【摘要】 为了解决网络诱导时延给网络化控制系统(Networked Control Systems,NCSs)带来的不利影响,在分析Ethernet时延特性的基础上,提出了一种不依赖于网络诱导时延精确数学模型的改进的神经元PID控制器。该控制器利用单神经元实现自适应PID控制,利用神经元良好的学习能力克服网络化控制系统中不确定的网络诱导时延的影响。改进的神经元PID控制器根据有限的时延信息动态调整用于计算控制输出量的参数,该控制器还利用一种在线调整规则对神经元的比例系数进行在线更新,并针对Ethernet这类网络进行了仿真研究。仿真结果表明,所提出的方法能有效提高控制系统的性能。更多还原

【Abstract】 To offset the adverse influence caused by network-induced delay,based on analyzing delay characteristic of Ethernet,a kind of improved neuron PID controller independing on the mathematical model of network-induced delay is proposed.The single neuron is used to realize the adaptive PID control,and the favorable learning ability of neuron is utilized to overcome the effect of uncertain network-induced delay in NCS.It is primarily through finite information about time delay that such controller dynamically regulates a series of parameters employed for calculating controlled output value,and takes advantage of a kind of online regulative order to update proportional coefficient of neuron.The numerical simulation on Ethernet shows that the proposed method can improve performance of the system.更多还原