基于径向基神经网络的Internet网络时延模型研究

以Internet为媒介进行控制信息的传送,是远程控制发展的重要趋势,而Internet传输时延的不确定性,使得这样的远程控制信息闭环控制系统难以实现.在实际测量网络时延实际数据的基础上,利用滑动模型和样条原理,采用滑模均值代替滑模的常规输入值,利用径向基神经网络建立了网络时延的模型.仿真结果表明该模型能够反映并预测该测量数据所代表的网络路径之间的时延特性,并能在基于Internet的闭环控制系统的设计中有效地替代实际网络进行研究;而所用的建模方法具有快速、准确的特点,能用于在线学习网络模型并对网络的时延值进行预测,为基于Internet的远程闭环控制提供了新的思路.     

Study on control information network and its real-time property

The control network is an important developmental orientation in the remote control system. As the control network and information network are comparatively alike in the framework and technology, we can build a control network which is similar to the common information network. In the era when the information network is becoming increasingly mature, it is a royal road to construct or rebuild a control information network in the development of the control network by relying on the achievements made in the information network or current information resources. This paper expounds the construction idea of the control information network, gives the idiographic realization method and then researches into the real-time problem encountered in the control information network, and presents a three-closed-loop control system based on virtualized reality. The feasibility of the idea is validated via experiments and simulations separately.