning-based control, a composite control is proposed based on multi-step prediction model of cubic spline rolling optimization and on-line adjustment of neural network. Tracking control simulation is performed on the complex, time-varying and nonlinear controlled object with external disturbances in Token network. Simulation results show that the adverse effect caused hy network delay on the controlled complex object can be overcome, and rapidity and high-precision tracking can bc achieved by applying the composite control strategy.

关 键 词：网络学习 三次样条 神经网络 滚动优化 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]